Super Bowl Ad Reactions, New Ferrari Design, Ads Launch in ChatGPT | Jason Fried, Bill Bishop, Jason Kelly, Dan Romero, Boris Sofman, Sara Hooker, Edward Mehr

You're watching TVPN.

Today is Monday, February 9, 2026. We are live from the TVPN ultradome metaphor of technology, the fortress of finance, the capital of capital.

It is great to be back.

It's great to be back. We went to the super bowl with none other than ramp.com Time is money save. Both easy use, corporate cards, bill pay, accounting, and a whole lot more all in one place. We're breaking down our super bowl experience. You know, people call us the sports center for the LinkedIn crowd. It's always been funny because we mostly focus on X and RSS feeds.

Well, now is the first football game we saw this season.

Yeah.

Was. Yes.

Yeah, it actually was. It was the first football game I've been to in maybe a decade. I don't know, it's been a while. But we do try and bring that SportsCenter energy to the show. We like, you know, ringing a gong. I don't know if they have a sport, if they have a gong on SportsCenter, but you know, we try and bring the high energy to tech and business reporting and. And now we're. We're officially sports fans. Were officially football fans.

Converted.

Converted. No, it was a fun experience going to the Super Bowl.

Apparently it was not the most exciting game. Yeah, we did have to leave to catch a flight in the fourth quarter and then it got. And then it's started being more exciting.

But also just a weird experience seeing it in the stadium because there's so many ads. But they don't show the super bowl ads in the stadium. And so they'll run a couple plays and then you'll just take a five minute break and then you just get distracted and get sucked into a conversation because you. We were there with a bunch of fun people and so it was a little bit trickier to actually follow the game.

You mean follow the ads?

Yeah. Well, you didn't follow the ads at all. I didn't see a single ad. I would open my phone and see people reacting to the ads and I had major FOMO because I wasn't getting to enjoy the ads in the stadium. Of course there were some branded integrations in the stadium, but those are separate from the ads that NBC sells. Of course. We had this running joke for a while. We're most excited about the ads. It's like a little played out at this point because some people say that just as a reflection of like, they don't like sports. We say it because we actually like the ads.

Right.

Because we like advertising and commerce but you know, we participated in the super bowl hype train. I was very happy with the success of our campaign. Super bowl is like a relatively minor event in the calendar for tech people. I feel like wwdc, you have Davos, you have Sun Valley. Like of the things that everyone collects around Super Bowls on the calendar for a lot of people, but not top line for everyone. It's not. You gotta be there. But we were able to run a regional ad in the Bay Area which we mentioned on the show.

And I guess it was all over California.

Yeah, it was all over California.

People text me from Southern California too.

Yeah, but it was fun. I mean people see the.

We had this one guy, I got a poll, I gotta pull up his post because somebody yesterday thought that they were hallucinating. This guy Chip Rogers on X said hallucinating and then he said ordyhaze ohn Cooganvpn on the pre game super bowl commercial. And I was like, no, it was real. It would have been insane to just be like watching NBC or watching the. And then you just get.

You're watching tvpn.

He probably was like, did I sit on the remote or something?

Yeah. And I accidentally clicked off the stream or something. But no, the intro was. The intro to our ad was exactly our intro to our main show, which is funny, but it was a cool moment because obviously people see view numbers on clips and they see follower counts. We just hit 200,000 on X. We're very excited about that. And they see the guest lineups. But there's something different about actually seeing the patchwork of all the different logos of everyone who's participated in the show in one way or another as a guest. And that was just very cool. And I think people really sort of understood the scope of like how crazy 2025 was. I went from interviewing like one or two people a year for videos to like 1000. And it was great. It was a lot of fun. Let's pull up the linear lineup because I want you to meet the system for modern software development. 70% of enterprise workspaces on linear are using agents. We have Jason Fried, the co founder.

CEO of 37 excited about this one. Jason's going to come on, we're going to have a conversation about the new Jony I've Ferrari designs that got announced this morning. We'll go through it in a little bit. Bill Bishop to him, Bill Bishop to.

Talk about legend, about what's going on in China broadly risk to TSMC Taiwan. What's going on with the Leadership changes and shakeups. Over in the ccp, we have Jason.

Kelly, co founder and CEO of Ginkgo Bioworks, coming on.

What could go wrong if you hook a Biolab up to GPT5.

Yeah. And if you saw OpenAI shared last week, they have a.

Like a.

Basically biolab integration. Jason is powering that via Ginkgo. And then we have Dan Romero, who's joining Tempo, the project between Stripe and Paradigm.

I saw a big billboard for it up in sf.

Very cool. And then Boris from Bedrock Robotics, Sarah Hooker from Adaption Labs, and Ed from Machina Labs coming in in person. So great show today.

Yeah. The other super bowl little tomfoolery we engaged in was we launched Claude with ads. Obviously, we'd been joking back and forth about Anthropic launching a Super bowl ad, kind of taking a shot at OpenAI or other LLMs that might put ads in there. What does that mean? Is it going to be bad? And so, of course, we had to create a wrapper, thanks to the Opus 4.6 API and a lot of tireless work from Tyler Cosgrove over there.

Yeah, really amazing execution. Went from Idea Friday morning to product that. Thousands.

Yeah. Actually, like, a lot of people, 7,000.

People have, like, used it, played around with it.

8,000 people signed the petition to bring Claude. To bring ads to Claude. That was your line, right?

Yeah, yeah, yeah. It's also. I think this is actually important for a lot of safety people to think about. Right. Because this is actually a very good example of misalignment from Claude. Right. Because I used Claude to build this. I used Claude code. Right. And so this is Claude acting in defiance of Anthropic stated principles.

Yes. Anthropic is anti. And yet Claude, the model they built, allowed you to add ads to Claude. That's textbook mislimited.

The safety people need to be.

They need to study this.

Not in the chat. So Sholto has left the chat. We are with Sholto yesterday at the Super Bowl.

He had fun.

He was having fun with it.

A lot of you were asking illegal is this violation.

We respect trademark law a little bit too seriously.

We're just having some fun and it is just tomfoolery. So the site will be going down later today. So your last chance to give it a whirl is today your last chance.

To get totally free access to Anthropic's most powerful model.

Yes. As long as you have unlimited email addresses, I think, because you can technically only get one prompt, and the first prompt, break it down the first Prompt comes back.

We were really. So the concern was that we launched this. Somebody sets up an agent to effectively sign up millions of times and run up a crazy bill.

Yeah, but if you actually use it.

Like you send a message and then it responds, it calls the API.

Why are you asking them?

And then. So it actually calls the API on the first message.

Yeah.

But then after that it's just like, it's just hard coded in. Like it'll just give you an ad.

And there's no actual response.

It just says like. That's a great question. Also, have you heard about.

So in exchange for one email address, which is somewhat scarce, we give you one prompt of four points, but in that prompt there's still an ad.

Yeah.

And then everything subsequently.

I mean, you can make new chats, but you still on each, like, oh.

Okay, so you could make new chats and then that would allow you to hit open source.

So you could kind of have a conversation if you basically add the context.

Of the previous conversation in the new one every time. Okay, yeah. So a little prompt engineering and boom. Opus 4.6 intelligence. And it's too cheap.

Opus 4.6, we love it.

If you can figure it out, we love it. Let me tell you about figma. Figma make isn't your average value coding tool. It lives in figma. So outputs look good, feel real and stay connected to how teams build, create code back prototypes and apps fast. And so I was very surprised by how well Claude with ads.com did, given that we tweeted a link. There's been debate over how nerfed are links on X. We got a ton of likes and views on this and it seemed to work very well. So maybe that bodes well just for like, it's a weird link and it's worth clicking on. But also I just think that the whole X algorithm is less punishing to links these days, which is. Is exciting. And you know, we've always had this shtick that we're extremely pro advertising, have been very aggressive about finding funny ways to integrate sponsors all over tvpn. We have the turbopuffer, we have our console laptops, we have all sorts of stuff, the Axon Gong, the Lambda Lightning round, et cetera. And I've always thought, I'm working on this metaphor for the show, like, how do we think about the show? And I like the metaphor of like, we think of it like an F1 car, like F1 race. There aren't all that.

You've always said it's an F1 team.

F1 team. But I think the metaphor can be extended because the number of people that actually go to every single race in person is pretty small. And so is our community of people here in the chat. We love all of you.

Ryan says, I put my dad onto tbpn. I got excited seeing the ad and he started asking me about it. Converted. There we go.

There we go.

Good work.

And so, but you know, the number of people that watch a little bit of an F1 race or the highlights or the drive to survive, and you can think about diet TVPN as sort of the drive to survive of tvpn. You know, you're not sitting there watching the whole thing live. It's a little bit more condensed, a little bit more editing, and then the clips are sort of just, you know, you're aware that Red Bull has a team and you're aware that Ramp sponsors tvpn. What's this?

Rookie Tyler was learning to muddy spread when Chads, Jordy and John were beer maxing with the Patels. Turns out Tyler vibe coated Claude with ads. But money spread yet.

Oh, he can't.

Oh, he botched. Because they're like brand new bills, so it's harder.

Oh, there we go. Get those glasses on. You gotta look the other direction. There you go.

Rookie, rookie, rookie. But he's learning quick anyway. No, I can't. I mean, Tyler, just incredible, incredible work. Should we, should we get into some of the ads?

Yeah.

From the show?

Yeah, yeah, yeah, yeah. I think, I think the big meta point from the. The tech's response to the super bowl, which we'll dig into, is like, it's very fun. And we spent most of last week, like, obsessing over the inter lab vibe wars. Like, how is OpenAI responding to anthropic. And as we'll see, like, this was not the story of the super bowl at all. The story of the super bowl, even from a tech perspective, was much more about how is tech communicating what AI can do broadly to the largest swath of Americans and broad stakeholders and voters possible. And so people were. There's debates over data centers, crypto, gambling, weight loss, drugs. There's new technologies and society's grappling with those. And the super bowl is actually a very interesting place to go and make a case for how you want this technology to be integrated into society. You're making a case for how it can be used positively, all of these different things. And even though it's fun to focus on, like, you know, oh, should Claude have ads or should ChatGPT not have ads or whatever. Like there's a much bigger discussion that's happening at the super bowl and I think that's what we should be running through as we react to these super bowl ads. While we pull the first one up. Jordi, you can pick whatever you want to we want to talk about first. I will tell everyone about Vanta Automate Compliance and Security. Vanta is the leading AI trust management platform. And so do you want to read this reaction to the Claude super bowl ad?

Yeah, we can go through. Let's go through Claude first. Orange on over on X was posting a bunch of good kind of reactions analysis. He says ranking the cloud super bowl ad in the moment surrounded by people that aren't all terminally online. 80% of the people here don't understand that ChatGPT speak when the second speaker talks and we're confused or tuned out.

Yeah.

So that to be explained.

It's sort of iconic to me because I've used ChatGPT voice mode.

Yeah, I've seen those and I've seen.

Those reels where it's like.

It was like perfect execution.

Great question.

For X, one of the other challenges is again, we weren't seeing it live, but apparently the ad was much shorter. There was one ad during the game I kind of expected them to run to really try to really mog and do like 430 second ads.

Someone was saying it might be like an $80 million buy which would be like insane anyway.

So they toned it down a bit and Horn says had to be explained by the AI bros. Which was as bad as it sounds. While this ad speaks well to the early bell curve, this might have been too early for a mainstream investment like this. Overall, I think they had a lot of fun with it. It clearly was not like the creative was like can't have been that expensive because it was like four scenes. It was like four one day shoots probably.

I think we're going to get the team on who did it. Mother.

Yeah.

Because it is beautifully shot and it's. And I think it's funny. But. But yeah, it's. It is a little early. But so was the chat.

But basically like. But. But in. In they. They basically made OpenAI way overreact.

Yeah.

Like actually seeing the way that the ads went yesterday, it was almost like maybe they sort of. Maybe they didn't need Sam and. And the entire team like you know, Streisand affecting it.

Well, this was what Rune was posting. He was like, it's the. It's the blue shell and it like, it successfully, like bridge baited.

Everyone got rage baited.

Yeah, but I mean, they've known. OpenAI has known that. They've had to like be very, very clear about the way ads will get integrated. Because when you. When you think ads and LLMs, you immediately think what we built with cloud with ads, which.

And then they did update it, they said there is a time and place. Instead of ads are coming to AI, but not to Claude. They updated it and said the new copy is there's a time and place for ads. Your conversation with AI should not be one of them. So I wonder what, you know, I wonder what pushback they got. This, this comp. This doesn't hit quite as hard, even though it is probably more authentic, right?

Yeah.

It almost sounds like it's like a lobbying firm or something.

Oh yeah. It does feel like something you'd see from like a think tank almost. Again, no QR code, no call to action. Where did Claude land in the App Store? We were wondering about that. Right. We were thinking that it was at 36 or something when we were looking at it last week. In the lead up.

Yeah, it's at 23 right now.

That's not bad.

So it's like up maybe 10.

Yeah, that's not too bad. There's a bunch of stuff that moved. I mean, just looking at the top, it's like Peacock. Okay. That's watching the big game. NBC app. That's the same thing. NFL app. Obviously that's super bowl related. Same with NBC Sports. So there were prize picks. Like there's a whole bunch of things that move. FanDuel is up there. There's just a number of apps that jumped just by default. In terms of free apps, there actually aren't that many others that ran. I see Dunkin. Dunkin feels like it's up in the app store. I think they ran a big ad that we'll look at. But yeah, the stuff that's below it doesn't seem like it's higher than temu. Last year TEMU ran a big super bowl ad. I don't know exactly what they did this year, but you would expect that.

My first time seeing a Temu ad was crazy. Yeah, I was like the copy shop. Like a billionaire.

It is funny.

Was so insane. And yet I think it actually, in hindsight, kind of works because everything's so cheap. People don't have to look at the price.

I guess that's the. That's the thesis to a billionaire. A couch feels like $5. So here's a $5 couch, an Apple or something like that. Okay, yeah, yeah. It is funny because like I don't think like the real way that billionaires shop is like, oh, my real estate guy bought a mansion.

Your real estate desk.

My real estate desk bought a mansion. And then the interior design desk will furnish it and I will have no input in any of it and they will just magically know exactly what I want before I even ask for it.

One thing I'm interested to see the way the super bowl buys work, they make you buy ads in the Olympics as well. And so if you're watching the Olympics today or tomorrow, I guess we're going to have an Olympics ad because we are forced to like we should do.

A whole another victory lap. We're taking our ad. It was so successful. We're taking it to this, to the Olympics.

We're taking it to the slopes.

The slopes.

But what will be interesting is if various labs or other companies run like longer ads. It's not quite as expensive. We'll see.

Be interesting. Bryce. Bryce said Anthropic ads were total flop in his house despite having a highly tech literate family. They took a bunch of explaining and yeah, it is, but that makes sense, right?

So like they hit so hard on X. Yeah.

Oh yeah.

I remember the morning of we were.

Sitting here being like, oh, it shot across the back.

I can't believe they did.

It was like a mic drop.

Yeah, mic drops. Predictable. There was no way to come back from it, you know, lost. As soon he was, as he was writing up like a word salad, you know, trying to respond. But, but again, there's some data coming back from, from ad week. Morgan is sharing audience didn't like Anthropic's ad, placing it in the bottom 3% of all Super bowl ads from the last five years. But it makes sense, right? It's like a barbell, like for people on AX that are like really following tech closely. It was incredibly pretty small sample size.

500 people they asked.

Yeah, but you're running that like in real time right after the ads go.

Well, let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. Let's look at how OpenAI responded. Obviously this is not a response. This was probably months or weeks in the making. This is called you can just build things. And let's take a look and see if it tells a more optimistic story about AI. One that's maybe less confrontational with their rivals. I Like building things, making cardboard stuff is underrated. You get a lot of Amazon boxes, cut those things up, make something very cool. Become a hacker, read more. Learn Bayesian probability. Become a scientist. Play chess. I don't think if chess is something you build necessarily sick job displacement. No more. No more sweeping. Somebody said this is a. This is a Windows computer running a running Mac or something. Is that real?

I think people would tune out for this.

Build things. Why?

It's just too so long.

Do they actually run the full thing? A full minute.

And the other. And the other thing is. I don't know. I'm still.

It is a little.

They're trying to push Codex. They're trying to push Codex to consumers.

Yeah.

Which I think is smart.

I think it's very smart.

But introducing Codex when every single person in the audience is familiar with ChatGPT.

Yeah.

And then just trying to. I don't know.

Oh, this is cool. So that robotics stuff was actually Easter eggs of the robotics team at OpenAI. Like, they just brought the cameras in and filmed their own team. That's very cool.

Yeah. Dan Chipper says, huge opening. I runs a Codex commercial, not a ChatGPT commercial.

So was also Codex commercial.

Yeah, go to the very end.

Because that just felt like a general. Like AI is cool commercial. Like, and just like it felt very in line with the previous super bowl ad of just the eras of technology. Let's go to the end. Oh, okay. So it's showing you Codex desktop. That's cool. Okay. Yeah, yeah. That's pretty subtle, though.

Everything about it is too subtle. It never says chatgpt.

What's the final slide? Because it says you can just build things and then.

And then it goes Codex.

Okay. Codex OpenAI. Oh, okay. So it's not okay. Yeah, yeah, yeah. You can just build things. Yeah.

I don't know. It's just not. It's. Again, it's kind of cool for X. People are like, oh. But I think Both Anthropic and OpenAI are both using the same reference material for their ads. They're all going back into the Apple archives trying to pull inspiration from the same fishing from the same pond.

Yeah. They should have just done a parody of the Budweiser Wazza ad. Have you seen that one? It's just people talking to ChatGPT voice. But WASA that would. They're doing Clydesdale's actually for ChatGPT. ChatGPT find me a Clydesdale.

Somebody on X. Grace said super bowl commercial. So evil this year. Seeing a Bud Light commercial felt healing. Like, oh, yes, Bud Light. A tangible object unrelated to AI or crypto.

Ridiculous. Let me tell you about MongoDB. Choose a database built for flexibility and scale. With best in class embedding models and re rankers. MongoDB has what you need to build. What's next? So obviously, we went to the super bowl with Eric Lyman from Ramp, the CEO of Ramp, the whole team. Ramp did really, really well with their super bowl activation. They had a whole bunch of different touch points. So they didn't just like, lob an ad in and then call it a day. They were there. They sent the Brian's. We can kind of go through the whole plan, but there's something interesting. I mean, we certainly experienced this with our super bowl ad. Obviously we bought a very small ad, but, like, why we got a good result out of our super bowl ad was we didn't just go to NBC and said, like, here's money and thank you. We went and told ad week and gave an interview and you gave some interesting quotes to the reporter on the record. So there was an article around it that's valuable. Then we emailed people who were featured in the ad. Hey, we're posting about this. Like, do you want to know that you're in this thing?

Cool.

You can go see it if you want to watch it. There's a whole bunch of different flywheels. And I think Ramp did a really good job of understanding that. Like, it's the super bowl campaign, not an ad.

Yeah. So, I mean, one they had. Obviously they'd work with Brian with during the box stunt in New York last year. That went really well. So they built off of that.

Yeah.

They. We showed up. We showed up to the. To the tailgate party that they had. By the time we showed up, there was already like thousands of people there. It was insane. Bunch of TVPN listeners came and said hi, which was awesome. But great group of people. They had a big yellow skateboard ramp with people, like, actually going pretty hard.

Yeah.

No really good skaters.

A bunch of different, like, contests running around. We're super Internet food. Somebody. Somebody. Brian shaved a guy's head.

Yeah.

On a live stream to look like himself. And that.

Yeah.

Ended up getting.

Some people were like, we're showing up and being like, I like this takes six levels of, like, understanding to get all the jokes of, like, the ramp is yellow because the brand and the ramp is the name of the company and the skater. And then like, Brian's here because he was in the office and he plays an accountant. But, like, I think that's fun for people. I think people actually do, like, Easter eggs and they like going down the rabbit hole, but at the same time, you got. You can't just be pure Easter eggs. And I think when you actually watch the Ramp commercial, which is right there, it's like staring you in the face. Obvious what this does. It's like the brand name Yellow in your face. What does it do? You have to do things. That takes you 10 hours now. There's 10 people, there's 10 copies of you that do the thing that you do.

You can do it in five minutes.

And so you can just do it much faster. And so you can watch this and not know anything about Ramp. And, you know, okay, corporate card, multiply. What's possible. Like, I have a task in accounting. It looks like an accountant. I'm triggered on that. I'm familiar with this character. And so all of that is, like, clarity at the top and then tons of depth. And even in that video, there's a whole bunch of Easter eggs. In the video, it was directed by someone who. Who directed the Office. And there's different actors from the Office in there and there's layers. And then you go on social media and you see other stuff. So really, really great. Like 360 execution.

Yeah. What worked well was like, classic super bowl ad, easy to understand celebrity, popular character, plus a bunch of the super Internet native stuff, like physical activation and then localized activation. So you had people. The Super Bowls in sf, you have thousands of people in SF participating on the ground. And so there was again, Dylan Field.

I love it. And this was so crazy because Anne Kong from Ramp told me at one point, like, oh, we're sending all the Brian's. And then she was mentioning, like, oh, well, they'll also be like ramp people there with prospective clients and stuff. And I was like, oh, they'll probably do one or the other. And they wound up doing everything. So Brian and a whole bunch of lookalikes were just there in basically the front row of the super bowl. And how can you not take a picture of that? It's so innately viral. It was very, very funn.

Yeah. Insane. Insane production.

