Meta Poaches Apple's Top AI Executive, Replit Partners with Microsoft, Unilever Goes Viral with AI | Alex Roy, Ryan Petersen, Jonathon Barkl, Jonathan Ross, Craig Piggott, Jacob Glanville
8 Jul 2025 · 144 min
This is a podcast titled "Meta Poaches Apple's Top AI Executive, Replit Partners with Microsoft, Unilever Goes Viral with AI | Alex Roy, Ryan Petersen, Jonathon Barkl, Jonathan Ross, Craig Piggott, Jacob Glanville" recorded in July 2025. The episode features multiple hosts including Ryan Peterson, Packy McCormick, Jordy, and guests such as Alex Roy. The discussion covers a wide range of topics including AI industry talent moves, corporate strategies in AI, EV and autonomous vehicle technology, and cultural phenomena like the Cannonball Run. The content is a multi-speaker conversation but does not include John Coogan as a participant.
You're watching TVPN.
We started 11am sharp with this show. No technical difficulties. No one can stop us from streaming. No one can stop us from podcasting.
Bad Day didn't want us to podcast.
Yeah. They said we couldn't do it. They said we couldn't do it.
We're still not sure exactly, but we.
Did and we're live, baby.
We're not sure exactly. The cause could have been a sophisticated state actor, could have been super intelligence. Super intelligence gone rogue. They know that we're the last thing standing between humanity.
People say this. The last jobs will be live streamer. AI, thumbnail artist.
Thumbnail artist. Those are actually the only two things. Eventually the AI research will just be automated as well.
Yeah, exactly.
Anyways, it's good news.
Speaking of AI research, we have a ton of stories in the AI in the AI trade deal world. The biggest one is that Zuckerberg continues to be on a tear. Meta snatches Apple's AI chief they went and poached from OpenAI. Now Mark is poaching from Apple 10 million plus pay packages. The rumor as Super Intelligence hires Mount Ruman Roming Pang, who led Apple's 100 person Foundation Models team is leaving for Meta Superintelligence group amid growing internal turmoil at Apple over its AI strategy. We've talked about this a bunch. Pong or P. Pang, I'm not exactly sure how to pronounce that joins a wave of top tier hires from Mark Zuckerberg including Alexander Wang, Nat Friedman, ex anthropic and OpenAI researchers lured by multimillion dollar compensation and personal recruiting pitches from Zuckerberg himself.
Phone calls.
We, and we talked about this with, with Dwarkesh yesterday. He's still, he's still like they're underpaying them like it's so valuable. And so yeah, the numbers are staggering. But if you're a multi trillion dollar company and how are you, how is the board, how are the shareholders not okay with you or like okay with you not offering like top dollar to actually get the best talent? Maybe we should pull up the chart of the Mag 7, see how the different companies are doing the big tech.
And yeah, the other thing here, second.
To last and Zuck is in founder mode. He's, he's acting like he's on top of that chart. He's acting like he wants to be on the top of that chart.
That's right.
So I think that's what he's doing.
And if you remember Apple's MLX team, a group of people almost Left only.
That's right.
A little bit ago. And Apple had to renegotiate their contract.
I remember that.
Yeah, yeah, yeah, yeah.
Weren't able to retain Pang Pong. Yeah but.
And yeah this team was once central to Apple intelligence now faces a major reorg under Z Fang Chen. As more engineers the exits they got to figure out the comp structure at Apple. I bet that they're thinking about this now. Well, Mark Gurman has a post here. We're back to printing posts folks. He says this is one of the most significant AI job changes since the beginning of the generative AI era started a few years ago. Ruming is Apple's most respected AI researcher and engineer and his departure amid Apple's AI terminal turmoil could have a reverberating impact breaking Apple's top executive and distinguishing engineer. And of course there was a jokey post from Gabe who says Apple's top executive in charge of AI is leaving for Meta's group. And Gabe says bearish for both companies. Very rude. This is, this is obviously a very talented individual. Well really quickly let me tell you about ramp.com time is money save. Both easy use, corporate cards, bill payments, accounting and a whole lot more all in one place. Go to ramp.com and then speaking of OpenAI, Didi D. Doss, friend of the show has a post. OpenAI paid an average of $733,000 per year across 6,000 employees in stock. Nearly three times every single other public company. Not bad.
Stock compensation was at 119% of revenue.
Yes. So this is non cash compensation. So it's not putting them in, you know, disastrous financial condition but there is a lot of dilution there and I'm sure investors are thinking about that. And what's interesting is that the information is an article that comps how the stock based compensation at OpenAI tracks with other major companies. So at OpenAI they're making, they spent $4.4 billion on stock based compensation. That's 119% of total revenue at Google. Before the IPO Google was spending 16% of revenue on stock based compensation. Facebook was only spending 6% of revenue. So the delta between how much more R and D they have to do, how much more salary do they have to pay before the revenue catches up is crazy.
It's not just stock based comp too. It's it's all everything going into training.
It's very expensive. Yeah, I mean everyone posts that like here are the Google financials. Every VC Everndell's posted this before From Kleiner saying like, oh, I just want just one Google in my portfolio. Because the company was incredibly financially stable, even pre ipo, like within the first few years it was like everything looked great, like amazing revenue growth, all the costs were under control, cash flow, like everything was great.