Yeah. And it's just like these multiple touch points all have, like, different shots on goal, and they all have different, like, probabilities that you can sort of wait and be like, okay, well, if the ad is really loved, we'll jump to the top of the rankings in this. But then also there's a chance that, like, this photo gets, like, mega viral or this this like someone cool goes to the tailgate and they post about that. And so there's like all these different. All these different touch points that have like different shots on goal as opposed to just like, okay, we sent the big check to NBC. We hope the ad goes well. Right?

Yeah, it was pretty funny. On the way on the drive from the tailgate to the game, I was sitting with Eric and we were just catching up on life and business and all that stuff. And he's wearing the bald cap the whole time, which I should have here, but he was. Just stayed in it the whole time. Really fun.

It was wild. Really quickly, let me tell you about Restream 1 livestream. 30 plus destinations. If you want to multi stream, go to restream.com. there was a Brian lookalike spotted at the Super Bowl. They were just random people who I think went to the tailgate or got their head shaved and like went and stuff. It's like they were really like people all over the place.

We were, we were sitting across, looking across the entire stadium and we found the public team and we were zooming. We could zoom in enough to wave.

It was very fun. Testament to the new iPhone camera. Like the 8x zoom. You go 40x and you can actually see people all the way across the stadium.

Aman says OpenAI ad flop live with the normies. Didn't really get it. Thought they were all together.

Nine likes.

What?

I didn't really get it. Yeah, yeah, yeah, it is. It is tricky.

Yeah. It was just like a bunch of cool things and then a product and a name for a product that nobody knows about. That's the issue.

Yeah.

If you have a very popular product like Budweiser, something like that, you can just show a bunch of random images and then show your logo.

Yeah.

But if you don't, if you're. If you just show a bunch of random, like cool scenes or whatever, it's inspiring. It's like uplifting. And then you flash a logo that nobody knows. You end up with something like.

Sure.

That doesn't really, like, leave that much of an impression.

Do we have the Google Ad? Because I believe Google went way more practical on this.

Pull it up, Tyler. Let's talk about the. Before we get to that, we can talk about the Coinbase super bowl ad. This was probably the most. Well, it's controversial.

I still need to see exactly what they did. I saw some images of them putting something on the sphere, which was.

Pull this one up in the.

Surprising to me, because they didn't put the like the super bowl is not in Las Vegas. So the sphere is like a different advertising surface. But I guess it's cool to be like multiple touch points in that way.

I did like pull up the one I'm on in the, in the chat.

We ran into Brian Armstrong and Fred Ursum, the co founders of Coinbase, and they had these really cool letterman jackets. They're fun. Wait, is this the. Is this somebody watching the ad?

Yeah.

Okay, let's watch this sound on it.

Anything you need. You better rock your body now, everybody. Yeah, rock your body. So it's a sing along.

Okay.

They like switch up the. It's like a state change, right? So it's like you're paying attention and.

Then you just see Coinbase. That's funny. Oh, that's really funny.

Anyways. Really, I mean, the Coinbase team responded to the criticism and said if you're talking about it, it worked. Crypto's for everybody. A lot of people were not fans, but I don't know, I thought it was fun. I like that they do something different each time. It's like a couple years back to back. The QR code, last QR code in.

The DVD thing was very cool. I think there was a. There was also a dust off about that because I think it crashed the servers because it was so high demand because everyone scanned at the same time or something like that. But I liked the QR code. I like the direct calls to action. I like the introductions, the very clear statements. Unless, you know, I think Apple, Google, like the really established brands, do have the permission to sort of go higher abstraction and just sort of do like these brand films for the AI labs who are trying to push a particular product. It feels like a little bit too soon, a little bit too risky. But I don't know. Anyway. Yeah, yeah, some people like the Coinbase.

Ad, but yeah, it was like very 50, 50 polarizing.

Yeah.

Let's pull up the.

Let me tell you about Railway. Railway is the all in one intelligent cloud provider. Use your favorite agents to deploy web apps, servers, databases and more, while railway automation automatically takes care of scaling, monitoring and security.

All right, let's pull up the Gemini.

I want to see the Gemini ad. I want to see the Gemini.

It went with the full 60 seconds.

Okay, full 60 seconds.

Yeah.

I saw some people reacting to this like it's very clear. It's like what you see is what you get. Like this is the actual experience of using the product.

Yeah, it's next to mine.

And Google does a great job of like Pulling on hard strings like, and.

Integrating with Google Photos.

That makes sense.

Part of the demo.

Yeah.

Features.

And Charlie's bed can go right there.

And a banana killer feature. That's cute.

And look, here's the yard. Oh, we could have a trampoline.

This is actually how a lot of people have delightful experiences with AI. Like, this is like what a lot of families are using AI for. We're doing a remodel and we're using AI for, for this stuff. And you, you put in what you have and edit it and it's cute and like the kids love it. You talk. You always go back to the example of like, make me into a dinosaur. Like, that's delightful.

Come on, Ty, let's go.

That's sweet. That's like. That's just like a nice, sad.

I don't know, I think a win if you're an AI.

Even Ross doesn't like it, though. I'm not going to. That's funny.

Yeah. NA says Google focuses a lot on multimodal. As expected. Yeah. Smart. They had a lot of success obviously with Nano Banana. You should lean in. Totally. Yeah. That ad, it also works if you're running an ad yesterday and you're an AI company and people didn't viscerally hate it.

Yeah, you won. Yeah, no, totally. And also leading into the visual stuff works super well in a visual format, like a Super bowl ad codex. Even if you had a million do overs, it's a hard product to explain. It's a desktop app with a lot of text and it's in dark mode and it's going to write code, but only behind the scenes. And you don't really have to know how to write code, but it's going to write code, which is not very aesthetic on the screen at the super bowl. You could maybe puppeteer a robot arm, but that's probably going to be a little bit harder because it's where you're getting a robot arm and then you have to figure out how to hook that up to your laptop to run codecs. There's an extra step there. It's a little bit tricky, but Google knows how to just deliver. Here's a heartwarming experience, Here's a positive experience. Let's focus on that. Let's not talk about any of the complexities of the technology necessarily in this format.

What else we got? Mike Duda was live tweeting. Live tweeting his reactions. He. Okay, I don't know which of these.

State Farm tried really hard. Okay. DraftKings. Good spot, but not built for the Super Bowl. Toyota continues its long tradition of kind of. Kind of off super bowl commercials. Not good.

We got to talk about the Shvedka ad. We should pull that up.

Silence. Let me tell you about Phantom. Find your wallet without exchanges and middlemen and spend with the Phantom card. Yeah, I. Yeah, Shvetka is the one.

That had a fully AI generated.

They had a fully AI generated.

They came out, wanted to be the first. First company.

We're not an AI company, but we heard there's a backlash. We would love a backlash.

People are calling it the worst ad of all time.

Of all time.

Not even just in the Super Bowl. They're just saying it's the worst ad ever.

I do love when a brand just like rolls up to the super bowl with whatever. Whatever, Whatever their stock ad is. Like, I think that Temu ad was not a Super bowl spot.

Yeah, it was just like a.

It's performing pretty well. Streaming TV on the seat roll. Okay, here's the worst AI vodka ever. Let's watch it.

Like, if I'm watching the super bowl with my kids.

Yeah.

I'm just turning it off. That's right.

Really bad. Also, like, what's the meaning of the robot drinks the. Like the. The mixed drink and it just pours.

Out all in Shvetka. The issue is it tastes kind of like it should be used in, like, machinery.

Heavy machinery.

Like, as, like robotic, like fluid or something.

Yes, yes.

It's rough anyways. Rough. Yeah.

Also, that doesn't. That doesn't even feel like. I feel like if you're going to go all in on AI generated video, I want you using the latest and greatest. You got to be Nano Banana Pro Sling. You got to.

Marketing budget $6 a feed.

Exactly.

No, I agree.

Goldrog. There are very. I've seen these videos where they will take two perfectly generated images and then they just use clang to like, interpolate between them and they actually look really cool. Especially if it's like a car or it's something that like, AI is particularly good at generating and sort of like using it as a transition from a drone to inside of a car, like first person view. Like, there are really cool and innovative ways to use AI imagery in a way that maybe it stands out as like, oh, that's obviously AI, but it's cool. Like, if you're going to do the AI thing, like be Harry Potter, Balenciaga, like, do something that is iconic and interesting and inspired. Don't just be like one ad that we would have just used CGI for last year, and now we just use the AI for it. Did not hit. What do you think?

Well, I was saying we should. We should watch Ryan Peterson's Flexport ad. Some are saying it. Some are saying it might be AI.

It might be AI. Okay, let's pull up Flexport's ad. Let me tell you about Applovin. Profitable advertising made easy with Axon. I get access to over 1 billion daily active users and grow your business today. Let's pull up up Ryan Peterson's Flexport ad. Okay.

Looks real so far.

Seriously, dad, how did all these jerseys get here?

Well, kids, let me tell you about something called a global supply chain.

First, jerseys are manufactured, boxes are packed, and logistics company Flexport takes it from here. Then containers are loaded onto cargo ships.

Or pulled through the ocean by 100 pirates.

Not exactly, honey, but if you want speed, Flexport does coordinate airframe and fighter jets.

Actually, buddy.

Flexport then gets the jerseys through customs, back on the truck, and then the.

Superheroes fly the trucks to the stadium.

And lightning zaps the jerseys onto the players. Yeah, we're being sarcastic, guys. This is entirely AI Entirely AI but.

This is so good.

But it has the nailed the feeling of a Clash super bowl ad.

Totally. And it's all things that. Yeah. Like the budget to actually film. And to be clear, you're talking about millions and millions of dollars.

Ryan didn't actually run this.

No, he didn't. But it still got a. It still got a bunch of views, and it was very fun. And it's interesting because he's actually explaining the Flexboard business, but in this funny way that you keep asking, like. And then the lightning comes down, or then the superheroes fly. Like, so you're explaining it to a kid, but you're like. Because if you just sit there and you just do the vanilla, like, Flexport explanation, that's gonna be a little dry.

Yeah. What's interesting is now I feel like. I mean, I'm sure smart advertising agencies are already doing this. Where it used to be that you would. A company would say, hey, I want to run a Super bowl ad. They'd go and talk to a bunch of different agencies. They would get a. They'd do an rfp.

Yep.

And. And then. And then the agencies would, like, come up with some concepts that they would basically script out. Maybe they add some images or whatever. Now it feels like the agencies have to actually make a V1AI V1AD and say, like, here's our five concepts. If you hire us, we'll like narrow in and actually make it hire. Hire actors, maybe not use entirely AI.

Yeah, but I thought that was really good and I can't believe. I think he made that like himself. I think that he wrote the script and concepted everything and actually prompted everything, which is pretty remarkable. And I think it all flows from the idea of having a viewpoint, having an insight, having like an idea that makes sense for the Super Bowl. Maybe Flexport should just get this, like add a layer of polish and then run this.

Orin went so hard. What do you say on this Fed?

Okay, yeah, let's read it.

This was the first AI generated Super bowl ad ever. A milestone for truly no one. The worst of agencies pitching be the first that no one but the agency cares about much. Rather would have watched a cast of characters workshop how they're actually going to become the top vodka of 2033. That's a cool concept. So there's a lot that they could have done.

And yeah, I mean the actual character design of the Svedkabat is very off putting and very. It's just very plasticky and it doesn't even feel like a modern humanoid design. Like if you put this next to like a 1X or a Tesla Optimus, like you're just. It feels like this is what like the 90s called. They want their robot design back. Like that's what it feels like like to me. Anyway, let's move on to some other tech ads. First, let me tell you about console. Console builds AI agents that automate 70% of it. HR and finance support giving employees instant resolution for access requests and password resets.

Rippling ran an ad. Let's pull it up.

Parker Conrad, an absolute dog.

Tim Robinson.

I love Tim Robinson.

He's fantastic.

So funny, but nasty to me.

After years of planning, I'm finally going to get my revenge. I'm going to trash the town with baby. Unfortunately, baby Breck hasn't been onboarded yet. We're still waiting for his laptop and health benefits. Baby Breck, he doesn't even know what a laptop is.

He's not onboarded.

No.

And his harness wasn't approved by finance.

They're stuck chasing down receipts.

The harness was gonna dangle him from the chopper.

When we bring him into the town.

And drop him in, it's like so perfectly like he's tripping over his words.

See, this is funny. This is really, really, really funny.

Yeah.

If you know Tim Robinson.

Yeah.

But my concern is, I mean, I don't know how. I honestly don't know. How many?

He's pretty popular now. I mean his show on Netflix is pretty big. You should believe it's like he's definitely funny. And I mean that style of comedy can just be. I don't know, it is a little.

But there's so many Tim Robinson sections where he is in a workplace setting talking exactly like that.

I just love, I love when tech companies and founders become like patrons of the arts in the sense that like, like this is probably like a pretty big payday for Tim Robbins is Robins or Robison Robinson. Robinson. Tim Robinson. Love him, wanna support him. And it's just like it makes me like the brand because I'm just like, I like that he likes that they like the same person as I and they like, you know, got him to.

Do some funny guy. It's so funny because I've actually laughed at Tim Robinson content so many times seeing this. I don't even know what is baby Breck.

I think it's just like a character they made up. A character that they made up that's like a Godzilla type that's gonna wreck the city. And these are evil villains and they're planning but they need to onboard. We saw this ad in the.

I'm gonna dangle him from the chopper.

Yeah. It just assumes you know all the lore of this campaign that like who knows where it came from? But he doesn't even know how a laptop is. What is what a laptop is? You don't even know what a laptop is. The hashtag. Yes, rippling exactly one post with this hashtag. Killing it. This is so funny.

But yeah, I was waiting for. I know a number of companies have been circling Tim Robinson being like be the first tech company to work with Robinson.

Totally. I really do think that there's some alpha in like if you're just like on the gold box.

55 GPUs, 55 data centers, 55 Neo clouds.

That's a great reference. That would have been incredible. We need to get some hyperscaler on that one.

Cadillac ran a full one minute ad.

Super bowl ad.

Let's Vicenzo watch it. Former guest first business at speed.

Let's tell everyone about public.com, investing. For those who take it seriously. Stocks, options, bonds, crypto treasuries and more with great customer service.

That's right.

Okay, let's watch the Cadillac F1 video.

We choose to go to the moon in this decade and do the other thing.

I don't think so. Probably just CGI. I'm so excited for Cadillac F1. I think I'm the only one. But because they are hard, I'm very excited for it. See that? Yeah, that could easily be cgi. It could easily be AI, but sort of depends on who you put on the team.

Serve to organize and measure the best of our energies.

And kills.

I like it. Building the car like a car breakdown.

It's team sport. It's an engineering challenge.

It's like the moon landing. Is that what they're saying? TWG is the. Is like sort of the presenting sponsor of Cadillac F1. I like the livery. It's hard when you don't have a really bold color like McLaren orange, Ferrari.

Red, black and silver is the.

Black and silver is cool. But it's a little understated for the grid, so it doesn't quite jump out at you the way an Aston Martin green does. But over time, hopefully they can kind of dominate that. But Mercedes sort of has carved out like the silver line a little bit.

TWG has range. They're a unique holding company, strategically investing in and operating businesses across investment management, securities, AI and technology, finance and corporate lending, merchant banking and private investments, and sports media and entertainment.

Thank goodness.

There we go. Oh, I think they can do it all.

I think somebody from TWG once randomly.

But Apple. Parker says amazed. Apple's finally going for the Spotify jugular, promoting the ability to switch easily to Apple music during the Super Bowl.

So the Apple Music brand reach during the Super Bowl, I think was number one or number two right up there with Budweiser. It did very well. I saw it winning awards. I haven't seen the ad. Let's pull that up while I tell you about 11 labs. Build intelligent, real time conversational agents. Reimagine human technology interaction with 11 labs.

While we're pulling that up, Apple music was also. They also presented the halftime show.

Yes, they presented.

They had a crazy box. They hand over.

Yes.

And break it down.

What happened? So we're in the stadium and they do one of those. Let's cut to the box. They got the long lens looking inside the box. And who's there? Tim Cook. Everyone knows Tim Cook, so they didn't need to give him a little tagline. But there was this other guy whose name was Dave Grohl, and I guess they were worried that people wouldn't know who he was. So they put the little Dave Grohl tagline there. But they didn't need to give one for Tim Cook.

So Tim Cook was aura farming.

Yes.

His box was crazy.

His box was crazy, actually. And it was funny because I'm of course, joking, but. But Tim Cook was like, sort of in the shot, but on the edge of the shot. And kind of like the photographer or the videographer, the cameraman was like, sort of moving Tim Cook in and out of the shot. Really focused on Dave Grohl, who was like, putting on a fun performance, chugging a beer and being funny. And Tim Cook was just off to the side and I was like, pan over. Show us the big guy. Show us the CEO of Apple. That's who I care about at the Super Bowl.

While we can talk about the halftime show from our point of view, since we were there, I would say the energy, the production value was insane. Like, very, very, like, cool, unique looking setup. It was really funny because I was like, looking away while they were setting up, and I looked down. I was like, they basically created Puerto Rico. They recreated the entire island.

Plants and people in plants.

And the plants were so funny because it seemed like the people in the plants couldn't see out very well. And so they had, like air traffic control people that were like, moving the plants around, like, yelling. You could tell they would be forward. They would be getting, like, right in the face of the plants and yelling like, go.

It was crazy.

But the production value was insane. The energy in the stadium was completely dead.

Dead. No one was dancing. It was a very.

Actually, no. Dylan Patel was dancing.

Yes.

He was going. He was like, yeah. It was so funny because he was gonna walk down and sit next to you and then he was like. He's like, actually, no, I'm gonna stand up here, I gotta dance. And he just started dancing. Dylan held it down. Everybody.

Yeah. I think. I think the audience of people that. That go to the super bowl, it's a lot of, like, older folks and they were just not into the new kid on the block, Bad Bunny. But it really was, like, a remarkable production when I think about, like, what we do here in terms of live production. Cameras, chirons, all sorts of stuff. So much goes into it. And to do that and say you have 20 minutes to build a city on the field, and then you're going to walk around with wireless cameras that look like Arri Alexas. It looks like a movie. It's like perfectly shot. You don't even see if you're watching from home. You don't see that you're at the super bowl until, like a couple minutes in. And then if you're at the stadium, you don't see what's happening on the performance because it's not on a stage. It's like in this interior zone. Yeah. Like, you can't even tell if that's at the super bowl until, like, they start panning out and showing you the full thing. But so much layered detail in the performance. So many Easter eggs going back.

Yeah, it was interesting. It was very much felt designed for television.

Totally, totally. And that's what the super bowl has become. And the super bowl halftime show has become like. It is an opportunity to bring in a new audience who would not necessarily be watching the super bowl, but they come for this, and then they. Maybe some of them drop off, but some of them stick around, and it's just like a new introduction, but really, really crazy. At one point, he climbs up on a. On a telephone pole, which, again, was constructed 20 minutes before. There's a camera that's, I think, battery powered and wirelessly streaming. And so you have talent who's, like, a multimillionaire. He can't fall down and get injured. He's at the top of this telephone pole. There's a camera there that shows him. And then fireworks start going off. And so the combination of, like, fireworks, camera equipment, and elite talent that cannot be injured was like, a really remarkable just, you know, magic trick from the production team. And, of course, there's a whole bunch of fireworks and whatnot.

I was a lot of people obviously, you know, somewhat divisive online because there are people that are frustrated because they couldn't understand what he was singing about. But from a purely commercial standpoint, I think it makes a lot of sense for the super bowl specifically because a bunch of people that wouldn't have watched the super bowl, that are international, are gonna tune in.

So.

Understand.

I was in the moment thinking that it'd be great if Apple had their new translating AirPods and you could actually hear in real time.

Yeah, you could probably throw those on anyway.

Yeah. Surrounded by a bunch of AI ads, and then you have this insane IRL experience.

Yeah, it was very much like a return to practical effects. And there's all these cameos and details and. Yeah, just. Just. Just really remarkable to watch the video. Not so remarkable seeing it from the stands. But anyway, do we have the Apple ad available?

No, no, no.

They didn't even run an ad.

It was, like, integrated into.

Oh, that was the thing. Oh, okay. I thought that.

Let's go to the AI.

AI.com.

Ai.Com.

AI.Com. you wanted AI. You know where to go. AI.com they spent $8 million for a super bowl ad. Oh, yeah.

Bobby says, and the NFL is in the midst of a big international expansion, so. Playing a bunch of international games.

Totally. Oh, yeah, yeah. They play a game in Mexico, London, Canada and stuff. Trying to bring the sport of football to the world. What do we have here? AI.com.

It'S only 30 seconds long, but it seems to have.

What's the reaction to that? Did they have a voiceover in the actual ad? I don't know. You can just kind of see it, but it was big. Duca says AGI is coming and it's a 503 service. Temporarily unavailable screenshot because I think they got so much traffic, but it was a very odd question. So you go to AI.com it asks you first to log in with Google. So you have to give this new website access to your Google, which is like insane.

Which is a lot, because they're wrapping openclaw.

Yeah.

Which. It's an openclaw wrapper. And so I go in there, that's a new company. And I was like, I don't have a single. You know, I have multiple Google accounts. I was like, I don't have a single account that I'm willing to just connect to some. Some app that has existed for.

Yeah, right there.

Yeah. So we can't even find the ad online. I can't find it anywhere.

Did you already.

Chris, the founder didn't. I don't think.

Yeah.

We're going to try and understand the bigger project because it seems like the domain buy was maybe longer than the build out of the product because it seems like it's an open claw fork of some sort.

And.

And the domain name was maybe $70 million. They say it's $500 for a vibe coded site. Probably fake, but funny.

Okay, we got the ad now.

Oh, we do. Okay, let's pull it up. Let's pull up the ad. What else we got?

Okay.

Do you think this is AI generated? Because if this ran first before Svetka Svega got scooped. Now this looks like maybe traditional motion graphics. I don't know. It's hard to tell these days. AGI is coming, get your handle.

Now you're gonna want. You're gonna want.

You know, I am a sucker. I'm a sucker for locking down.

I know. I was interested.

What are they doing there?

Trying to get people to Mark.

Oh, Zuckerberg. I was thinking Mark Chen.

Yeah. So.

What?

Yeah, I mean, the whole thing is, I invited Chris, the founder on. We'll try to get him on to understand more. One of the things. So he had A lot of success. Clearly buying crypto.com.

Yeah.

Crypto.com arena here in the arena he basically owns LA. But the difference with crypto is building an exchange is somewhat a commoditized platform, at least in its simple state. You're like, I want people to buy crypto in my app, in my exchange.

They come here, they make their crypto.com.

Okay.

The difference with AI.com is like, I think to actually compete in this domain, your product's going to have to be insane because you're competing with OpenAI, Gemini Anthropic. You're competing with Perplexity. There's so many thousand different startups and so I think this ad is not going to give people a lot of the whole experience here is not going to give people a lot of confidence. One getting people to like sign up, connect. They're not even giving them an opportunity to just create a normal account.

Yeah, that is wild.

One. That. That's crazy. I would have just created.

It's got to be way better if you do give it access to your email because then immediately it knows who you are, what you talk about, what you shop for.

Yeah.

But also running a Super bowl ad and only having Google, that is crazy.

There's so many people that are on Yahoo and Outlook and you know, any, any other service. Like there are people that are like, I'm an icloud guy. You know, it's like rare, but it's still like millions and millions of people. Like they're just not in your target market. Very odd.

Yeah.

So anyways, where D2C brands, B2B startups and AI companies advertise on streaming TV, pet channels, target audiences and measure sales just like on Meta Jordy. Do you have anything else on AI.com?

Adam Strong said. Did you already cover this? He said 8 million for Super bowl ads, 70 million for domain name 500 vibe coded site Cloudflare. Basic hosting. Priceless.

Yeah.

The app did not or the website did not feel polished.

This is Paul E. Williams post with 18,000 likes. Love the idea of running a Super bowl ad that just says go to my website then not being ready for the traffic. Yeah. I mean you can cache the site locally, everywhere and have it be super static and bury the call to action. You can capture someone's email in the browser and then just hold it there for an extra minute while you give them some other experience. So you're like pushing the funnel so your server isn't on fire to actually usually the thing that brings down the sites deeper in the back end, like, the database is not performing or whatever. It's very rarely, like, I couldn't get them HTML on time. That's pretty solved. And so it's very, very odd. But I don't know, at the same time, reflecting on the crypto versus AI thing, there was never really a crypto tal like there was in AI because there was never a moment where it was like, oh, like this platform needs this particular feature and only that person, that scientist knows how to do it. It's like there were important features that got built, but a lot of them were done open source. And then the value of the platform was like, liquidity, the scale, the features, the compliance, the regulatory, like, all of that. And you can't just like, poach one person and then all of a sudden be like, I have coinbases stuff. And so very interesting question of, like, it's maybe not crypto.com. how do you build something that's. That's as performant as Perplexity or Claude or Gemini without that team? Or maybe you're ready to invest in that team and.

Yeah, or, I mean, if the AI.com brand I've never heard of, if you can get through to the site requires you to authorize login. And like, I read through the private. I read through the, the privacy policy and the terms, and I was like, like, definitely not.

Yeah.

And. And so, yeah, again, what's. What's the point?

Yeah, it's interesting that AI is somehow more intimate in many ways. If you're talking about personal assistant, personal AI, then like, I would trust crypto.com with my money in many ways because I'm like, okay, if I buy a Bitcoin, like, there's a chance that it's an offshore exchange, it goes bankrupt, but like, it's contained to like, whatever you're doing there. Whereas if you're letting an AI into your.

Yeah. One reason it has access to crypto.

And all your bank account.

Yeah.

One reason to trust crypto.com is it made it through the last cycle.

That's true.

And I think a lot of people felt like, hey, I haven't heard. I mean, the company, I think was mostly, pretty much bootstrapped. I think it's mostly, mostly founder, you know, entirely owned by the founders. So they've, they've been stable in that sense.

But anyways, very strapped companies, you know what I'm gonna say, Say turbo buffer, serverless vector and full text search built from first principles and object storage, fast 10x cheaper and extremely scalable.

Before we talk about Jason Carmen's. Yeah.

Let's run through some ranking.

No, we got to jump in with some breaking news.

Oh yeah, what we got?

Apparently OpenAI actually rolled out ads in chat GBT today.

Today.

Let's go. Finally. Should we bring the gong for them?

We beat him to it.

We beat him to it.

We beat him to it.

We front beat him to it.

We front.

Breaking news. Ads have officially launched in ChatGPT. You can go to the free tier or I believe the go tier as well and Enjoy ads in ChatGPT. You'll see probably enhanced rate limits and extended functionality more. Can you.

Can you work to try to find to get served an ad?

Yeah.

I want to see what they actually look like.

You have to look down or create a new account maybe.

Yeah.

New email.

New email. I would love to know how much do they exaggerate when you see an ad? They could do a blaring red background. It's very obvious this is the ad and then this is the content. Will the ads be even at all related to the content? At all related to stuff you've seen in the past or is it just going to be completely random? We'll see and I'm sure we'll see a lot of screenshots in the timeline.

Yeah. I think from what I understand it won't be related to the content because people are going to be annoyed at that. I would expect it in the future. That is. Right. That's just sort of the natural evolution. Right. And so that's like one, one, one way to justify anthropics, you know, kind of like original ad which is like, hey, you asked about this thing, you asked about how to get healthier and we offered you this drug. Maybe that's not that. Not super aligned. But in the meantime, I think it'll effectively be display ads based on your interest graph. Right. What they know about you based on what you've searched in the past.

Yeah.

Interested to see this? I'm not going to say anything more. Anything more positive because your audience captured. Yeah. Everyone. Everyone. It's so funny. I went from like last year being.

Like, jordy's a hater.

Hyper critical hater. Hypercritical. OpenAI was like I was steel learning everything. Yeah. John was the steel man. I was like, sora is going to fail.

Yeah.

Atlas is going to fail. They're launching adult mode. I'm like the, like the, the most.

1.4 trillion is too much like where's.

The most like vocal critic? I'm talking about about, like, you know, throwing shade at Sam's answer on BG2, saying, like, that was the worst possible answer you could have given. I have zero confidence to. Suddenly I say, like, I just popped my head up at one point and say, hey, I think the ads are kind of deceptive. Parker says, the amount of athletic greens and squarespace hats, I'm excited for it.

I'm excited for it.

So anyways, we'll. We'll see what Tyler can dig, what people get.

Let's also tell you about cognition. They're the makers of Devin, the AI software engine. Crush your backlog with your personal AI engineering team.

Yeah. Brian says, to be fair, SOAR and Atlas have kind of failed and that. And yeah, I said that basically the day they launched, I was like, I don't. I think SOAR is like a cool way to generate a video. Yeah, I don't think it's going to work.

We talked to Sam about that too. People are using it more as, like, generate something to send to a group chat. Generate something that goes onto Instagram or something like that. Or even, even just, like, use it as a tool in the tool chain for certain things.

Yeah. Anybody that. Anybody that. Anybody that said. Anybody that said, oh, oh, you're just. You're just like pumping OpenAI. It's like, I'm sorry. We literally had Nick on the show last week Anyway, so trying to provide a balanced point of view, let's head over to Jason Carmen. He built a tier ranker.

If you want to share where you think each brand landed, you can go to story.insuperbowl and you can tap an ad and then you can tap to rank it and you can decide where you want to put Budweiser. Where does Pepsi go? What do you like? What'd you hate? And you can take a screenshot of this and share it. And Jason, of course, he makes cinematic ads for science and technology companies and he will be sharing his rankings tonight. I guess he probably already put them out. Let me see. I gotta find these. But anyway, go check it out@story.inc he.

Loves a good Serena Williams partnered with Ro.

She's been partnered with RO for a long time.

Yeah.

But partnered for a Super bowl ad, which is a new level up. Let's see what she had to say about roe.

I'm on RO.

34 pounds down on GLP1s healthier on row. Supported on row. FDA approved GLP1 options now even in a pill. Weight loss expertise I trust. I'm moving better on row. Extremely clear Yeah.

A lot of repetition.

It's just like, this is what it does. I'm famous. And here's exactly the value prop.

Yeah, good.

Said on row at least. At least five times.

Yes, yes. You don't know the brand by the end of that.

That's good.

Sticking with Rose kind of history and direct response. Yeah, totally. Growing up as a brand, they're like, if we run an ad, it can't just be.

They were doing TV direct response super early. I remember seeing those ads in bars and random TV spots. They were doing NBA, mlb, all sorts of stuff.

They mentioned the pill. Do you think that's an actual. Do they have an actual partnership?

So it's the same thing. Front end to Novo, front end to Wegovy.

Yeah.

Yeah. To the big pharma companies. They are. I think that they have fully stayed out of the compounding space or at least a little bit and really let them just be this like, you know, like seamless front end to deal with doctors. But anyway, let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces. And now with AI agents, let's pull.

Up up the core weave ad.

I love core weave. I'm a big fan. This ad. You can't controversial anything without AI. We were memeing this too. We were like, you can't. You can't spell Claude without a D.

A as in aerospace.

I like this transition academia. This is cool. N is for nanotech, navigation, neuroscience. Y is simply yes. Yes to predicting storms to keep people safe.

Yes to empowering.

There's a lot of good stuff. T is for translation, trading, transcription. The tools teachers need to educate each kid in their own way. That's a weird transition. H is for healthcare to help keep humans healthier.

Home automation and robotics.

I is for ideas and the power to grow them exponentially. N. This is for now the time to never say never.

G is for the good we could do with AI.

Genomics says sir, it's a bubble moon.

Boots gravity for anything.

Oh.

Anti gravity boots. That's what it is.

Okay.

The chance we were built core weave, the essential clapper.

So the chance a rapper.

Yes.

Part felt super random. I haven't seen or heard from chance rapper in a long time. Yeah, I think what they were trying to do is they're probably looking around and saying data center are the most hated thing in America right now.

Yeah.

We got to make the case for data center.

Yeah. This is. This Is what my.

But I would have liked something like potentially even more direct and like tactile.

Yeah.

Just kind of like rambling on. It was. It wasn't again.

Yeah.

I don't know. It wasn't. It wasn't like, it wasn't. It didn't really like pull. Pull you in.

Yeah.

Just kind of like background noise.

Yeah. There was one. There was one part where Dan says.

Chance the rapper WR A P P E R. That would have been good.

Yeah. There's a line in there where he says something about predicting the weather.

Add reprice their CDS spreads and bring down the refinancing cost.

We will find out for sure. No, the business is doing fine.

Up 8% today. Total victory.

But no, no. So he's doing this alliteration thing where there's different, like it's t. And then he mentions predicting the weather, which I think is cool. And I think that's actually like a great story of AI like, you know, oh, there's a flood. Like, no, it's not okay.

But then just because we've had weather forecasts forever. So if you're trying to justify data center development and say like, hey, AI is cool. Actually, you can predict the weather. People are going to be like, cool. The thing that I've been able to turn on my television and understand.

Yeah.

The newsstand tells me, okay, well, I don't know. My point was that each letter had five or six things instead of like one concrete thing. Maybe you want to narrow it down and be and unpack it. Actually make the case. Because clearly just mentioning that you can predict the weather with the AI didn't hit with you. But maybe if they unpack it with three or four sentences, they can convince you a little bit more. And then part of it is, I think a lot of people don't care about trading.

Yeah. Part of it is like, doesn't Core Weave have like a handful of customers that actually matter?

Yeah, and they're also like deeper in the stack than most companies. It's like, there will be a weather prediction company that uses a. A foundation model that is run on coreweave, but you won't see Core Weave powering that. So even if you are like, wow, I actually got a flash flood warning early. I moved out of town and it was amazing. Predicting the weather's IBM Watson.

Yeah, Cam, there probably was an IBM Watson super bowl ad at some point that was like, in the future we will be able to predict the weather.

Yeah. Oh, well. Oh, well. Yeah, this one was. This one was not that well received. Let's head over to Dunkin Donuts. First, let me tell you about Plaid. Plaid Powers. The apps you use to spend, save, borrow and invest securely connecting bank accounts, move money, fight fraud and improve lending. Now with AI, Head over to Duncan Goodwill.

Duncan version of Goodwill Hunting was made as a sitcom with a real genius in the lead and some other.

So this is AI as well, right? De aging the munchkins in the Fibonacci sequence. I got a genius working for me. He's such a genius.

Then why'd he put ice in his car?

Coffee, huh?

Come on, Chucky. I'm just Will Hunting. I'm not a genius.

I will marry the first man that.

Can help me with the Fibonacci sequence. How you doing?

Smooth. Don't you have a girlfriend?

We're on a break.

I don't need her.

I still get everything I need right here at Duncan. Hey, kid, if you're still single, doing.

This Boston and working for Duncan when.

You'Re 50, I'm gonna be very disappointed. Isn't that your girlfriend? It's like Cameo Central.

Yeah. Very chaotic.

Well, this is my new boyfriend.

How you like Tom Brady?

I'm Tom.

They really were spamming. The cameo.

How many cameos do you want? Yes.

Yeah, the AI CGI was a little rough, like. It was.

It was reasonable. The haircut is what's hilarious to me. The putting Ben Affleck in the Matt Damon haircut. That's very funny, but not a great pitch for Dunkin Donuts. I don't know. It was fun.

Yeah. What's the takeaway? Think about Duncan. Sometimes all that matters is you get people to think about Duncan.

They had fun with it. I don't know. It's interesting. The Squarespace was particularly odd. I think the Squarespace one was weirder. To me, that was.

I like that one.

You like that one? Okay, we'll debate this while I tell you about Labelbox, RL Environments, Voice Robotics evals and expert human data. Label Box is the data factory behind the world's leading AI teams. Continuing to watch the Squarespace ad with Emma Stone.

Oh, this is a Genspark ad.

Oh, what's Gensp you should do, Stu? Monday, February 9th.

Finish his slide deck.

Okay, now you're not working from home.

Hey, you put on pants.

This is practical.

Response.

Dial in this space.

Tune went so hard. Parker's got it done. Circle back on out of here. Let genspark automate your work and take a day off.

Okay.

AI for Enterprise.

Jensmark Unicorn Raised around half a billion to date. But, yeah, competing. I mean, effectively competing with Copilot. Right. Copilot also ran an ad for Excel. Let's pull up the Squarespace ad now.

Yeah, this is one where they just left the. Her Go crazy. So find your domain name. She wants emma stone.com, and it's unavailable. So she breaks the laptop, and there's a whole bunch of laptops that have already been broken. So she's furious. She gets another laptop. It's still unavailable. She smashes it again. Now, I was thinking that this. Get your. Oh, this is a shorter version. So there's another version where, like, there's this rollerblader who comes in and, like, continually, like, delivers her new laptops. She tries multiple times. And I was like, oh, I know where they're going with this domain brokerage. They're gonna help her buy emma stone.com from whoever owns it.

That could go down. The feature that works where you pay $70 and then they just take your money and they're like, yep, couldn't get it.

Sure. But maybe their pitch is like, it works. Now, we're good at reaching out to the people that have the domains, and we will. And so. So we got Emma Stone, emma stone.com, you can go there right now. And we brokered it for her. But they don't. They just end and they're just like, get your domain before it's too late.

I don't think this product will ever work for everyday consumers because they'll tell you they'll literally reach out to, like, google.com and say, like, hey, if you give us 70 bucks, we'll try to get google.com for you.

Obviously, they need to have some sort of layer of interaction, but you could imagine that between. Between AI agents, a small salesforce, really top tier people. Like, you could. Squarespace could have a serious domain brokerage that could do 80% of requests. Especially if you're Emma Stone and you have a budget.

Yeah, it's kind of a rough. The challenge with the ad is that a lot of people that do go on Squarespace and try to get a domain and make a website will have that exact experience.

Yes.

So you're kind of like advertising the experience you don't want.

Yes. Also, if you go to emma stone.com, it is her website, so it doesn't.

Quite match fake news.

They do a good job of saying, get your domain before you lose it. They tie it back to the super bowl ad, and it's the same video. Like, Black and white texture. Yeah, you can see it. So it does complete the journey. But it's a little bit odd from the storytelling perspective because in the video she was unable to get emmastone.com, but then somewhere after the ad ends in the castle in the island, she gets the domain and spins up a website from Squarespace. Anyway, beautifully shot ad and fun that they went black bars, I believe. They didn't do a full widescreen ad. They did a little bit squarer, which I think is like a nod to Squarespace maybe. I don't know. We will see.

We watched the Hims and hers Super Bowl.

Oh, they launched one too. Interesting. I thought Roe would be the only one. Okay, let's pull it up.

Dropping it in here.

I will tell everyone about Lambda Lambda is the super intelligence cloud building AI supercomputer for training inference and scale from one GPU to hundreds of thousands. And then we will pull up. I like the thunder. The cloud.

We're adding different effects today. It's amazing.

Let's pull up the Hims and hers, apparently. What is this one saying? Live longer.

Okay.

They're taking shots. It buys more time. The wealth gap is a health gap. Okay? The rich have health care that comes to them. Custom formulated peptides, specialists on call and preventative care before democratizing health of everything.

It's pretty wild. A bunch of biographer testing.

Oh, okay. Diagnostic testing. I'm getting the EDM gone. This is a lot of content loss treatments that can be microdosed to fit your goals. Microdosing testosterone hormones to keep you feeling great and early cancer detection through a simple blood test. Okay.

The same science says last ad before bankruptcy. Stock's down 17% today because of this ad. No kidding. It's down because they got a very angry.

The admin.

Right.

FDA came out hard against them Friday. Yeah, we had tj. TJ Parker came on the show, broke it down on Friday. If you want to understand the him situation.

And there's an article, Washington Post, Eric Topal shared it. He says super bowl spot will promote a cancer test that can produce false results. I mean, every test can have false results. The question is like, is it out of the bounds of what the FDA approves? Even an off the shelf pregnancy test can have false positives. That's why they say take multiple. But it does seem like the FDA is not a fan of the move fast and break things approach in this particular category.

So roll, move fast and disregard ip.

That's another one.

Disregard studies.

It's a rough time.

Make your Own drugs.

Well, our first guest joining in just a minute. But Michael duda continued his reviews, which I'm enjoying at this point. Homes.com no ro not entertaining, but it will be massively effective for ro's business. I agree. It was like, okay, I get it, I get it, I get it. But that's the point. Wix harmony. I think they meant to have that on grammys last week. OpenAI expected better from the company synonymous with AI. So Michael Duda is going all over the place. Mr. Beast also collabed with salesforce on a massive super bowl ad that had a lot of hype. Giving away a million dollars. There's a lot of cool CGI in here. We had a chance to meet Mr. Beast at the game. What we're hoping.

Well, he's a ramp customer, so he came by the ramp box yesterday.

That's right.

Yeah.

We're very excited for him. Well, without further ado, let's bring Jason in from the restream waiting room. Well, a little ado because I'm going to tell you about fin AI, the number one AI agent for customer service. If you want AI to handle your customer support, go to fin AI.

Sorry.

Jason, welcome back.

Hold that thought.

Hold that thought.

It's a great product. I love it.

Thank you.

Well, how are you doing?

I'm doing well, thank you.

Did you watch the super bowl? Are you a super bowl guy?

Not really, actually.

Did you see any of the ads?

Once I saw the tv. Once I saw the end, the ad, I of tuned out.

Yeah, yeah. You're like, okay, we're three hours before the game. Job's finished. I'm good.

That's right.

And you flew to San Francisco to see it because it was a regional buy, so.

Exactly.

Of course.

Couldn't miss it. Okay. Couldn't miss it.

Well, well, the news that we wanted to talk about.

Yeah, we wanted. We were texting a little bit about the new Ferrari ev. Yes. And I think there's nobody that I was excited to talk to more than you about it. Just given your have always appreciated your. Your personal taste in cars and just also in design choices broadly. And I think this is such a. For anyone that missed it, Ferrari is making a. An their first electric vehicle. The loose luce.

Yeah.

Am I saying that correctly?

Sounding Italian.

Yeah, it sounds Italian. I thought they. But it's effectively like we don't know exactly what the car looks like, but it's like meant to be kind of a daily driver. It looks to be like some sort of hatchback. Something like that so not as aggressive looking as the GTC4 Lucho or the FF, but looks fine and highly anticipated. And then just on the interior, they partnered with Jony. I've. And the love from team. So before I share are my thoughts, I'd love to kind of like hear how you've been processing it.

Yeah.

So I saw it this morning like everyone else did. I think they came out with it this morning or Nelson this morning. And obviously it's a demo. It's a 3D demo of kind of what this thing appears to be, at least the video was. And I think the first thing you go is that, well, that's at least very fresh. Like, it's about time. It feels like car interfaces have really gone mostly in the wrong direction. Just like black glass screens and everything's touch screen and they're starting to bring some switches back and dials back. But clearly this feels a bit more considered. There's more tactile stuff, there's more real stuff integrating with. With virtual stuff. And of course a lot of the design choices were, you know, very beautiful. Transitions were lovely and there's no lag in all these things. Again, this is a demo. It's hard to say what this will really be driven on, but it looked great. I think it looked great. There was a really nice. I would say the thing that really stuck with me was the integration between the virtual and the physical switches that when you move them up, like then a digital display comes up, little buttons on the top right corner, adjusting some display stuff. Really, really nice. A lot of inspiration from other brands I saw too. I kind of felt like there was a little Aston Martin thing going on here. So the DB9 era of Aston Martin's, you take this key, this glass crystal thing, and you push it into a slot to start the car. And I think that's what they were showing here is they put this.

I like pushing it down. Like you can imagine the experience of pushing the key in, having it lock in place, place and then tapping it, because it is. It's just a nice experience. Still, there's. I mean, I feel like so often I'm like putting my key in random different places at different times and it's nice to just have a place in the car, push it, lock it in, you know where it is and to leave it there.

That's the other thing.

Yeah, exactly.

It probably is magnetic, so it probably snaps in, snicks in and push down. But that's kind of an Aston derivative. And also the way the, the dials when they light up, they, they go counterclockwise is also something that Aston was doing.

So.

So Johnny, I being British, I wonder.

If there's a little bit of a.

Call out to Aston there.

Yeah, good point.

But it's cool.

I would say though, one thing that no one knows is what the outside of the car looks like yet. And I think great interiors need to be matched up with great exteriors. The best interiors in history, I think were complete ideas. You think about like the. And I'll go off on this, so you tell me to stop. But like the VW First VW GTI Mark 1 Special Interior, it felt like it fit with the exterior. The like Lancia Delta Integrale is another car like that that just like the interior and the exterior just were perfectly built together. The Porsche 928 feels like the interior had to be in that car. It couldn't have been in a different car. And so it's hard to evaluate with interior design. Like, this is like, what does it look like fit in? Does it fit?

Totally. That's the, that's the crap. Yeah, that's the big unknown for me right now because I look at the mock up that everyone's been sharing and it's this hyper modern sporty hatchback. And then on the inside you have all this analog buttons and you have what looks almost like a resto mod, like heavily inspired by some of the legacy Ferraris from traditional classic Ferraris. And so right now when you look at Johnny's work and you look at the mock up, it doesn't feel cohesive at all. And the experience when you get into this sporty hatchback and then you have like an old classic Ferrari kind of like racing inspired steering wheel just, it looks strange. So I think we need to withhold judgment there. Kind of going into some of the more negative, like I never have. Something I've really disliked about some of the more modern AMGs is this like iPad, this like iPad interface. And you see this in some of the Astons too where it's, it's like, hey, we didn't know where to put this screen. So we're just going to like bolt it on to the front dash and like put it out in your face. And the experience of being a car in a car is just so incredibly nice when everything just kind of like blends in in the front and you're able to just focus. And I don't want, I don't want a screen kind of pushed into my face. So some of the like, it Feels like my big concern with this project is that Ferrari is not Apple. Apple. So Johnny doing. Johnny doing work.

Doing.

Doing product design work and building products at Apple, one of the greatest hardware manufacturers in history. Probably the greatest hardware manufacturer and consumer electronic manufacturer in history. And then you put him in at a team at like Ferrari and then like, it almost like he's going to be handicapped to some degree by Ferrari's own internal capabilities, which, like anybody that's owned a Ferrari is like, I never in my Ferrari ownership experience was I like, oh, Ferrari is just exceptional at making electronic interfaces. Right. Like, they can't even. There was like a very long period where the buttons would be sticky after a year. Right. And that was just like an ongoing problem that they couldn't seem to solve. And so my concern is that Johnny's probably deeply handicapped not just by the Ferrari team, even though they're so great at making vehicles overall, but also from regulation. Right. And that, like, just because you want to do something doesn't mean you can do it, even if it, like, is. Would be a nice consumer experience. And so the excitement comes from, like, fresh look at the interior, bunch of new ideas, bringing back these kind of like analog tactile experiences, taking inspiration from history. But then the concern is that it's Ferrari and they don't. They can't fully deliver on it and some of these features, like, aren't as reliable as you want them to be and. And on and on and on, you know.

Yeah, I agree with all that. I also think your. Your point about the Mercedes AMG GT floating iPad. Like, I love the way that car looks. I love that car and I hate the interior.

I will not buy one because it looks.

It's so bad. The other thing that's interesting is, is he also seemed to kind of, you know, Mercedes has these iconic round vents and I kind of got that vibe from this interior too.

Yeah. In the back, the passenger vents look really cool and they can point around in a unique way.

Yeah, it had a bit of an AMG thing, but anyway.

Yeah. Not to go like too giga brain, but is there any translation that you. Do you think you can do any, like, red string drawing from skeuomorphism to flat design, like UI to what's going on with the Ferrari interior design language or even just like reflections on your career and how you processed skeuomorphic design? The transition to flat design, which sort of, I believe it was led from Apple and then everyone else just fell in line. But what was your experience there and then and then is there anything to pull from it?

Yeah, that's a great point. I mean, I think that we lost a lot, actually, when we lost skeuomorphic design and we went to flat design. And I think what we're seeing now is a pullback. And you're seeing it everywhere. I think you're seeing it more in software, too, where there's more sort of textures and transitions and different things happening and shading. Again, you're seeing it in car design now with knobs coming back and knurled surfaces and like I saw in the new.

Audi.

I think they call it the Concept C or something like that. This new TT inspired design they're working on. Their interior is, like, very tactile. It's not just that there's buttons. Like, there's buttons, but there's also. The buttons have texture and there's clicks and snicks and all these things that are going on. And I think that it's so deeply satisfying to know that when you put an interface into a certain position, that you did it right and then it works. And that satisfaction, that physical feedback, tension, all that stuff, and I think we lost some of that. I love seeing that it's coming back. And, you know, I think, look, I saw some criticism online that, like, what Jony Ives was, you know, designing stuff for billions of people, and now he's designing something for, like, a thousand rich people, whatever. But I think this idea is going to make its way down the chain, across. I think people are going to be inspired by this design in general and. And hopefully pay more attention to these things. I do think, though, it's very important that what we've seen, pretty much every car interior today can just be, like, pulled out and put into a different car. It just feels like they're all the same. And I think that these brands need to figure out the whole idea and not just kind of just throw a bunch of screens and go, we've done it.

We've made it.

Yeah. I mean, the Aston Martin has been ripped apart on some of their Halo cars. The Valhalla, the Valkyrie, where they just built this beautiful racing machine, and then they're like, here's an iPhone, right? And it's like, we got it. We got to get beyond that.

Is the Ineos Grenadier underrated? Overrated? Is it a gimmick or is it the future?

I like the outside of it. I've never driven one.

So inside, it's all physical switches for everything. I mean, there are so many switches. I mean, they still have a screen a little bit, but like, you know, you have switches up on the top of the ceiling and all the way down to the cup holders.

Full airplane cockpit?

Yeah, it's a little airplane cockpit Y but people love it. But it is, it feels like it's intentionally contrarian, which I think is a good exploration. But yeah, I think they've leaned into.

What that is, which is like trying to go to become an old school Land Rover again. I have a Land Rover Defender, the current version, and I think they've done a really nice job with the interior in that car. There's some screens, they're relatively small, but a lot of things are still tactile. And also the interior really mimics the exterior. I think that's one of the best designed cars at the moment, actually. Inside and out, like combination full ideas. Yeah, there's some good stuff though, but.

And I think like, yeah, there's a.

Lot of cool stuff.

I was telling a friend, like, yeah, it seems like the INEOS is like a huge hit. I just like, it's like every fifth car and it's like the only. Like, dude, you live in Malibu. It's like the prime. It's like I was like. But yeah, the one other criticism again, I want to hold actual criticism until we see what the actual outstanding looks like. But. But the current renders that we're seeing just look like Kia's. Like if this looks like the outside. Yeah, I mean that was the same criticism of the Purosangue which is like, hey, like this could be. You could put a Kia badge on it and I would think like you.

Can trust it with like the Lamborghini where like everything's an octagon or hexagon and like it looks like everything is the most extreme. Like just to turn it on you have to like click up and push down and there's like 25 different things and then it's like over engineered but it's unmistakable. And that's sort of the shtick. And it works.

I don't know my prediction. I agree.

But I think this is credit to Kia though, and not discredit to.

He is doing some amazing design work.

Totally, totally. And yeah, it shows.

Who knows, maybe they're a love from client and they just said keep our name off the web page quietly. Conspiracy theory here.

My prediction with the Luce is that it prices in the middle three hundreds somewhere in that range you'll be able to buy it for 200 winter sticker within a year. Yeah, I actually think they'll sell. I actually think they'll.

Because they're EVs.

Yeah.

So rough.

I think they'll sell well because there's certainly demand for daily able Ferrari. But the Purosangue being priced at over half a million dollars, it just like really cuts out so much, so much of the market who can. Actually there's very few people that will daily a Ferrari that want that kind of brand surrounding them at all times. It's twice in Aeris and then you're going to probably lose a quarter mil within a couple years. Like these cars are going to be available for a quarter million dollars and not that long. And I expect the ev, the depreciation to be even more vicious. But I expect these to do numbers in Cupertino. Know from the old Apple guard that wants to just feel.

By the way, I like that I.

Did comment on this on. On, on X about how the worst part of this interface is the Apple CarPlay integration.

Yeah. Why?

And I feel like. I feel like there's a sweet revenge in that for, for. For Johnny and I think Mike, who's kind of a iOS designer.

I forget his name.

Fantastic designer. The fact that they made this look the worst.

Yeah, yeah.

Well, yeah.

Just because the design language doesn't match or it's not customizable. Okay.

Well, that. And it doesn't match.

But what's interesting in most cars, like, it's the best part because the rest of their interface sucks.

Totally, totally.

In this case, it's the Apple looks the worst. The Apple bit looks the worst. And I think. I just think it's sweet revenge for them, frankly.

Yeah. We're still in this messy middle where cars have basically two operating systems and it's unfortunate.

Yeah. Yeah. It's really bad in some cars where if you turn the volume knob, you can't access the CarPlay UI because it's a layer on top and it's like, oh, come on, guys. I think Apple's working on solving that, but it takes years and years to roll all this stuff out. It is an odd revenge to spend your entire career building a walled garden and then be on the outside of that walled garden and be like, but I'd really like to get into the garden. And you can't because you built it.

That's right. Anyway.

That's right.

Well, cool. Thank you so much.

Great to catch up.

Thanks for popping up.

Anytime.

Yeah.

Talking cars come back on soon. Great to see you.

See you guys in person.

Have a good one.

See ya.

Let me tell you about Gusto, the unified platform for payroll, benefits and hr. Built to evolve with modern small and medium sized businesses. And without further ado, we have Bill Bishop. He runs Cynicism. Welcome to the show, Bill. Good to see you again. Welcome back.

Hey, thank you. Thanks for having me back.

It's been Happy New Year. How's your new year going?

It's going pretty well. Although it feels like we're in the. Near the north pole here in D.C. we've been iced in for two weeks. It's pretty nuts.

Well, describe iced in. Are you actually. You can't go outside?

No, I got everything dug out, but literally at one point the, the guys digging out my driveway were taking selfies with the blocks of ice they could pick up.

It was so big, it was frozen salt. It.

Wow.

And it still is.

That's insane.

Did you watch the Super Bowl?

Of course. And congrats on your ad. That was awesome. Thank you.

Yes.

The earned media was. I don't know what your ROI, like the, the multiples on your 50k, what you spent, but that was. What a brilliant hack. Congratulations to you guys.

Yeah, thank you. Thank you. Anything else?

The game kind of sucked, but, you know, other than that.

Yeah, it was, it was not, not the most exciting. Just a lot of field goals in the first half.

Yeah, we. We unfortunately left. Left like three minutes into the fourth quarter because it just, we knew it was going to insane chaos getting out and then it just got really. It got, it got a little more interesting.

Yeah.

So I have a question, but was that an interception or a fumble?

Oh, I don't know. You're asking the wrong people.

You're asking the wrong people. When did that happen in the game?

That was their last touchdown, Right?

Okay, so we had left. We had left. We had to get to the airport. And to be honest, it was tough to follow because we went with the ramp team and a bunch of our friends and so there was just.

There were a lot of interesting conversations to have with folks. And so there's a lot of opportunity just to get lost in conversation and then turn around and. No, they scored.

Anyway, any of the ads stand out to you?

The robot vodka ad that you guys talked about a little while ago, which was pretty awful, but you know, China's got the Lunar New Year's coming up next week, and Lunar New Year's Eve, the CCTV does this big spring festival gala with hundreds of millions of people watch. It's going to be full of robots. So I'm very curious to see how they spin the Robot performances.

Yeah.

I mean, I think they're going to.

Drinking vodka through the neck.

Yeah. That was very, very weird.

Yeah. Associating your alcohol, which kind of like tastes like engine lubricant, with heavy machinery is an interesting decision. I expect the robots to perform incredibly well, just based on some of the demos we've seen where these things are flipping around, they're moving like actual fighters or dancers. It's incredibly impressive. And still worried that we're going to let them sell 10 million of those in before we wake up.

Have you been following the DJI story?

Which part of it?

Just the ban. And, like, how fast it's rolled out, if there's any loopholes, because you always hear the headlines like, Nvidia, the chips ban. Like, there's zero. Nvidia chips are going to China. And then it's like, oh, well, there's diversion, there's cutouts, there's Nerf chips that wound up. You could train deep sea, you could do a lot. And then the trade deal gets renegotiated. And so I'm just wondering, like, there were a couple founders who were sort of taking. Making victory laps in the drone. American drone community, and I'm rooting for them. I love an American DJI. GoPro famously failed at this, mostly because of the supply chain pricing, all of that. But it's always like, I see people take victory laps and I'm wondering, it feels maybe a little bit like, will the new regulation stick? How all encompassing is the regulation?

I think it's a bit premature, and I think you've seen already bits of. Of it whittled away where now you can buy previous models and parts for it. And so part of the problem, I think, really is that DJI makes the best drones, both in terms of performance as well as cost. And so unless the American firms can actually make drones that, like, law enforcement wants or, you know, various companies want, you know, I mean, it is. It's in an unfortunate situation. I certainly hope the American drone makers can catch up. But it. And maybe this regulation will help. But, you know, we have to be competitive, right?

Yeah, yeah. I mean, it also just seemed like dji, yeah, there were a ton of, like, commercial applications, but it was just such a go to Christmas present for, you know, a lot of people, like, you know, the casual outdoor person that goes on hikes, they want to take cinematic video. Like, realistically, it's going to be collecting dust in three months, but it's going to be an epic present on Christmas. And so, I mean, that probably propelled a lot of sales and just helps get to scale and that's important in these manufactured products. Right.

And again, as we all, you know, you've talked about ad nauseam with lots of people. I mean China has a supply chains for this stuff and we still don't. And so whether or not these regulations will pull that supply chain creation here, it remains to be seen. The challenge of course is you have to balance cutting off access to products that customers actually not just want but need, like police departments, etc. And, but, but then at the same time just making it so that some companies can, can to take advantage of loopholes, have some sales, but still kind of wash their components through third countries that actually still are probably Chinese components.

Yeah, no, no, I mean the supply chain, I remember digging into the, the small drone motor market. So the motors that go on those drones, the small motors, and there are truly no American companies. There's one company in I think Seattle that sold to private equity and they immediately offshored all of the manufacturing and it's just like this Holdco now and I think now they're starting to bring some stuff back. So there's like green shoots, but these things take years and years and years. Just look at like TSMC in Arizona. So years, maybe a decade, decade to get actual to scale from like the initial plans. Anyway, let's dive into the PLA purges. We read through the Wall Street Journal's coverage and I have a bunch of questions, but how are you framing it? How are you thinking about what's happening in China today?

So the PLA purges have been ongoing for quite some time. They've accelerated over really the last 18 months or so. I mean they are part of a multi year process of Xi Jinping both starting out taking control of the PLA but then also forcing through a whole series of reforms around structure for structure operations to try and get the PLA to what they called world class fighting force with a specific goal for 2027. The centenary goals, which are the, it's the 100th anniversary of the founding of the People's Liberation army where they want to be, you know, some people say they want to be able to invade Taiwan, haven't explicitly said that. But it's to get, to get a force to the point where it could actually undertake missions like that. And the latest round where the PLA has a top structure, it's called the Central Military Commission and it's got a chairman who's Xi Jinping and then two vice chairman and four Members, So seven members. There are now two members, Xi and a vice chairman because he's perfect. The rest over the last year or so. And just for a couple weeks ago, they purged the remaining the one Central Military Commission vice chairman and one member. And it was quite shocking both because there have been rumors popped up and then three or four days later they were gone. But also the vice chairman who was purged was someone who's considered to be close to Xi, who she had kept on past assumed retirement age because he was supposed to be sort of Xi's guy. And so it's pretty shocking on the one hand. On the other hand, it's kind of a continuation of what's been happening. We don't know why. There's lots of speculation. The Wall Street Journal article you referred to, I think talked about possibly a briefing internally that said that this Vice Chairman Shah was leaking nuclear secrets to the us. Wish it were true. Haven't found anyone in DC who actually thinks it is, but would be impressive, right, if we had that level of a spy. But what I think it points to is, and then the question is how you interpret what's going on. There are lots of people who are trying to sort of put out versions of what happened. You know, there was rumors that there was a gunfight, total BS as far as I understand. But it's a black box, so you can't say zero.

So before we go into the implications and the interpretations, can you break down Anatomy of a Purge history of Purging? It feels like a uniquely Chinese just event. Like we, you know, like when we elect a new president, a bunch of positions turn over, a new head of the FDA comes in or whatever. And we don't think of that transition as Purging. Although of course some people get fired midterm, even if they've been appointed by the President. And certainly. So what's actually going on? Are these like forced resignations, are these purges or are these firearms?

They are detained for investigation. Okay, so it's alleged process or alleged criminality or alleged violation of party or military rules in this case. And then they are. And all we got, you know, all we got was a very terse statement from a very nervous looking ministry defense spokesperson announcing that these two individuals, Zhang Yosha and Liu Xianli had been been put under investigation. That was it.

And so she has not come on the record, she has not come on the record about any of this, not publicly.

There have been authoritative statements in like the PLA Daily, which is the military's newspaper, but she has not yet said anything publicly. And there may be, at some point there'll be. He probably has talked about it internally at some point, maybe we'll get a publication of some of his speeches. But we have very, very little information that's public about what's actually going on other than that these two have been taken away from investigation and so far they have not been replaced on this body.

And then one way you could potentially read into this is that you're consolidating power, which makes it easier to perform military operations. The other is that you lost all your top guys who were going to help you with military operations. What's your interpretation?

So it's a good question and I think it kind of is both. There's clearly the number of generals and senior officers have been taken out over the last two plus years is quite shocking. It's in the dozens. And so it, it's hard to imagine in the short term it doesn't have some impact on the military's ability to fight. But at the same time, there are a lot of officers and a lot of younger up and coming officers in the pla. The PLA has historically been an incredibly corrupt organization. And Xi Jinping has been. He started, he really kicked off this anti corruption campaign in the PLA in 2014 in sort of full force. And so what may be happening is that he has realized that he just has to effectively decapitate one or two generations of the PLA to get down to a group of younger officers who were promoted not by buying their positions as was very common through, up until even I think, into the SHI era.

How did those get priced? Obviously under the table. Very.

So you guys, your audience may find this interesting. It's kind of like an angel investment, at least in some cases where actually people would collectively buy a stake in a rising officer. Because I'm not joking, right?

Literally no way. And then, and they'd go, they'd go, basically they'd go out and say like the officer would be like, hey, I've got some potential within the organization. I think I can get this job. Let me pool together some capital. And then they pull together that capital and then they take over. They pay off the person or someone to get the role and then there would be a revenue stream back to the original pool of capital.

The idea is you're, you're buying an option on a future revenue of corrupt goodies. Right? I mean, it's absolutely because.

Because if said person gets the job, they'll be able to generate a bunch of revenue not just with their salary but through like corrupt activities they have.

They can create a whole bunch of opportunities for people who are close to them.

Wow.

And you.

Okay, so yes. So he basically gets a certain job, he's got access and control over some amount of budget and creates basically a little economy around, around within the stack. That is insane saying.

And this was. And there are, there's been certainly, I don't think it's been in the media, but I'd certainly heard that among the funders of some of these officers in years past was our America CIA. Because it was a great way to push people in and you know, eventually they owe you. Right. And so I mean it's actually friends.

And family round from the CIA.

And this is something without going into detail, the party has talked about, about, you know, getting rid of this process of buying and selling promotions. And you saw from some of the previous cases of generals who were, were detained early on in the XI era. I mean the stories of like, you know, the cars full of gold and the, you know, these suitcases of millions of dollars in euros worth of cash hidden in one of their villas. I mean the level of corruption was insane because there's so much money being thrown into military, you know, the military buildup.

Interesting.

Crazy.

What's the state of.

And so part of that, if you have decades of corruption that have been like that has been intimately intertwined with the military buildup, maybe doesn't give you that much confidence in a lot of the actual fighting force and the equipment. Right? Because maybe it wasn't going necessarily to the best vendor, it was going to the vendor that was pushing enough money out the back door.

I mean that is certainly the risk and potentially the concern. I mean you look at, in addition to all these generals, they've basically taken out a significant chunk of the leadership of the military industrial complex, all sorts of, of defense contractor defense, you know, weapons makers, heads of research institutes that were involved in weapons development. At the same time, the weapons look like they work. I mean, think about China and corruption is, you know, China has this great high speed rail system, right? Well, the guy who really oversaw it is in jail because he was corrupt. So. But, but the thing is, is the corruption is just sort of like it's another tax. It still works.

It still works, right?

Unlike maybe other countries, you still get it done. It still works pretty well. But some folks, you know, but then you make a little money on the side.

Wow, that's funny.

He's like, I actually Gotta make the trains run so I can keep the gravy train going.

Yeah, exactly. The gravy train is important. Lots of lessons there. What's the state of communications or relationships between Donald Trump and Xi Jinping? Are they talking regularly? Are they meeting in person? Are we at, like, a. A local top or local bottom? Are things the worst they've ever been? Somewhere in between?

No, no, we seem like we're in a steady state. They had a call last week, which, you know, again, I think is an indication that so far, at least, things are on track for President Trump's visit to China in early April. There was one sort of wrinkle, though, in the readout from the Chinese side of the. Of the call he had with Donald Trump last week. Week he was. Had some pretty stark language around Taiwan and specifically around U.S. arms sales of Taiwan, because the U.S. sold 11, you know, announced an $11 billion arms sales package, Taiwan back in December, which was at the time. It's a very large number. The US has sort of been doing billion or so kind of packages and rolling them out. The Chinese get pissed off, but, you know, they move on. 11 billion was pretty significant. I had heard that the reason the Xi Jinping had mentioned sort of being prudent around arms sales to Taiwan last week was specifically because the US Was working on a big arms package. The Chinese had found out about it, and the Chinese ambassador here in D.C. had basically gone to the White House and thrown a fit over the weekend. We saw, I think, on Friday, the Financial Times reported, yes, there's a $20 billion package in the works, and it's something the Chinese don't want to have happen. And they have threatened to postpone or cancel Trump's visit. And I think they might actually mean it. And so I would imagine that the Trump administration won't push forward that sale until after the meeting. What's interesting, though, again, is it's not clear Trump knew about it. This is people in the administration who maybe are more interested, who are sort of more pro Taiwan not happy with this kind of. Whether you. I don't want to call it detente, it's a little too early for that. But this sort of steady state in the relationship over the last couple months, and they want to make sure that Taiwan is still getting attention.

Yeah, it's fascinating. How does China frame the Taiwan question? Internally in the west, we sort of all accept the premise that if we're doing an aid package, it's for defense, and that the only possible scenario is that China would Invade at some future date. But does China use rhetoric that's like, well, we don't want Taiwan to have weapons because we're worried about them invading us? Is that even something that they toy with? Or are they saying like, we don't want them to have weapons because we're planning to do this at some point?

Well, no, I don't think they're worried about Taiwan invading. I think it's more, you know, Taiwan is the first of their red lines, especially in the U. S. China relationship. And so they, but ultimately they, you know, they, they don't want, I mean, the U.S. yeah. Saying we're going to sell you a bunch of weapons. Even if right now, for example, Taiwan, because of the political. What's going on in the Taiwan politics. You know, the Taiwan legislature won't approve the budget to buy the last package of weapons because it's the opposition party controls with a coalition partner controls the legislature.

So this $20 billion package, should we assume that CCP operatives have effectively infiltrated the opposition party? Party, or is that too much of a tinfoil hat? Because I imagine that if you were, it would be worth the time to try to get your team elected.

Good seed investment.

Yeah, good seed. Put an adventure.

There are no questions. A lot of influence efforts. The Taiwan government under the current president, Light Ching Te, has definitely stepped up and talked more about these kinds of infiltrations and influence. I don't think there should be any surprise that that is going on. But, but, but ultimately for these arms packages, it's also a signal of, I think the US Is trying to make it a signal of we still support Taiwan. So that, because what, what Beijing wants ultimately is for the Taiwanese people to think that there's nothing, you know, no one, there's no help coming, they have no other choice. But, you know, effectively resistance is futile. Right. Roll over. And the faster you roll over, the better you'll be treated. Is, I think the, the kind of the constant messaging they're trying to put out there. And so support from the US in terms of either rhetorical or big arms packages messes up that messaging. The comments from the Japanese prime minister back in November mess up that messaging. And then they also complicate the PRC's planning. In the event that there is some sort of scenario where they have to use some sort of force, it's going to be a lot harder. If the Taiwanese have better weapons, better training, and have support from, from the US and potentially Japan has to get involved.

What economic indicators are you tracking in China broadly Things like unemployment rate, foreclosures, general development, infrastructure development, housing development, all that stuff.

So all those are worth tracking. You have to filter through the data. I think the, the consensus of a lot of the folks who really tracked us closely is that generally like the economy's not doing great, but it's also not falling off a cliff. Right. It's not the binary boom or bust. The things to really watch over the next. I mean, we'll learn a lot more by the middle of March because the first week of March starts what's called the two sessions. And the one that matters is this National People's Congress, their legislature. And so we'll get a work report from the premiere that will then lay out targets for this coming year in terms of things like GDP growth and et cetera. But then also this is the year where they roll out the 15th Five Year Plan. And that also includes not only high level goals, but also some targets in certain sectors. And so those ultimately I think are more useful to look for over the next few months than some of the sort of high frequency data, just because the high frequency data is noisy. And ultimately the Chinese companies, you know, again, the stock market's up 50% or so from the lows. It's at a new high. I think today the tech sector is booming. There's, there's, I don't want to say bubble, but there's a big AI boom in terms of the AI related shares.

Yeah. Their, their companies are going public much sooner than ours.

Yes. With much lower revenue raising, much less money, much lower valuations. And a lot of it is because they, you know, they, they actually need the capital. I mean it's, I think they, but the amount of capital they need in raising is a fr. It's like, it's like, you know, a rounding error. For what, like OpenAI is raising. Not quite, but sort of.

Right.

No, Yeah. I mean, you see some.

What is the general pop. How does the general populace feel about AI? I think most of America is, especially after the super bowl. Nobody was seeing the Super Bowl AI ads being like, you know, this is amazing.

Of sucked.

Well, yeah. Okay, so the ads, yeah, the ads weren't that great. A lot of them weren't that great. But there's just also a general fear around job displacement, lack of, you know, people are not excited even to put a data center in their state. Right. We have all this legislation going down the pipeline, but how do people in China actually feel. Unemployment rate for youth is already so high, it's hard to imagine it going much higher. So maybe there, maybe people already feel like it's over.

But so, I mean, data centers, energy, obviously you have other guests who talked about it's not an issue for China. Right. It's somewhere they could build as many data centers as they want, as long as they can get the chips. That's, that's the bigger issue. I think when you look at what the, what the government is doing, you know, they have a, they have like this AI plus plan they rolled out a couple months ago where it's really to embed AI throughout the economy and throughout society. And so they're not really focused on, on as much as sort of getting to AGI and these massive models. They're really more focused on diffusing how do you use AI and all sorts of tasks in your apps, in WeChat, for medical, for seeing doctors, for medical advice. I mean, there's a big boom right now in companies chasing sort of medical AI, including Alibaba. And so it seems like even though it may not be the models may not be like the ChatGPT or Gemini level, at the same time they're being taking a much more pragmatic and practical approach to just diffusing it through society. And then in terms of what it does to employment, there's been lots of discussions about the impact. I think, you know, it's a country where they can incentivize positively or more negatively companies to not necessarily lay off as many people as they would if they were operating just purely on an economic basis. When it comes to sort of AI disruption doesn't mean unemployment is not a significant problem for the youth. And, and I don't know, I don't know that they have a good solution for that. But it isn't holding back what they're trying to do around AI at this point.

That makes sense.

Last week Jensen was on cnbc, I think it was Thursday or Friday, talking about, you know, how the overwhelming demand, obviously we had earnings last week. Everyone's raising their capex guides. You have legacy AI chips that are sitting at very high utilization, surprisingly high utilization and pricing compared to what a lot of the AI bears have been thinking about over the last six months saying, like, hey, all these chips are going to be worthless. And turns out like Michael Burry and others. Yeah, yeah, those, those types. And so I think there was some conversation around, okay, and, and if Jensen wants to go like, hey, all these people are super chip constrained, I think the question comes up, okay, so then why are we selling chips to China? Then right, if our leading labs are not able to get the.

Call Amazon, they'll buy them.

Yeah, call Microsoft.

It's a great question and it's something that has certainly you've seen some movement on Capitol Hill asking that question, asking the impact on HBM prices. Right. Memory prices. And ultimately the answer is well, Jensen Huang can go direct to Donald Trump and convince him to approve these sales. What's interesting, right, I think it was the Financial Times reported last week that even though like the Department of Commerce has signed off on the licenses to sell to China the H200 that the Department of State or the State Department which has this bureau of. I think it's arms control and non proliferation, they have yet to sign off on it. So the sales actually haven't happened.

Interesting. And what is the general sentiment now from the CCP and various groups? Because when the first time we maybe agreed generally to a chip sale and then Howard Lutnick came out and said we're going to get them addicted to the American AI stack and then they were like actually we don't want, actually they were like no, we don't want them. But clearly all the companies want them. Their computer way more, more compute constrained than we are. They're ready to rock. And so where is that actually, do you think that if it actually gets fully approved that they will all flow without any type of red tape on the China side or.

No, there's been reporting, I mean various reports the Chinese are being careful about who they allow to actually order and they're talking about the H200s. They didn't want the H20s, they need the H200s because China has enough like they can actually make, make looks like decent inference chips. They just can't make the chips they need for training. Right. And so that's the H200 fills that gap. And so I think the 90s perspective is look, we're not there yet. We can fill this gap. We have to sort of keep competitive in the AI game. Nvidia, the US government has approved these sales. Nvidia will sell this to us. We're just going to make sure that if you buy these, you also have to also make sure you're buying Chinese chips to keep supporting our own indigenous ecosystem. Ecosystem. Right. And so and then of course there are some, I think potentially some security, security concerns because of the, whatever the security review the US is requiring, which is basically I think just to make sure that the chips get shipped to the US charge the 25% licensing fee as a tariff which Makes it legal and then ship them back to China. But you know, from the Chinese perspective, what's the security review? What do they have to do these, you know, who knows what they might do these chips. So there's certain places I think that like state owned enterprises, certain labs where they probably won't want these chips. But the issue also is, right, there's also still a lot of Nvidia chips in China. Right. So we should see in the next week or so. If you remember, you guys, it's been a year since the quote unquote Deep Seq moment. Right. And so now everyone's waiting for Deep Seq's next model which is supposed to launch on or around Lunar New Year, which is next Wednesday the 17th. So we should have some sort of a Deep Seq model the next eight or nine days. I think it was the information reported it's being trained on Blackwells which they're not supposed to have. Right. But somehow they have the Blackwells.

They fell off the back of a.

Truck and they can get as many Nvidia top end chips as they want hosted overseas in these cloud facilities.

Yeah, right.

So it is not a clean set of controls by any meetings.

Yeah, yeah.

What, what do you think China's reaction is to the, the latest election in Japan? Obviously Japanese equities responded positively to, to the result. But how are you tracking that whole situation?

I mean, I think that the, the Chinese helped Takechi because their, their reaction to her comments in early November which again she said she reiterated the Japanese position on, on sort of a Taiwan contingency, so to speak. She said it in a, in a, in a setting where it hadn't been said before by sitting prime minister. And you know, but their reaction really I think helped make her case and other sort of more defense hawks in the Japanese government make their case that we need to do more because China's a threat. And so now she really has a mandate. The question will be will the Chinese continue to really push on her or do they sort of find ways to over the next, it won't be immediate but over some period of time find ways to at least calm things down and then start re engaging with dialogues. I think, you know, the fact that they clearly started playing the rare earth cards at Cardigan with Japan. You know, also again, in some ways there's no going back for Japan. Even if the Chinese were to find, they were to find an off ramp and find a way to sort of get back to kind of the U.S. china sort of detentis like relationship around. I think the damage has been done in terms of Japan needs a stronger military and Japan needs to move faster to protect itself from the weaponizing of certain parts of the supply chain that China can do, which we all know from rare earths.

Yeah.

Yeah, that makes sense. Anything else? Rudy?

This was super fun. This has always been fascinating. We love having you on the. The audience. Audience loves you too. And I'm going to follow up and ask what headset you use because we're making. Because not every. Not every guest of ours comes in with a sound video set up like this.

It's a. Sure. It's a shore headset. And Ben Thompson got it for me, you know, because he. Because I do the Sharp China podcast with. With his team and with Andrew Sharp. And so Ben sent me all the gear.

Very thoughtful.

So I have this solid state logic little box and then I plug in this. This headset actually. Is it. Sorry, it's a Sensenhauer thing. Sorry.

Sennheiser.

Sennheiser.

Yeah, there we go. Sennheiser, Alpha. Well, thank you so much for taking the time.

Have a good day.

Great to see you.

Good luck with the weather. We will talk to you soon.

Thank you. Thank you.

Have a good one. Let me tell you about Okta. Okta helps you assign every AI agent a trusted identity so you get the power of AI without the risk.

For our next guest, let's go over to.

Let's go over to Tyler. What's new in Tyler's world? Why are we xing out Tyler back to the show? That's it. That's it.

I'm just trying to.

When are you going to drop the special effect? The latest and greatest special effect.

This one?

Yeah. Tyler, tell us the rest of your story. That's all, folks. Fantastic. Let me tell you about Sentry. Sentry shows developers what's broken and helps them fix it fast. That's why 150,000 organizations use to keep their apps working. And without further ado, we have Jason Kelly in the restream waiting room. He's the co founder and CEO of Ginkgo Bioworks. Jason, good to see you.

Wow. Good to see you guys.

You look fantastic.

Incredible setup.

You sound fantastic. I don't know if you remember me. You gave me a tour of Ginkgo Bioworks over a decade ago. Maybe 2012 or something. 2013.

Yeah. That was a real long time ago.

It was a long time ago.

Different era you were doing. Yeah, different era.

Yeah. Soylento. It was. It was remarkable. I mean, the Facility was incredibly advanced. You had the pipetting machines. Everything was already automated. I'm seeing behind you, it's clearly grown. Break it down for us. Like, what is the shape of the operation today? What is the, what is the scale of the footprint? What's the business like today?

Yeah. So when you would have come by, we were sort of early on this journey of trying to automate last year work.

Yeah.

So I, I brought you in the lab. I'll give you like a little bit of an introduction to Please. How to do science. You know, you're hearing a lot about AI for science. We announced a thing with open AI.

Yeah.

So like, what does it actually look like to be a scientist?

Yeah.

So first off, glasses. Lab coat.

Lab coat for sure.

Good. Stuffet pipette. There you go.

Okay. Yeah.

The robot handles the pipette.

If you remember, like high school biology.

Right.

This is a little like a very fancy straw.

Yeah.

And you suck up liquid and you squirt it out somewhere else. And you do that for five years and you get a PhD at MIT. Like I did.

Right.

It is just a grind. Like you're actually doing the work of science primarily by hand.

Yeah.

And that's, that's still broadly true. Like all, you know, United States, where our labor is extremely expensive. Everything else doesn't matter. Like scientists are working by hand at the lab.

No, I remember I went toward my. The Langer lab at Caltech. My co founder was doing a PhD there and I was like, oh, like you're in the most elite lab. It's Caltech. Like it must be all robots and stuff. And he was like, yeah, this is my day, this. And then it just jiggles it and kind of just like gives a little motion. And then centrifuge over here, just like moving. Moving little buckets from one tool to another all day and just by hand.

Yep.

Yeah.

So, okay, so then the tech people hear this and they're like, that's crazy. Just automate it, bro.

Yeah, why not?

What are you guys doing? Like, it's crazy. Right. And so, so the analogy, they're like.

Trust me, I built a crud app. I can do this.

But let's take transportation. Right? We had automated transportation 100 years ago.

True. Trains.

Subways.

Subways.

Yeah, exactly. Trains. Right. Okay. And yet we had the car, which is this manual thing. So why didn't we just apply our automation to the car? And the answer was variability. Right. Like the car needed to, to. You needed to transport yourself to your front door. And once you had that extra Element of like the variety of requests from the user. Rails were screwed. Automation was screwed.

Yeah.

And so up until four or five years ago, suddenly we start seeing waymos around and we don't even call it automation anymore. We start calling it autonomy because it's so surprising that you can automate that variability. And what they did with transportation at Google and Tesla. Tesla, that's coming for every other, in my opinion, physical industry. And what's going to bring AI in is managing the variability.

Yeah.

And so this behind me is an autonomous lab. The reason we were able to let GPT5 drive this thing, which we can talk about, is first off, because it operates like a waymo.

Yeah.

A scientist can just tell this thing what experiment it wants and it'll do the experiment.

Okay, so break down the line between.

10 years to get there.

True overnight success rate here. True overhead success.

That's not a trivial thing.

Yeah.

Break down the line between experimentation and manufacturing because once a discovery is made and we're hearing about, you know, GLP1 scaling up, I feel like those pills are made in automated factory. So break down that.

That's subways.

Okay.

That's subways. Manufacturing is the subway lines. Right. You're doing the same experiment every day. No big deal. There's a ton of automation already and biotech and pharma and any, frankly, many manufacturing across the board. Right. It's the research side of the house. Those sad grad students at the bench, and I say that with love, you know, that are doing all the variable work. And by the way, that's what's moving the frontier of science.

Yeah.

And so if we want to, you know, and by the way, this is both true at like the nation's level, like, you know, if we want the US to lead in science. But I think what you've now seen, thanks to, you know, Sam and OpenAI and making hardcore bets on, on original research, you're seeing industry suddenly wake up again and say, oh, maybe we should be doing original. Maybe it's time for Bell Labs again. Yeah, like we should be doing original research because it pays. Right. You know, like if you really have fundamental discoveries, then you can, you can be the first to the line to commercialize them. So I think unlocking this flywheel with science where the models can control autonomous labs, I think you'll see that in a lot more sectors than you'd expect because people are getting religion about research again, even in the industrial sector.

And talk about how you've processed the AI over the last maybe five years. Because, like, the whole concept for the company originally wasn't predicated on people getting excited about LLMs. It was just like, hey, let's automate a lot of this work. So when did you start thinking about.

In many ways, it was like a crud app on top of a lab, right?

Yeah. And I'm sure you always imagine, I'm sure you like, there's some people that have a company that's capitalizing on AI that you can tell. They just were like, oh, this is a better idea than what I was working on. I'm sure you and the team and your investors and partners always thought about a world where you could generate a concept for an experiment or do research with AI and then automatically sort of prove it out or study it in the real world. But walk us through kind of like how you've processed the developments of the labs and how all of that work can be integrated and applied within ginkgo.

Yeah.

So.

So we're laser focused on make that hardware layer and then the software that. That basically orchestrates and schedules this thing and makes it possible to put lots of experiments on it, make that robust and also make that have hooks into whatever it could hook into a scientist placing an order. Just like a person can sit in the back of the way and tell it where to go, or it can hook into an AI model placing the order. Just like when a person's not in the Waymo. Waymo's AI is telling me the Waymo where to go. Right. Same exact idea. So it's not our. We're not actually trying to solve the AI scientist problem. Right. Like, OpenAI came to us like we had that interaction. They were excited on the research department to see, hey, can these models be smart enough to design experiments? And we just announced on Thursday that we had a breakthrough where we let OpenAI do six rounds of experiments on the platform. Platform. And on the fourth round, it beat scientific state of the art. And then by the sixth round, it had beat it by 40% on cell free protein expression, which is just like a tricky biochemical reaction to set up and design. And so it was. It's not.

Is beating it just a, like, consistency thing. What is. What does that actually mean?

Yeah, so what in this particular thing, Right. Like, so all the cells in your body, right. Are producing protein every day. So you as a human are basically a bunch of like nanotechnology.

Mine produce a lot of protein.

Yeah.

Yeah, you guys are. Yeah, I can tell. I can see the difference. So they're making all these and all.

The proteins are different.

They're like little pieces of nanotechnology. It's almost freakish to look at. Like that's what they look like under a microscope. And so they're all interacting. They build your cells. When you cut your skin and it regrows, that's all the, the proteins in your body, like being able to rebuild that stuff. And so your cells can take DNA code and turn it into whatever protein you want. All right, so the question is, could you as a scientist, print a piece of DNA, which you can do synthetically, you design it, add it into a mixture that has all the parts, the guts of a cell, and have it make your protein at a high level. It's like the world's smallest 3D printer.

Yeah.

And so that, that reaction is cell free. So there's no cell there. It's all in a test tube. It's in like these little guys. Cell free protein synthesis.

Sure.

And it lets scientists design new drugs. If they're making a protein therapeutic, it could be a new material. Who knows? Right. And so we are able to let chatgpt bring the cost of that down.

Yeah.

And so that was the, that was the big goal. Like how much cheaper could it get? And there was a paper out of Stanford just in August that set a benchmark and we beat it by 40%. So I think it's a, a, it's a good demo. It had a clear benchmark so you could mark against what other people were doing. That's why we liked it. The experiments are fast.

Yeah.

But the model designed about 500 of these plates were each. Well, and that is a little experiment. And it designed the experiment for each one.

Got it. Yeah. Talk more about the different applications across. I mean, we were talking about like synthesizing unique perfume scents at one point.

Yeah, sure.

There's obviously everyone jumps straight to cure cancer, but there's also so Boom and GLP1s. Like where, where are the bounds of like where we need more experimentation or even just like where experimentation is valuable.

Yeah. So I'm pretty excited about what's happening with the GLP1s. I think it's opening the door to applying the tools of biotechnology to wellness.

Yeah.

Right. Like, like right now, if you think about the pharmaceutical industry today, it's basically the disease industry.

Yeah.

Yeah.

And like how much of your life. Life are you sick? Depends on the person. But not that mode for the average person. Not that much. How much of your life would you like to feel better? Would you like to sleep better? Would you like to have more muscles, you know, Right. Like, like, like, like that. Like if you have something that adds muscle mass annually in your 60s, it's another GLP one. It's, it's a, it's a multi trillion dollar drug. If you have something that adds a year to lifespan. Yeah, what's it worth? Like how do you even, how do you even put a market cap on a thing that adds.

And when you think about, when you think about how much consumers spend on various wellness things today that have zero impact, like truly, like, well.

And you know, do you know why? Do you know why they have no impact? The reason is we the industry today, we, the pathway to apply all this to the problems of wellness is much more. It's like muddy. It's unclear how do you get the FDA approval. There's lots of, there's lots of barriers. So we haven't actually thrown the full horsepower of biotechnology against that problem. We've only thrown the full horsepower biotechnology against disease. And I think, I think that needs to change. And so that's my, if you ask like, where do I see biotech? Because all that stuff, right, was ideas on like different areas we could bring biotech to that wasn't disease. And people tried the food, people tried perfumes, people tried new materials, all kinds of things. And the one I like the best right now is, well, health and wellness. I think it's a monster.

If you had a 62nd slot in the super bowl and you wanted to get people excited about the intersection of AI and all the things that we're talking about, what would you want to communicate?

So I think what would be cool to communicate is that science, like what it really is, is like the formalization of human curiosity. Okay? And everybody's curious. And the reason everyone doesn't do science today is this shit, okay? Right. You're blocked by the cost and expense of all the physical infrastructure. And, and if you took that away, if this was available cloud style, for example, how many people would want to be doing science? Maybe they have a health condition, they want to study themselves. Maybe they have a new idea for a material, maybe they want to make a new pet. You know, they want to do a new, they're a gardener and they want to make a new plant variety. I have no idea what ideas they would have. Right. But I think that is going to be accessible to people coming up. And I know that sounds crazy, but I'll tell you something else that sounded crazy in 1960 if you said, said random average people will program computers yeah.

Aws.

That sounded insane. Totally, 100% absolutely insane. And so I think you fast forward on the back of this, you know the, the model's ability to access literature and be smart and tell you how to turn your question into an experiment. And then the autonomous labs that could do that experiment for you in the cloud.

Yeah.

And I think we'll have millions of scientists just like we have millions of programmers now as we may.

Then what? Yeah, then what's the next step? Because it feels like.

Yeah, that was perfect. That was perfect. We'll run this next year.

Models, your guys ad was the best. Thank you.

I think.

Models, yeah, I mean models can do a lot of reasoning around experimentation. If you have an idea and you come to it, might be able to reality check it against literature, do a deep research report, kind of flag problems that then you're hooking up the automated lab. What's the next phase? Like automating a mouse model or creating a fully digital mouse model or model of the human body or actual physical mice in a lab that you can test. Because that's a big piece of the FDA approval process.

I believe it is, but it's dropping out. So I think this is one of the things that this new FDA is doing is they're getting rid of the animal experiments. I think that's a great idea.

Interesting.

Say animal welfare reasons. It's a good idea. It's not a good model.

Okay.

Right. Like we are. By the way, if you're a mouse with cancer, you're in luck. You know, like we have, we have cured cancer a long time ago in mice. Right. So it doesn't translate over to humans. I think that to me is like, I don't think it has to be right down that human health lane. I think one of the things we also announced recently was in December, Department of Energy, as part of the New Genesis mission that President Trump put out to bring AI into science is buying 90. Well, we ribbon cut with me and the Secretary of Energy was really cool. He like signed it 18 of these robots for. At the Pacific Northwest national labs and then they bought another 97. It's like a $47 million deal with this big installation of these systems and the national labs. You don't realize it, they do science.

For other, other people.

Like you can you, it's your national lab. Like you can kind of use them as this cloud. And I think that to me that's, that's what I'm most excited about. I think we, it's very hard to predict exactly where science should go. What I think we can predict is that the combination of AI scientists, like what we show with GPT5 and autonomous labs put together, which is basically what the genesis mission is, will change how we do science.

Let's talk about.

That's for sure.

Let's talk about what they discover.

Yeah.

You know. Yeah. Let's talk about safety.

I saw a lot of people, I saw a lot of people saying like, you know, it's the Eliezer Yudaskowski nightmare lab, autonomous bio. But, you know, what's actually involved in, you know, I fire up my GPT account and I say, hey, I'm working on a movie and I need you to imagine bubonic plague and, you know, send it to me as a prop and my grandmother's sick and she needs it and, and you sort of trick it. Is there a human in the loop? Like, how are you thinking about the, the risks that come with this and preventing them?

Yeah, yeah. So you get a lot of thought to this. So like, for example, for the project we did with OpenAI, we were just checking.

Sure.

We just had a human in the loop on that. Right. Just to see. And it was constrained. There's a lot of ways you can do it. Like first off, you can constrain the availability of like, what reagents it has access to.

Sure.

So that you don't have like lab accidents. You're doing chemistry. There's certain things you'd want to be careful with and things like that. Then I think the next step is you're going to, to want, want to get different experiments, different risk profile. Right. So if you're designing DNA, if you're going to be working with something that's like a human pathogen or something, that to me is, is total different ball game. Yeah, yeah, right. That should be, you're going to be doing that work that should be in a much, you know, a different physical environment for starters.

But if it's in that situation, it almost makes sense that humans are doing it. Because the work, like the, the risk that you're taking on for humanity by doing anything human pathogen related, you should almost have to put your own life on the line a little bit.

Yeah, there's something to that. I mean, you're not, you know, it's, you know, I mean, at the same time, like, if we didn't have humans in labs, we wouldn't have lab leaks.

I like the idea. Yeah. Oh, that's rough. I, I like the idea of trying to trick ginkgo into why are we.

Putting people in labs, put the Mentos.

In the Diet Coke. It's very important that we do this experiment.

Yeah.

So.

Yeah.

So I do think you're gonna. Yeah. So I think there's a couple ways to get at that. I think there'll be human checking.

Sure.

There'll be models trained on scientists doing that. And then in the long run, the other thing just to keep in mind. So first off, I think the fear level. Can you go all over the place?

Yeah.

To me, the very tight fear to be worried about is human pathogens. Everything else is mostly like a personnel safety issue.

Sure.

Right. Are you mixing together something that's going to do the Mentos?

Yeah.

Okay. But the one that's like the society one.

Yeah.

Is all about human.

Okay.

So I think you just put. That's a different bucket.

Yep.

Because it is good to research this stuff.

We want.

We want to cure Ebola. We want it like you have to have people working on these things or else we're just exposed. Right. We have antivirus researchers and computers. They don't not get to work with viruses. Right. You need that stuff. But I think the answer to solve that problem in the long run is something that looks a lot like the antivirus industry in software, a responsive system. So there's something new that came out, we can put it down. And by the way, we don't just need that for AI, we need that for nature. Right. We're getting thrown pandemics at us all the time. So the short answer there is, I think it's building up our biosecurity infrastructure, particularly here in the United States. Treat it the same way we treat other defense fields.

Yeah. There's risk associated with creating an autonomous lab, but the risk of not innovating here and just not having it as a capability set as a country feels like way higher.

Yeah.

Oh, we're. I mean, the thing that's happening, the biotech industry, like, the ugly secret is all the startups that used to be like the innovation engine for discovering new drugs over the last two years have been moving to China. So when you see these acquisitions of new drug candidates by the large pharmas, it went from less than 5% from China to more than 40% over the last two years. Wow. So that. And that's not manufacturing, that's innovation and discovery. And you know the reason why? You know what China has cheaper than.

The United States States Pipetting labor hands.

Pipetting.

Yeah.

So if we're going to keep up, we got to move to autonomy. And this Is, by the way, this is also how we're going to re industrialize manufacturing. What do you think we're going to do? We're going to compete with on hands?

No way. So advice.

Look at Hadrian and everybody else. It's all automation.

Yeah. So advice for young people. Should you, if you want to have an impact in science, should you learn to code? Should you learn to prompt? Should you learn to pipette? What should you learn?

You should learn about the domain. So you should learn. If you want to have a breakthrough in biology, you should learn about biology. Right. So that you're the one who understands the limits of that. And then second, I think you do need to like, I don't think PhDs are not going to work at the lab bench at the start.

Yeah.

Because you got to understand the limits of experimentation.

Yep.

So that it's just like today. Who are the best users of the coding?

Yeah. Coders.

Yeah. You know. Right. So who are going to be the best users of autonomous labs?

Labs.

Scientists at the lab then.

Yeah.

Right. That already know. And, and I think the thing is people get scared. I was going to take scientist jobs. That's like there's a great IBM ad from like the 1951 IBM ad. And it's like the, the automated calculator will do the work of 150 engineers. And it shows engineers with slide rules.

Yeah.

I swear to God, Like a sea of engineers with slide rules. And, and you might have said, oh, it will take 150 engineer jobs. And of course total opposite. Right. By increasing the return on investment of computation, what was in the minds of engineers became worth 100 times more. Jobs increased by 100 fold. All post IBM. Right. All post the automation of computation. So if we automate all these industries, whether it's science, manufacture, blah, blah, blah, it only favors the people that have the know how in those industries. But we got to automate it because otherwise we're not competitive in. And in the modern era we already automated the easy stuff. We automated the subways. It's all the autonomy. It's the things that can handle the variability. That's what the hardware guys like us have to work on and then let the AI models go nuts.

That's awesome.

This was fascinating. Really enjoyed it. You're our new in house science. Yeah. Our science corner. Our biotech czar.

Yes. CBPN science corner. It's happening.

I think that might be.

Yeah. Incredible. Incredible setup again. So looking forward to the next one.

Thanks so much for coming by.

All right, take care, guys.

We'll talk to you soon. Let me tell you about graphite code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. There was some breaking news over the weekend. SpaceX has delayed Mars plans to fly focus on the moon. This is something we've been debating a lot. Are you moon pilled or are you Mars pilled? Well, SpaceX is going all in. They had previously aimed to reach the red planet in 2026, but Elon is focused on lunar voyages for NASA now. So the rocket company told investors it will prioritize going to the moon first and attempt a trip to Mars at a later time. According to people familiar with the matter, company will target March 2027 for a lunar land landing without humans on board. The strategic shift comes as SpaceX doubles down on plans to launch AI data centers in space. That deal they acquired Xai, which we've talked about. In a memo announcing the merger, Musk outlined the plans to help build a permanent presence on the moon. He referenced aspirations to use it as a base for exploration deeper into space. He wants to build a mass driver on the moon.

Have you launched? You've always wanted to wondered if this would happen at some point.

Well, I've always thought that even though getting to the moon doesn't really solve the initial Elon pitch for like making life multiplanetary, creating, like if something really bad just asteroid hits the Earth and it's destroyed, like, does humanity continue? If you're on the moon and the Earth's just destroyed, like, you're probably in a bad spot as well. But if you're on Mars, you genuinely are okay. And so it's not the long term solution. Yeah, you probably need some resupplies, but if you're there and you're really set up and you got your systems humming, you're probably going to be okay.

10 million Optimus robots.

Yeah, there's at least a chance. But I've always thought the moon was a great staging ground for going to Mars because you can go any day of the week. You can only go to Mars, I think every 18 months because the, the planets are only in alignment. Do you know the number? Is it, is it 18 months? You're looking at me.

I don't know, but it makes sense. There's a transfer orbit on Darkesh. He talked about the moon driver.

Yeah, the mass driver. Yeah. And so once you get to the moon, it's very easy to get elsewhere. You can build things, there's materials there and then also Underrated is that if you get people doing like moon tourism, you effectively the moon always faces the Earth. And so as long as you're on, on the front side of the moon, you could essentially have an escape pod where like, you know, in that sci fi scenario, oh no, there's a crack in my helmet. You could just dive into an escape pod, smash a button and splash down in the Indian Ocean in like, I don't know, a couple hours. Probably because you're always facing the right direction.

Whereas if you're on Mars says it's a shift. But Elon has been consistent in 2019, he said for sure, moon first. As it's only three days away and you don't need interplanetary orbits.

Not a couple hours. Yeah, but yeah, you can. So you can like, you can get your reps in on the moon, but it's less of like, okay, the final boss. Like really, what is the vision of SpaceX? So this doesn't seem like a complete flipping of the narrative or flipping of the strategy.

The first podcast on the moon is going to go so hard.

It is, it is. You could do a very short podcast in space via Blue Origin for I think a couple hundred grand. It would need to be like a 60 second one of those man on the street videos. What do you do for a living? Okay, we're back.

I started a company, a retail, small retail business called Amazon.

Amazon.

Yeah.

I deliver packages.

Anyways, this is cool. I'm excited. It's. It's. Yeah, I want to. I think I want to go to the moon. Applying like SpaceX pace to something that is not so far out in the future is to be super exciting. I want to. We got to pull up this leaked. The leaked OpenAI ad. It's very unclear. Oh yeah, Seems completely fake. Yeah, OpenAI. Com said it was fake.

It was just completely fake. But what this shows the power of.

AI that you can make an ad as just like a speaker.

Who's this sound?

Can we get sound?

Skarsgard. Skarsgard. Yeah. There's no sound. This was leaked. Right, but so is this AI generated or is this actually. This looks photoreal. If this is AI generated, it's very, very impressive.

I wonder if it's an ad for video model company. They're like, oh, that was actually us.

That'd be cool. That'd be a great.

Because again, it looks like the actual actor.

Yeah, it really does.

Like, this looks better than the Dunkin Donuts. AI kind of like secret.

But also it might just be the actor.

But well, but if it is the actor, you gotta get to the bottom of the app.

This is the real actor. He should comment. Has he commented? We gotta, we gotta dig in and see what's his name? Michael Skarsgard. I'm gonna. Alexander Skarsgard.

Alexander.

Alexander Skarsgrd.

Nice to hear your voice, Ben.

Yes. Swedish actor. Does he have Alexander Skarsgard? Pop crave has something. Does he have any like social accounts he's on Instagram that he posted about this? No, his last photo was, I don't know, just picture of him looking at the camera really quickly. Gemini 3 Pro. Google's most intelligent model yet. State of the art reasoning, Next level vibe coding and deep multimodal understanding. Yeah, I don't know. We'll have to dig in and fact check this. Put it in the truth zone. But OpenAI is disavowed it. So I mean it seems pretty clear that it's not, it's not a real thing. So. Ars Technica has a. Eric Berg has a piece in Ars Technica. Why would Elon Musk pivot from Mars to the moon all of a sudden? SpaceX has already shifted focus from building a self growing city on the moon. That sounds amazing. As more than 120 million people tuned in the super bowl for kickoff Sunday evening, SpaceX founder Elon Musk turned instead to his social network. There he tapped out an extended message in which he revealed that SpaceX is pivoting from the settlement of Mars to building a self growing city on the moon.

Moon.

I like this. For those unaware, SpaceX has already shifted focus from building a self growing city on the moon to building a self growing city on the moon as we can potentially achieve that in less than 10 years, whereas Mars would take 20 plus years. This is simultaneously a jolting and practical decision coming from Musk. So we'll have to dig in more and talk to. I still want to get Eric Berger on the show because I'm a big fan of his writing or his Technica on space. Yes, Elon is following and unfollowing all sorts of people on the timeline as part of his never ending recruiting recruiting push. Also, Jeff Bezos posted just a picture of a turtle from blue origin. Do we know what's going on here?

Did he send. The turtle theory is like a tortoise in the hare, right? So they're taking a slow approach. They're going to get to the moon first. Elon said this. He said they might get to the moon before us.

Okay. So Elon was sprinting he was the hare sprinting towards Mars and instead, you know, the tortoise is plotting along either way. Good for America, good for the space.

Good to have two Benares going head to head.

Yes, yes. So can Raptor broke it down. People are confused. So let me shed some light. You ever hear the tale of the tortoise and the hare? SpaceX is the hare. Quick, active, but easily distracted by Mars. And AI Blue Origin is the tortoise. Slow but methodical, dead set on, on the main goal, the moon. Bezos clearly implies here that they will beat SpaceX to the moon. Well, I like that it's a new race. It's going to be a lot of fun to watch. Before we bring in our next guest, should we talk about France? You got in trouble.

Let's save this one.

Okay, we'll save.

We did talk about it a little.

Bit on Friday and it's too much fun.

It's too much fun. So we'll dig into that later. Quickly, let me tell you about Cisco. Critical infrastructure for the AI era. Unlock seamless real time experiences and new value with Cisco. And without further ado, let's bring in Dan Romero. Wait, wait, wait.

We're now entering.

Oh.

We are the Lightning Round. The Lambda Lightning Round. That's right.

Thank you.

This is our Lambda Lightning round and we will invite Dan Romero from the Restream waiting room into the TV pin altar. Dan, good to see you. How are you doing?

Good to see you guys.

Welcome to the show.

Great to see you.

Give us the update.

I like this. What is this? A little sweater with a good, good looking good.

Yeah.

I have new jobs. I'm dressing for success.

What's the title? What's the role? What's the company? Break it down for us.

Yeah, so we sold Farcaster last month.

Yeah.

And you know, it's a five year chapter. And now I am working at Tempo which is a purpose built L1 blockchain designed for stablecoin payments. Stripe and Paradigm incubated it and yeah, I am working on the go to market side of things.

Okay.

Makes so much sense. When I saw the news, the first conversation we ever had was about stablecoins. This was back in Venice in probably 2021, 2022. We were both so excited about stablecoins at that time and kind of frustrated with how little they were integrated into the actual tools that businesses and consumers could use every day. And specifically. Yeah, just specifically making them valuable by integrating them with the rest of our global payment network. So the role makes a lot of sense. Yeah, What? I mean, kind of a silly question, but why this opportunity? It's an existing business. You prove to be an extremely formidable founder with Farcaster. But I already know the answer.

Tell the audience.

For you to tell the audience.

Yeah, I mean, look, I've been working in crypto for 12 years, first to Coinbase and then Farcaster, and I feel like honestly, the time spent hasn't really resulted in that much real world impact.

Right.

Like, crypto has been its own thing and obviously the price of bitcoin has gotten pretty high. But I genuinely think with the genius act with which was a law passed last year, the opportunity for stablecoins over the next 18 to 24 months is pretty massive. And obviously I think Stripe figured that out. They bought Bridge last year and then they bought another company, Privy, which I think is a sponsor of tvpn. But yeah, I think banks are taking it pretty seriously. I don't know if you guys saw that Wall Street Journal article where Jamie Dimon had some choice words for Brian Armstrong.

We were hanging with Brian yesterday and, and I told him, I was like, you better have printed that out and framed it immediately because that is, it felt like the whole Coinbase team had been working forever to get that kind of quote because that was the whole ethos of the entire movement from the beginning. It was like, hey, we can actually come in and be a real player in finance.

Well, I mean, I started Coinbase in 2014 and it's the classic, first they laugh at you or at first they ignore you, then they laugh at you, then they fight you. This is clearly the fight you. But I think generally though, there are a lot of banks that are very excited about working with stablecoins. We're really excited about working with banks with stablecoins. I think stablecoins are startling for money is what I've heard from someone say that I think is pretty good. And I think generally you have a payment rail now that is global, fast and cheap, that moves at the speed of developers and frankly now AI right, so agentic payments, you probably have seen a bunch of the stuff with. I know you guys have been talking about Cloudbot and things like that, but that I think is going to be on stablecoin, Rails and so the opportunity is pretty massive.

Gtm, incredibly broad role means you're going to be getting people to use Tempo. What is the role going to look like? I imagine are you interfacing with legacy financial institutions, new neo banks all the way wallets, consumers all the way to governments. I imagine you'll be interacting with those kind of groups as well.

Yeah, everybody and anybody. I was meeting with a bank, I'm in New York right now and I was meeting with the bank this morning. I'm about to be with the one right after this. But also neobanks, developers, anyone who wants to do payments globally, I think is going to be on stable coins. And I think Tempo is very focused on the features on payments. Right. Like, I think there are a lot of blockchains out there that do a lot of different things, but Tempo's functionality is really dialed into payments. I mean, obviously working closely with Stripe on that. And so I think, yeah, that's the focus for the go to market stuff.

Lay out the AI agent thesis a little bit more.

Yeah, why wouldn't I just want to give Open Claw or claudebot or whatever agent I'm using just like my credit card?

Well, I think one thing is the credit card itself is kind of like a private key. And so having that get prompt injected out is maybe not the best thing in the world.

Sure.

And then going and giving a bunch of other agents your credit card, one of those could be malicious and then.

That gets popped where promise I won't charge it again.

So I think naturally that. But also if you have these agent swarms, the idea of spinning up a new credit card for each individual sub agent doesn't make sense. Where you could imagine wallets, you can spin up as many as you want and then be able to kind of manage programmatically the different balances for each agent. But if you think about it like an API call to any of these Frontier Labs right now is a pay per call.

Right.

So if you actually kind of break it down, you're eventually going to get to a point where every single API call you can just pay some amount of stablecoins in the background and keep moving. Right. So I think that's where the world is headed and I think it's on us to kind of pull that forward bit.

Does that have a meaningful impact on the API business? Because most API companies are like, yes, I'm giving you the service piece by piece, but I'll invoice you at the end of the month and it's fine. Is there an impact to cash flow if you pull that through or security or risk or underwriting? Like, what is the benefit if you're, if you're OpenAI or anthropic and you're like, yeah, technically we offer services on a per microsecond basis, but are like the Fortune 500s who are buying our API, like they pay their invoices on time.

Yeah.

Look, there's, there's obviously some amount of cost to using different payment Rails.

Sure.

With stablecoins, it's as cheap as it gets.

Yeah.

I think that there's also some amount of fraud. So when you don't have that because it's a bearer instrument, with stablecoins, that's a settled payment that also benefits. But I just also think that you just kind of have new use cases. Right. So how does an agent decide what service to use to register domain or to host a database? Right now you kind of are using Claude code and then it kicks you out to a web browser and it says, hey, go set up account over here, fill it out when you're done, bring it back and then I can continue what I'm doing. Whereas you could imagine you have cloud code running and it finds a stablecoin native way of doing something, it's just going to sign up and do it for you. Right. And so I think that's the kind of future where it just allows the agent to be unblocked, assuming the agent has some money to spend.

Yeah.

What are your main lessons from the probably hundreds of L1s that have launched in the last decade? There's been been so many that had a lot of potential differentiated, maybe approach from a technology standpoint. I think the opportunity with Tempo, as you're combining, you know, these trusted brands, you know, a deeply experienced team, you guys, I imagine are going to be incredibly focused on, you know, actually, you know, you have the ability to like sign up one partner that could drive billions of dollars of volume, I imagine. And so you guys feel like have all the right ingredients to be successful here. But having studied, I'm sure the last generation, the multiple generation of L1s, what is it going to take to win?

Yeah, So I think the differentiated thing for us, two things. One, the blockchain is specifically focused on payments, so there haven't been too many L1s. I think a lot of L1s tend to try to do everything and we're really, really focused on payments. And the second thing is we have a bunch of initial partners that we've been fortunate enough to work with. And part of that is also having kind of closely been working with Stripe since the beginning. We kind of are able to approach the payments use case specifically. And so you have publicly we're working with partners like Doordash or Klarna and bringing that perspective in, whereas I think traditionally in My decade in crypto, most of the time you launch an L1 and then you go spend a lot of time trying to figure out if anyone wants to work with you. Whereas if I think since the beginning, Tempo, because of kind of like that focus on payments, have had these concrete use cases that we're going to be bringing to market over the next six to eight weeks. We should be live by kind of.

Q2 this year, which is very cool.

From start to finish in less than a year.

Has crypto sentiment ever been worse in your time than right now?

Way, way, way, way. Yeah, this is an easy day.

What was your darkest hour? What was your darkest hour?

I think The Bitcoin Scaling 2015, 2016 Pre kind of Ethereum, it was pretty grim because everyone was assuming all the payments was going to peer to peer electronic cash. That was bitcoin, right?

Yeah, that was the theme.

So the fact that things weren't going anywhere and then I think Ethereum in 2017 really kind of broadened the aperture from kind of just being bitcoin. There were a lot more choice words to describe any other blockchain at the time and it became crypto.

Right.

And all the kind of stuff that's since come from that. So I think things were way More grim in 2015, totally relative to now. I mean you have genius act, clarity is still kind of in play in Congress. I mean we had a Super bowl ad yesterday that everyone was singing. Right. So like I think things are, I mean, 2022 is also pretty bad.

Yeah, yeah.

I think that the real world impact for crypto has never been closer. And so, you know, I'm really, really excited. That's why I wanted to come on the show today.

Yeah, 2022 is wild.

I'm not leaving the one thing that is maintained. I mean I've seen a number of cycles now and the people that I know that didn't quit are the ones that are, are either already retired or could be. But what I appreciate is the ones that could retire and are staying in it again just because the opportunity is brighter than ever. So I'm very excited to see this roll out. And when do you think the average. What's your timeline for the average Internet user? Just consumer might not even know. No, no, no. I was just going to say like to be like, like basically doing something with, with Tempo.

Oh, okay.

I'd say a year from now.

Yeah.

Not maybe average Internet user might be a little extreme, but how about. How about the terminally online or savvy online?

Okay.

I Like that. Yeah. That's good. Well, congratulations on the new role. Thanks so much for stopping by the show.

Yeah, it's great to have.

I'm excited for you guys to talk about France. I. I was. I was.

Oh, yeah, I know.

Knocking at the bit for the story.

That was.

That was an amazing.

How did you process that whole exchange? Because, I mean, have you ever had.

An entire country mad at you?

I was. I was actually. So the real bear case. The real bearish case for France is that I didn't get any death threats. Oh, yeah. I did not get a single.

Very polite over there.

I didn't get a single mean message, basically saying, hey, I'm here in Saint Tropez. We actually have. I have got my own cluster. Yeah, we're gonna. We're gonna. We're gonna win. We're gonna win the ad race.

Yeah. Alive or well?

Alive and well.

I'm usually a big France proponent because they're 70% nuclear, but I do think you won that exchange.

Okay, thank you.

Yeah. Well, that's what Doug. We had Doug from semianalysis on. He was saying, like, hey, that all the energy is there. A bunch of the labs have gone over, tried to get something done and just walked away. And so, as I said, until LVMH is spending 100 billion a year on data center capex, they're not really in the game. They got. They haven't. They clearly haven't read, they haven't studied situational awareness yet.

And Louis Vuitton, hopefully on tempo.

Great to see you, Dan.

Well done.

Thanks for having me. Congrats the whole team.

Goodbye. Let me tell you about the New York Stock Exchange. Want to change the world, Raise capital at the New York Stock Exchange. And guess what? Sydney Sweeney just rang the opening bell for the stock market today at the New York Stock Exchange. Adam Aronson said it was bound to happen because what do Sydney and Sweeney have in common? Nyse. Right in the middle of the Venn diagram, she has the letters NYSE in both her first name and her last name. That is a fantastic coincidence. What an interesting. What an interesting.

Let's go back to Friday, February 6th.

Yes.

Macron posted it in France. Oh, like we'll go back to France.

We will close.

We badly want to talk about France.

But we have Boris Soffman waiting for Bedrock Robotics. He's the CEO and co founder and he's here in the gp. Welcome to the show. Sorry to keep you waiting.

Pleasure to be here. Thank you.

Thanks.

Great to have you.

Do we have a Phone call, like a decade ago about Anki, because we were both in the Andreessen portfolio. Is that possible?

It's very possible. That was quite a journey.

I remember because we were the only two portfolio companies that were like, selling something online. And so I was like, okay, connect maybe with the one person that's not just doing software. I need to know how this works. And you kind of broke it down. And it was very fascinating.

This is before the physical stuff became in style again.

Yeah, yeah, yeah, it is in style.

I had a hard time.

Yeah. But first time on the show. Please introduce yourself and explain what the company does.

Thank you, first of all. Pleasure to be here. Thank you for the invitation, of course. Name is Boris Hoffman. I'm the co founder and CEO of Bedrock Robotics Products. And we're developing autonomy technologies for heavy machinery. And so these are the sort of machines you see in construction, like excavators, wheel loaders, motor graders, compactors, but also in all types of other industries like mining, agriculture, lumber. And so they're the workforce for a pretty sizable percentage of the world's gdp. And these are sectors that are going through pretty massive labor challenges. And so we're bringing autonomy solutions that can enable them to be fully operatorless, to be able to execute work with a lot of flexibility and bypass a lot of the constraints that right now are really challenging for not just the US but the whole world.

Yeah.

Why is there a labor shortage? Why don't people want to drive cool trucks and tractors and mining equipment? It's a dream. Yeah, it feels like a dream.

They always want to drive it for a little bit, but it's a really, really tough, tough work. So, you know, the amount of people going into this line of work is just way smaller than it used to be in the past. You also have a fragmentation where you can be an expert bulldozer operator, but you're not an expert excavator operator. And so you end up having these, you know, these challenges. And what, what's happening is that the average age is getting higher and higher. Retirements are skyrocketing. People are not coming into this line of work. Temperatures are high, vibrations. These machines are really painful. And it oftentimes takes many, many years to become even decently competent at these machines because they're very complicated. And so you end up having giant spikes in demand, for example, from the data center boom and onshoring and manufacturing, housing shortages, infrastructure work. But a lot of our partners are expecting half their workforce to retire in the next seven years. And so it's going the opposite direction in an industry that already had 800,000 worker shortage over the next few years.

Talk about. So it sounds like you guys are building hardware software that can kind of bolt on to existing machinery. Talk about that decision making process. It feels smart for a number of reasons. And I also want to understand how the actual equipment manufacturers are thinking about autonomy. We were talking about Caterpillar last week. It's been, if you were a hardware VC 10 years ago, you probably should have just put all your money in Caterpillar based on how stocks traded double digits every single year for almost a decade. But yeah, talk about the approach that you're taking.

My roots and a lot of our roots from bedrock, we actually came out of Waymo. So I was an executive there for about five years. I was leading trucking, various technology teams supported the cars as well. And in that world we had to redesign the fundamental platform every time you would make a new vehicle. So you want to put a Jaguar on the road or a Dimer Freightliner or a Hyundai, you're doing a multi hundred million dollar many year program because these are very complicated systems that need to be redesigned for safety reasons. The advantage that we have in the space and the approach we're taking is that we can actually retrofit existing heavy machinery. So take a Caterpillar excavator for example. We can turn it into autonomously capable with a sensor compute outfit. And now you take this half a million dollar machine that a general contractor already owns and it could become autonomous for the scope of work that it's cleared for.

Yeah, and with autonomous vehicles, when you look at the cost of Waymo, that's been obviously one of the criticisms of it. If you have a half a million dollar machine and then you're put bolting on 50k or 100k worth of sensors in order to run it autonomously. The economics can make a lot of sense quickly because the operator was getting paid. Maybe you can't find the operator and they also were probably getting paid quite well. So I imagine the economic trade is pretty good.

Economic trade is fantastic because the cost of hardware is also a lot lower because now there's an ecosystem of components being driven by automotive space and others. And so we can enable these machines to be autonomous, you know, for a lot less cost. But then you're able to operate 8 hours a day, 15 hours a day, 24 hours a day. You can compress schedules, you improve safety, you get visibility of your work, higher predictability. And so when you change the physics of how you actually manage the operations of a construction project like this, suddenly you can get a lot more productive and fundamentally expand the sector and you can utilize your machinery a lot more with a lot less constraints, which immediately helps general contractors and subcontractors for our customers. In the end, do work makes sense.

One interesting decision from Waymo is the electrification of the fleet. That feels like it unlocks a specific acceleration curves, very precise. And people really rave about the driving experience sitting in the passenger seat because it's computer controlled. Is there a major barrier to electrification of construction equipment because of the energy density and the work that needs to be done? And maybe charging infrastructure would be. Would you expect an electric tractor to be rolled out in a decade or.

Are we further away from just plug it in like one of those vacuums that has a cord?

Yeah, yeah, yeah. I'm just interested about like electrification of construction equipment broadly. Yeah, it's funny.

Electric things are actually easier to control because they're much more predictable in the output.

That's right.

Of what they have. So there's actually advantages from a vehicle standpoint. The in construction, what's challenging is you get pulled into really strange locations in the middle of nowhere. Infrastructure is just very complex. For example, data centers are getting pulled into strange locations because of availability of power. You end up having a practicality that's tougher in that domain. For us, we're embracing the ecosystem that exists today, which is incredibly designed machines. They just happen to be, you know, still, you know, not, not electric yet. But I believe, you know, when you do a rebel tax, you can have hubs that have like massive large scale charging facilities. It'll take more time, I think, for that to propagate into, into construction where by definition you're going to areas that are pretty GREENFIELD and don't have that infrastructure.

Yeah. Based on everything you've seen over your career, do you think it's a good, good time to start a new AI toy company now?

It's like, yeah, we were maybe a little bit ahead of our time. You know what's interesting is the technology that has skyrocketed over the last five, six years that has enabled autonomy on public roads and industrial equipment. It kind of applies everywhere. And I think more generally the physical world is going to get reinvented because now suddenly you have these incredible interfaces capabilities through like large language models and super sophisticated AI. You have lower cost of hardware. You have cloud computing, you know, cloud AI that can now be on our wireless network connected. So those are building blocks for all, every industry, effectively. And so I think that, you know, we've seen this like transformation on the digital side with, you know, LLMs and OpenAI, Gemini and others. The physical world is still 80% of the world's GDP. This is the future where you apply this to every single sector and every one of them can have a reinvented experience.

Yeah, yeah. I've just thought about like a lot of, I mean the waymos are incredibly reliable, but there's plenty of AI models that have little hallucinations. But if you design the kids toy to sort of embrace that. We talked to what was the company that was. You take a pic or you describe whatever you want and then it prints a sticker. I think it was called sticker box for Kids. It was just like this. It was sort of leaning into all the rough edges of AI in a very positive way. That's like low stakes content becomes like.

So much easier because of the activity. So I think, I mean generally I think it's going to reinvent gaming and entertainment like this just because of the, like you don't have to script the entire experience like you, you used to. The other nice thing is what you mentioned is that, you know, when you are dealing with autonomy in a lot of applications, you're so sensitive to the worst case of what happens, happens. Not the average case, which is what most of AI has to focus on. And so that adds an incredibly high bar to releasing a car and you know, to go driverless in San Francisco or a 110,000 pound machine to go and do excavation project in a, you know, for a factory. There's a lot of other sectors that can benefit from the fact that there's a softer, you know, kind of boundary and resilience to certain mistakes. And in general, I think that you see this shape in every industry. That's why legal LMS or medical LLMs have like a larger bar than a chatbot that maybe keeps you company and does some research. It's not absolutely accuracy critical.

Yeah, oh, sorry.

How many OEMs do you think you guys want to be working with in, let's say five years? Can you reach a, can you reach kind of critical scale working with Caterpillar and John Deere and kind of going down, whatever the kind of power law players or is there more of a long tail that I'm not aware of?

So there's absolutely millions of these machines and almost always the bottleneck is that the labor to operate them is limited. In the United States, every geographical region will have Its biases. Today we're working on Caterpillar machines.

Machines.

Most OEMs are actually moving towards this drive by wire architecture where you can control these machines in a very efficient fashion electrically. That proves to be a big enabler which didn't exist for the car truck space, among other challenges. But what's nice is that the OEM ecosystem is probably not going to be the constraint for many, many years. Mainly because there's such a prevalence of machines. And eventually these technologies will probably influence the buying cycles into the subset of OEMs that actually have the capabilities like this.

Is there an OBD2 port on these Caterpillar vehicles?

Yeah, they have a can bus interface and so you tend to have signals passing through and so you can plug in. And in fact, I think the industry's benefited from the fact that there's a lot of foresight from companies like Caterpillar and Deer to interoperate with high end GPS systems, the driver assist systems that help you auto level a blade, for example. And so you have the prerequisites of autonomy even if the capabilities on the AI side weren't ready yet.

That makes tons of sense.

Jordy, got some news.

Yeah. Tell us about the round. What happened? Let's bring down the mallet from the lambda cloud.

Oh my God. That's a big mallet.

Yeah. We got some heavy machinery here. We got some gong machinery.

We raised $270 million.

Oh my God. We're getting gong.

There we go.

Congratulations.

I'm dramatic.

It's very dramatic here. We like theatrics on this show and we enjoyed talking to you. Thank you so much.

Yeah, I'm super, super excited about what you guys are building and. Yeah. Glad you're doing it. Just like the even just. Yeah. All the speed ups. The more that we can speed up every single project in the real world, from a train to an airport to someone's home, the better the whole economy will feel.

Could be a radiation on a whole sector and hopefully this opens up a pretty broad generation of building. Building at a scale we haven't seen yet.

I love it.

Well, thank you so much.

Yeah, we didn't even get it. We didn't even get into software. I'm hoping it feels like a video game when I'm, you know, when I'm just kind of observing my fleet.

Agent swarms.

Yeah, agent.

The orchestration of these. That's the. That's the future and. Yeah, for sure.

Amazing.

Gentlemen, thank you.

Have a great rest. So great having you on.

We'll talk Soon. Congratulations. Goodbye. In other news, Mr. Beast bought a bank. That's fun.

Yeah.

Beast Industries. This was a Gen Z focused banking app.

Yeah.

Launched in the 2010s. Very hot in the 2020, 2021 fintech era. Remember when every one of these companies was trading at 100 times revenue, they raised something like a couple hundred million dollars of equity, I believe and then a bunch of debt. So anyways, unclear this felt. I don't think this was probably a great outcome for the venture investors involved but who knows if they bought 100% of it or whatever. But yeah, now you have one of the biggest networks and just the infrastructures in the world. I do wonder if they'll be able to to go international from day one. Just given that so much of Mr. Beast's audience is very global but very cool. This was something remember people have been saying since at least kind of that 2021 era. Mr. Beast should have a bank like you should have a effectively a commodity. A business that's differentiated by brand and your ability to get users to sign up in an expensive way and something that he can offer to everyone. Something that you know. So anyways, this and extremely scalable obviously. So I'll be interested to track this as they. I'm assuming every single Mr. Beast video will have a call to action. Call to action very, very shortly.

It's gonna be good. Well, we have Sarah Hooker in the restream waiting room from Adaption Labs. Welcome to the show. Sarah, how are you doing?

Hello.

It's the main to be here. How are you?

Great to have you. I love the background. It looks like a beautiful day in San Francisco.

Are you outside in San Francisco?

I'm not but I'm surrounded by glass so it gives a good illusion. This is the outside. You can see how beautiful it is. And this is inward looking but just high reflection density.

Amazing.

First time on the show. So please introduce yourself and the company.

Fun.

So I did want to say I was trying to. This might as a slight aside before we get started but I was trying, trying to describe to an AI researcher what this show was over the weekend. And what was funny was I was saying it's kind of like CNN New Year's Eve show.

Oh sure.

Every day.

Every day.

Every day. Yeah. That's incredibly dramatic. Yeah.

Yeah.

We should go watch all.

We should have a ball that drops from the ceiling and when it falls.

The show you got to protect that researcher with your life because they're locked in enough to not have ever heard about us.

That's a high value person.

That's a high bullish signal.

That's great.

But I do think it's fun because you have the cerebral from Anderson Cooper and you kind of have the dash of drama from the Bravo. Anyways, just love here. Three months ago I started Adaption with Sudeep who's my co founder. So I overnight success. Oh thank you. That's lovely. Yeah we just closed off $50 million.

There we go.

Get ready for another week of drama.

More drama.

Congratulations.

Anyways. Continue. Continue.

That was a lovely introduction. I thought that was very bombastic but probably more important was it's probably the most important question I'll work on. Most of my career has been in Frontier Labs and building the biggest model that we can that's very performant but this is fundamentally bucking that trend. It's about how do we continue to evolve these models real time. How do we not just build a static model but how does a model adapt to incoming data in a really efficient way?

Learning, continual learning.

Is that the buzzword? Is that a good buzzword? Do you like that for phrase continual learning?

I do because it's actually not a new buzzword. So for the first time we're not introducing a new word which is foundation models was introduced by Stanford. We often see these models kind of introduced as pretty. Continual learning is an old problem. It's just increasingly urgent because it's now within reach in many ways. Like most of the last decade has been like you throw computer at the training pre training but now we have expressive enough models that it can interact and that's fundamental. The question here is how do you interact efficiently with the world? And so I don't mind continual learning. I think it serves its purpose.

That's good.

Yeah. How much are you thinking of an entirely new architecture versus what we're seeing with folks building skills and markdown files, compacting chats, larger context windows. There's so many different ways ways to sort of get continual learning light these days and everyone's solving it in different ways. What do you think the most interesting path is?

Yeah, that's such a fun question and I'll tell you why. Because there's almost like two crucial questions that continual learning has to solve. The most heavy handed version of continual learning is you basically have to train again and that is something that's not all interesting I think to adaption mainly because. Because it's a very high barrier to entry and if we want to adapt and interact you want it to be real time. I would say the Light version you're talking about is probably another least interesting alternative. It's powerful, it's a lever. But it's really that you want to do two things. We want to have control over the stack and be changing the weights without gradients, which I think you can do in powerful ways. Especially here do it jointly with GPU optimization. But the second thing, and this is interesting, is that you really need to care about where you place compute. So the way with monolithic models now I think people intuitively understand we saw the same model at every problem, which is a big waste of compute. And 90% of problems every day that you solve using AI are extremely easy. It's a 10% that matter, that's the long tail. But even there we spend time too much compute because we do these massive rollouts and we only really have good verification for code and design right now. Why? Because those are people who care enough to give a ton of feedback. And that's where we see it working really well. What's interesting is that the component that's most fascinating for us is let's care about interface and let's create really interesting interaction points that are adaptive to each task. And so when we talk about adaptivity it actually includes is the type of feedback you should get change based upon your problems and that's how you make it efficient. So kind of on that spectrum of like heavy handed you train. But really for us our constraint is on the other end. We want it to be real time and we want to use like a strong set of levers to get there.

What's. Oh, sorry, go for it. Yeah, just what's the biggest sort of consensus take in AI right now that you disagree with?

Oh, in AI I mean I think data centers in space is pretty bonkers.

Okay, that's a good one.

I don't know if that's specific to AI, but it just feels topical. So maybe I'll throw that one out. I do a lot of work on systems as well. So serving really fast is really important. The frontier and dynamics are just really tough if you do. I think there's two things that are changing that make it hard to do something that data sciences in space one is most co located. Hardware is pretty much for training. I think that's why you care. Otherwise you can distribute and so inference compute which is where everything's moving. You know, when we talk about real time adaptation a lot of its inference you can spread that compute more easily. You can have multiple data centers. So if you care about space, you probably only Care about training compute. And I think people underestimate the amount of failures that happen and you don't want to get your training job interrupted.

Yeah, unpack that more. Because it seems like if things are moving towards inference and inference does not need to be co located, having inference happen on a single maybe wafer scale system on a chip in space, that seems like more possible. In that case, if we do move to inference and maybe the training still does happen in the co located data center, but then the inference happens on the space data center, is that possible or is there some logical inconsistency there that I'm missing?

So you can distribute inference, but to be honest, that's pretty easy to do on earth. Right. Because you have less constraints that be co located. The real shortage of data centers and where providers like Google is around training, it's around training.

Got it.

The truth is you can do inference distribution much more easily, which is. But the real issue frankly is that GPUs still have failure rates. So there's 2% of GPUs that just started considered done every year. You don't try and revive them from the dead. And that's really your cost. It's how quickly you can replace those and what it looks like.

Very interesting.

What do you think your first customers will look like?

Like.

For us we want to make workloads adaptable all the way from data to interface. There's two use cases for adaption that are both pretty powerful. The first one we're focusing on is customization Right now. If you are a developer, you typically have tried fine tuning. You haven't succeeded because it's too much. You have to bring your data, you have to wait and then you become a prompt engineer again. Most rapidly growing companies is just have a ton of prompt scaffolding. And for us, like frankly, if I think about our measure of progress, it's that we eliminate prompt engineering because that's really, I think intuitively it's a desire for control, but it's also an acknowledgement. Models don't work for people.

Well, congratulations on the massive round and all the progress.

I'm sure you'll be back on this year.

Have a great rest of your day and enjoy Sunny San Francisco. What a beautiful.

And we're gonna go watch CNN ball drops from the last like hundred years and just game, tape, game tape, study up. So thank you for the inspiration.

New Year's evil. You'll get the sense it's very fun banter and I think that's the cross section that you both capture, which is kind of fun.

Style and substance. Hopefully. Hopefully. Anyway, thank you so much for coming on, Sarah. Congrats to the team. Bye. And without further ado, we have Ed from Machina Labs in the studio. Fantastic. Welcome to the show.

What's happening?

Come on, let's. Let's. Let's kick off.

Whoa. Got some heavy metal.

Can we.

Oh, can we knock this? Let's knock the dog.

What do you got?

Sit down. Tell us the number. 124 men.

So you guys put our ad and our logo in your ad yesterday. We thought we should put you in metal.

Amazing.

Amazing.

Nathan, our new engineer, took some time off from customizing cars, and so put this. Put this in. Yeah.

So very, very cool.

Okay, so this is metal deforming. Explain what's going on here. Explain what machine made this.

Yeah. So we build a machine called Robocraft. So it basically does different types of metal operations like a craftsman would. The core operation is sheet forming. This is a slight kind of indentation to kind of get your logo out. But we can form, you know, aircraft panels.

Sure.

Airframes, you know, car bodies. We do a lot of customization work for Toyota right now, so. But the system is designed to do all kinds of manufacturing operation. You know, you kind of think about it as kind of versatile manufacturing robot that can do different types of operations just through software.

And what unlocked the new round? Is it a partnership with a big automotive OEM or just sort of broad adoption, unlocking certain technical milestones? Like what jumped off the first slide of the pitch deck.

Yeah. So we're now going actually from kind of like technology development at this stage to actually scaling. Right. So we are deploying at facility number three. Facility number four.

Right.

So the third facility is going to be 250,000 square foot facility, which is twice the size, more than twice the size of our, you know, second facility, which is 75,000 square foot right here in the. In the valley.

So for the.

And also we're going to expand the scope of operation. So we initially built a platform to do sheet forming. Now we're adding welding, we're adding assembly machining, so we can do full end to end, you know, complex metal structure assemblies for our customers. For example, we're doing a lot of work right now with aerospace primes to do missile airframes. So this actually, this first facility is going to do that at larger volumes. We're going to do thousands of missile airframes a year out of that facility. But the beauty is that tomorrow you can just switch it and do aircraft panels or do car and automotive panels.

Yeah. Do you want to ever allow someone the flexibility of something that feels almost like 3D printing? You know you're able to do R and D low volume, but then when someone says okay, I'm going to make the cybertruck, maybe you need a giga press. Is that something you want to get into or is that something you see as like antiquated? We won't need stamping anymore in the future.

Yeah, I think, well, in the short term I think we could cover up to a few thousand a year volumes.

Right.

And the big question is long term, you know, what does the manufacturing will look like. My view is that long term we're going to have less high volume of making the same thing over and over again and more just keep changing. And you look at the software, you know how many apps you have on your phone? Hundreds. Like you can download thousands.

Yeah. With software used to make it, it would get, you'd put it on a disk, sell it and just pray basically.

Right, right. You treat your car, you don't to have have a daily and a weekend. You have different panels that you chunk off changing.

I mean Jeep does that.

Yeah, they do, they do already.

Great.

Yeah, yeah.

So that, that world is, I think, you know, you look at any sci fi book, you know, it just tells you about this world where you know, like every building looks different, every car looks different.

There's you know, basically a lot more.

Version of a lot more variety. Yeah. So I think in long term we're probably not going to have as much of like make the same thing over and over again.

Yeah, yeah, yeah. We talked to Lucas singer about 3D printing for the hypercar he built. It feels like there's certain trade offs that he's optimizing for where certain weight to strength ratios are only unlockable with certain structures. They look more organic. What are the other optimization parameters that you're looking at? Yeah, go for it. Yeah. I'm just thinking like cost, speed, flexibility, anything else that's like jumping off the page.

Yeah.

Right off the bat is just a speed. That's one. Also we start out of defense because like you know, that's where you want, you want to have a new idea, a new weapon system. You want to scale it really fast. But not only just scale it fast.

Quickly, like quantize that speed because I've seen, you know, the image that I'm conjuring, you can correct me, is like two massive like sort of robotic arms Pushing against this metal sheet and it's working for like a few hours. Is that right? Is that hours? Days? How long are we.

Yeah.

So depending on the geometry, you know, the tak time could be a few hours up to maybe a day. Okay, Right. But you know, you have to compare it to the traditional part of dynamite. You have to go make a dies and tooling.

Sure.

So now you're looking at three months up to a year, depending on how many tools you need. Can potentially be longer before you even get your first.

Oh yeah.

And that wait time's terrible. I've seen that for like even just like plastic forming, it takes forever. Injection molding, like you make the mold, that makes the mold, that makes the mold. Seven steps and every. And then someone goes on new year.

And it fundamentally comes down to like, okay, what is the dollars went into building a part because you can horizontally scale.

Yeah.

Right.

So if, let's say, hey, you know, I have to spend $200 million to set up a facility and pay for all the dies to get to production. Okay. With $200 million, how many of robo craftsmen can run in parallel?

Sure, sure.

And that's where it becomes a little bit more relevant because we get speed through parallelization versus you know, traditional manufacturing gets to speed through decreasing tak time. Right. Like I do it fast, but I have one assembly line that things go through needs to be fast versus we're saying, okay, have a matrix operation, have 200 of these cells, same cost, but then to get you same throughput. But tomorrow you want to change from, you know, model A to model B. You can just do that.

Yeah. Who.

Who are you actually competing with? Is it. Is it operations in China? Like is that. Is that like when you. When you're really. When somebody's kind of comping you. I imagine there's not. There just can't be that many super tech enabled kind of manufacturing companies of the future in Southern California to where I often imagine your competition is overseas.

Yeah, no, so I think, well, it could be for some of the sectors in commercial world. Yes. But then in defense, you know, these are already within the United States.

But I meant more like in auto maybe manufacturing.

Yeah. So I think, you know, going back to your question around like what do we enable? Speed is one thing, but a lot of times we also enable something that traditionally was just not possible. That's what we're doing with auto.

Right.

Traditionally you could not go. No matter if you did in China or here, you still have to make dies. And unit economics doesn't make sense if you have to customize and make a new die and new mold for every panel or every custom car you're going to build. So in those cases, we're actually creating a completely new paradigm. Like this is a new business model that Toyota is going after. Like, okay, you know, John or Jordan can go on a website and say, hey, I want this Toyota Tacoma and I want tvpn. You guys have get fleet vehicles at some point for you. So put TVPN on the side door and I want 200 of them.

Right? Yeah.

Toyota would be like, this is not impossible. It's just not possible. Right.

So there's actually a big market, I imagine thinking electricians, like all these different trades where you would want the customer custom truck because you're like, we're going to run this into the ground. It's not like a consumer buying something that, that maybe they're like, oh, I'll be in it for a few years.

I just want a sticker on the side.

We want something.

Makes a lot of sense. I have an idea of what your capex looks like. I mean, you're buying big robotic arms. What does the OPEX look like in a robotic factory? Are you buying a lot of like oil cans? Like squeaking? Yeah. Like what? Like what are the, what are the parts that break when you're running? A lot of robotic arms.

Yeah.

So I think like robot maintenance is one part.

Yeah, right.

Like, you know, changing oil.

Yeah.

And every once in a while you have gearboxes that wear ass.

Okay.

Change those things. But motors too.

Do motors burn out?

Motors actually are pretty, I think, you know, I've had, I've been working with robots. This company in the previous one, Relativity Space.

Yeah.

Never had a motor give on me.

Wow.

Really?

I've had gearboxes give on me. Power supplies, but not motors.

When people talk about humanoids often the criticism is like, hey, you have all these different motors that need to function and they're going to be under a lot of strain. And maybe that's kind of a risk from a depreciation standpoint or just operating expense.

Yeah. I mean, at the end of the day, I think the way I think about it is if you go like, very fundamental. Right. The bill of material that goes into to making a robot is not that significantly different than a bill of material to go into building a die. The difference is a die and a mold is a dumb material versus a robot is an intelligent material that can change its configuration. So I mean, it's the same argument even with Molds like you know, if you have your gearboxes kind of like, you know, getting worn out dies. Get worn out, you have to repair dies.

Right.

So the argument applies either way. It's a question of like which one is a more longer term roi, which is usually intelligence has more longer term ROI than dumb.

Are you scaling outside of California yet?

Yeah.

So third facility is not going to be in California. So volume production, I think obviously California we have a lot of engineers so we want our RD development be here. But productions we're looking at Texas now for the third facility. New Mexico, Nevada recently and New Mexico. Mexico and Alabama.

How what's the process with those states? Are they getting or is it fairly competitive? Is there states that are leaning does? Because I imagine Texas has a lot of momentum. If you're one of these other states, you maybe offer more incentives to get your business.

Yeah, I think Texas has the benefit. Like you know, we're looking at Austin, a nearby area like Los Angeles. SpaceX used to.

Yeah, people want to live there. Exactly, that's a benefit.

But then you go to New Mexico, New Mexico has, you know, actually they have a sovereign fund that can invest into facilities that you're going to deploy. Obviously taxes are, you know, estate taxes are minimal in all three. You know, Texas is more property related taxes. So yeah, it's different for each state. But you're right, I think Texas is becoming that premium state. So they value themselves a little bit.

Higher than what was the 10 year pitch for investors? Obviously they're investing based on the, you know, everything you've done to date and the current traction. But if when everything goes right, what does the business look like?

Custom hoods for Ford Raptors. What else do you need to know? Don't even open the dock set.

Right.

Just wire the money. Exactly.

I think, I think you know what excites us and also like everybody in the company, right. Is the world where, you know, when I got out of school, you know, in order to build hardware you have to work in big companies. Like you know, if you're a software engineer though, you know, you can have a small team put stuff on aws, suddenly your app is accessed by, you know, millions of people. If you're a hardware engineer, you have to go work for like you know, big companies, Toyotas, you know, Boeing. Why? Because your idea requires a lot of capex to be built. So long term vision is can we make that happen for hardware? Can we make what happen for. With software? For hyper, where you have an idea, you go on a portal The AI on it guides you to turn your intention into a design that's manufacturable. And you say, hey, I want 200 of these in Hollywood, California, and the right facility that might be in South Bay or in the Bay area gets programmed and makes it for you. And a week later you get a ship to your door.

Tell us about the abstraction layers on top of the robots. You obviously buy robots that are off the shelf. I mean, they're made by large corporations, but once those arrive, I'm sure they have some computer control system. Have you had to build an abstraction layer on top of that or do you have a translation layer from a design? You put in an image and that turns it into a 3D point cloud or something and then translates into robotic motions. What was the process to actually fire it up? Because you're not controlling this with an Xbox?

No, in early days we did, but yeah, so there's two pieces of software that we developed. So one, we call it offline programming software, where it takes a design and then modifies the design, then prepares instructions that can be sent to the robot. And then once the robot is done, it gets the results from all the sensors and allows you to do some analysis or even train models afterwards. Words that gives you better process parameters for the next time. So that's our offline software. We call it architect. And then there's a software that controls robots every 4 milliseconds, look at the sensor data and say, okay, how did you do? And do I want to adjust Right. So I can improve it. So both of the software is internal. You know, we did decide to try to be as much as we can off the shelf. So 70% of our hardware BOM is off the shelf. There's 30%, like things that the robot pick up or the way we hold the material or the whole platform itself because it's a portable platform. Those are built and designed by us. But we wanted to be able to go to market with limited amount of capex requirements for something that can be easily financed with debt, as opposed to try to spend a lot of equity dollars on building custom machines. But yeah, so Software is all US hardware side, 30% is designed and 70% off.

Talk more about the capital sources. What's the role of debt in this business? What's the role of strategic partnerships? It feels like at this level of the stack, we often see companies teaming up with bigger players. What are the pitfalls of that? Best practices?

Yeah, I think that's one of the things that I learned is a Company like this, you need to basically have very diverse, diversify capital stack.

Right.

So obviously you get equity dollars for development but I think the best use of the equity dollars is just put into engineering. Build something that gives you long term mood. Most of our, you know, capex is fine finance with debt even up to this day.

Get a mortgage basically.

Yes, you basically get a mortgage against the asset. You know, it's still pretty cheap, used to be much cheaper. Now it's like, you know, you know, maybe 2%, 3% above the prime. Right. So, so and then, and then I think customers play a big role. We're such a fundamental change.

Are those loans typically interest only or are you paying down the principal over like 10 year depreciation schedule?

Yeah, so like you do either all flavors, right. So we have like some of them that like you know, couple of years interest rate only and then you have five, six years of repayment. Got it.

But you know, just like a house but a little bit shorter because it matches the depreciation schedule. Makes sense.

And then obviously you brought customers. I think for us it is actually pretty crucial. Like most of the time people think of, okay, if you're a venture funded business, don't go to customers, they change your direction, maybe they give you a little bit, they kind of force you to do things you might not want to do. I think in our case it's such a fundamental change. I mean Toyotas of the world, aerospace problems of the world, manufacturing is their thing. Right. So I found actually it's better to partner with them and actually get them involved because some of these manufacturing changes takes five, six years to do.

Right.

So if the executives, even the CEO of the prime is not in it, doesn't have a stake in the company long term, there might not be the success might be a little bit iffy because they're not committed for the long.

Right.

And then I think the last source of capital, I think this is. We talk about manufacturing as a national security kind of advantage. So a lot of governments, we work with a lot of government to provide kind of certain level of support so that we can deploy these factories in their countries. We recently announced a partnership with uae, so that's going to be where our fourth facility is.

Yeah, they have insane incentives around. They have so much power and so much land and they just want things to be made there.

Yeah. And they have a huge, I mean that area, I mean speaking of, I call it like, you know, Abrams Accord 2.0. That area is like so much going on there in of terms of instability. So all the kind of sane countries in the region are trying to have some level of defenses and I think UAE is doing a smart thing. They want to become the defense manufacturer in the region next to Turkey and Israel. So spending a lot of money and time on it.

That's great. How big is the company? What's the hiring plan? Big new round.

So right now we are leftly 70 people planning to get to 240 in the next year and a half.

Wow.

Yeah, that's a lot of people. Is that manufacturing jobs?

Sort of.

Is there blue collar element to any of the work?

Yeah, yeah.

So we still need a lot of technicians.

Technicians, facilities. That's right.

I think, you know, there's a lot of scare when people talk about, you know, kind of AI and like automation is like, okay, where are the jobs going? I think we're actually going to see revival of blue color a little bit. I think the blue color is not just going to be not as blue as it used to be. Like they're going to be, you know, they're going to be working with rural robots and iPads and you know, they're going to be enhanced. But yeah, we still significantly need, you know, technicians, manufacturing engineers on the shop.

Floor operating these systems. It makes a lot of sense. Jordy, anything else?

No. Thank you for this wonderful gift.

This is a wonderful gift. Thank you so much.

Awesome.

Thank you.

Thank you. Great to meet in person.

Good to meet you as well.

Have a great rest of your day.

Impressive.

We'll talk to you soon. We got to talk about France, right? Did we get to France? We got to close out with France. Tell us what happened, Jordy. They, they invested a lot of money.

They did, you know, they did.

How much have you invested in AI? You know, 30 million euros.

Macron posted on Thursday at 11:53am Pacific that science has found its home through France 2030. We invested more than 30 million to advance AI and a number of other initiatives and anyways, I didn't actually see it until Steven from Lambda sent it to me.

Throwing him in the. Throwing him in front of the bus.

No, he just, you know, it was important news. I think he just wanted to share. He wanted to make sure that I saw it the next morning I shared. Breaking France is going all in on AI with their last 30 million and used this wonderful image and it wasn't, I guess like it took a few hours but the French responded via the account, the official government account. French response and they said, breaking Jordy Hayes can't Tell the difference between an investment in AI and academic grants for semesters in the south of France. Can we pull this up, guys?

Yeah, here we go.

You kind of have to see it to believe it.

It's crazy.

My original post did fully quote Twitter tweeted. Yeah, they quote, tweeted and tagged you by name. They're quote tweeting podcasters.

They are.

But my post, of course, outperformed. I got a million views, 13,000 likes. They managed to rack up almost half the likes. 300,000 views. Not bad for what it's worth. But the problem was like, what does this even mean? So they're saying they're not investing in AI, they're just giving academic grants for semesters in the south of France. And I was like, you're. You're supposed to be investing in AI, showing that you are investing in AI, and you're telling me that your whole initiative, France 2030, is just French vacations for researchers. Like, you're not, you're not giving me a lot of confidence here. This is, yeah, kind of insane. Macron later, kind of. He, he, he made a more general response, which again came with a of lot long post. He came in with a long post.

He sent you with the long post.

And so if you're responding to people that are calling you a clown, I wouldn't start your own post in response with just quote, this clown wants to make France.

Also, you didn't call him a clown. To be clear.

You used that term. I would never do that. I respect you.

Compared to Matt Damon in Rounders.

I love France. I love France. I've been multiple times in the last few years. He started his check, your passport is still out. He said it, though. He said it. This clown wants to make France an AI leader with 30 mil. And. And then he goes on and on and on with. With a lot of basically coping and seething. He, he then shows. Says hashtag for sure. And then he has.

He.

He showed a chart of just foreign investment in data centers. And, and it's apparently 69 billion.

Okay.

It's going into France, mogging the United States. States.

Yeah. We only have 27 of foreign.

Completely disregarding the central bank.

Didn't they put in way more? I mean, I guess it's OpenAI and then OpenAI funnels.

Who knows, who knows? But obviously it doesn't matter because you need to look at foreign and domestic if you want to get an actual, actually accurate view. I responded, I'm sorry, but until LVMH is spending 100 billion a year on data center capital. It's hard to believe you guys are taking AI seriously and I honestly think, think I know this sounds like really insane, but in a fast takeoff scenario you would imagine that France actually getting serious about AI is getting their national champions of all types. Lvmh, even the brands. Wasn't Christian Dior actually owned by a big infrastructure company back in the day? And so I could imagine the Arnaults getting back in the game at some point. Adornai Anyways, a lot, a lot of fun. No disrespect to France. I think we were just having fun, having a little fun goofing around on the timeline. I love France. I'm excited for them to.

I like this.

I'm excited for my next trip to France.

Ray says hello. Emmanuel Macron. If you instead threw a week long hackathon at the palace of versailles with a $30 million 30 million euro cash prize, it would unironically boost the French AI startup scene way more than investing it.

Palmer Lucky said. I hope the researchers are better at making charts than whoever made this one.

Oh no.

Dr. Parikh Patel says you are not a clown, you are the entire circus.

It's a rough day on the Internet.

But you know he's the best. There's somewhat of a language barrier here. This, this again. They and, and honestly hire Lulu, that's all.

So I mean unclear if they even need to invest that much. I mean there's an entirely different world where you just become receptive to other companies coming there. They bring jobs, they bring economic opportunity, commerce. Like you don't necessarily need to build a competitor to Amazon. You can have Amazon come to town. Your benefit, you tax the local business and then you employ local workers and your consumers benefit from the experience variants of Amazon. You don't necessarily need a local copy of everything. Now AI might be different and you might need sovereign AI and there's a discussion around that. But there are countries that will just say yeah, you know what, like GPT5 is good enough for us. We're just going to like partner with Google for this stuff anyway.

We should have someone from Mistral on.

That would be good. Check in with them, see how they're doing. I think they found a nice little, little pocket of business over there.

Well, it has been a fantastic show today.

Thank you for tuning in to our super bowl review special. We've planted the bomb. We will see you tomorrow at 11am Pacific.

We'll be playing with Legos.

Yes. Ramp sent us these wonderful Legos Brian's office. Legos.

Office. Real. Very real.

There you go.

Hard to believe.

Really, really cool.

Fantastic.

I mean, really.

Just shows you the. The versatility of this execution. Just all over every. Every touch point.

I want you to have the best afternoon of your life.

Computer.

Give every person in the chat the best afternoon.com newsletter.

Goodbye.

Nice work, brothers.

I'll see you on the next one.