The meme was that they were spending on crazy food and purse massages because they wanted to be. We're like, hey, we don't want people to think we're too much of a monopoly. The thing here that does make sense is, you know, OpenAI is the first real credible threat to Google search monopoly. And it makes sense that in order to compete with a multi trillion dollar company, you're gonna have to not necessarily act like a five year old company. You're gonna have to spend in a bunch of these different categories well beyond what it would a traditional startup.
Yeah, yeah, I completely agree. It's like it's a new category like LLM chat Interface, but every Mag7 company is threatened by you and so they're all like, I'm going to pay so much to fight. And even if that's, even if the, like the aggregation theory and the default app of OpenAI and ChatGPT on the home row, like even if that's not really that much at risk, it's just distorting the hiring market and you gotta hire people.
Yeah. And some people were calling out too that OpenAI, while it serves as a functional tool for a lot of people, it's also serving as this social product as well. And people are spending many hours in the app on a daily basis and that becomes a threat to Meta. So purely from Meta wanting to. They're competing for consumer attention, right?
Yeah. And I mean the same dynamic plays out in the GPU market. If there's a massive capital war for GPUs, Nvidia is supply constrained. The price goes up and that's what's driving the cost of all these GPUs way up. OpenAI is forecasted to spend 6 billion this year with compute costs. And talent retention has become expensive as inference itself. Investors now face mounting dilution risks plus the wild card of potential Elon Musk settlement. So there's a lot going on over there. Anyway, let me tell you about figma.com, think bigger, build faster. Figma helps design and development teams build great products together. Get started for free figma.com if you're.
Not on Figma already. What are you doing?
What are you doing?
What are you doing?
Well, you're probably not working at Jane Street. Jane street is $4.3 billion India Options Trade triggers market ban and crackdown.
Jane street was like, you know what we need another PR crisis.
We got the apology form here. We got the apology form to Jane Street Capital. Why, why, why were you hating on Jane Street? I thought Indian regulators were like, good. I never actually looked at the Indian options market. Look, that South Sudan thing kind of shook me and I had a moment of weakness. I have a poor understanding of what counts as arbitrage. Wait, what are options? Again, I just like dunking on prop trading firms and then at the end you sign. I hereby respect Jane Street Capital and will not talk down on the future first ballot.
Hall of Fam post from Alice at JustPlan.
Yes. Fantastic. And so.
Yeah, there's been a lot of pushback here. Matt Levine broke it down, but also was getting tremendous pushback on his posts. People were accusing him of being in the pocket of big Jane street high frequency trading.
Big high frequency trading. Well, here's the high level on the story. So a secretive arbitrage strategy that netted Jane Street $4.3 billion in India's booming options market has now led to trading ban and a 5 and a $570 million seizure order by India's security regulator Sebi. The firm allegedly manipulated the nifty bank index by moving thinly traded underlying stocks to profit off of a far larger options positions trading up. Up to more. Up to 7.3 times more in derivatives than cash. While SEB frames it as extending, marking the close, Jane street calls it textbook arbitrage, exploiting a massive price gap between overhyped options and underpriced stocks. SCBI launched the probe after Jane Street's lawsuit with Lo Millennium exposed the strategy's existence. The Indian markets imbalance, where options trading volume can exceed underlying stocks. Cash stocks 353 to 1 left retail traders vulnerable and regulators on edge.
I had no idea that options were so popular in India.
I had no idea. Yeah, so Jane street did not want anyone to know about this. Basically what happened was there was a few traders, Jane street that discovered this strategy. They were studying the Indian market. They realized that there was this massive, massive options market relative to the underlying stocks. And so there's an opportunity to arbitrage by buying certain. Buying the derivatives and selling the stocks.
Or vice versa and so getting tricky at the close.
Yeah, so, so, so they were.
What happened? The, the, at least some of that initial group that discovered this trade and were executing it went to Millennium.
Yep.
Then Jane street sued Millennium for trade secrets and through the lawsuit everybody was being very vague about.
No, no, not everyone. Just, just drain street was, Jane street was like we were operating a unique strategy which we won't tell you about in redacted market. They didn't even want to say India. Then Millennium gets in the courtroom and is like oh, our India strategy where we do exactly this. And the Millennium lawyers are like we, this here's exactly what we did. And then the Indian regulators are like wait, wait, both of these firms are trying to do this over here. Like get out.
Is interesting that they, the regulators had to discover this through the courts.
Yeah, I mean America. Yeah, yeah. You wouldn't really notice it because it's just like oh, okay. The, these two assets are, you know, pushing closer to alignment value.
Basically saying, your honor, it's called finesse.
It's called finesse. Well, speaking of finesse, nothing has more finesse than Vanta Automate Compliance. Manage risk, Prove trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation. Whether you're pursuing your first framework or managing a complex program.
It's time to soc too.
It's time to soc too. In other news, Sean McGuire, friend of the show, multiple time guest, is in hot water with, I don't know, the left, I guess he is being attacked for attacking Mamdani, the Democrat nominee for mayor of New York City. Pavel Asperuhov, also a friend of the show, has been, has a case study here.
Let's pull up the polymarket by the way.
Oh yeah, yeah, I want to see. So people are calling for Sean McGuire to either be disciplined or step down or leave Sequoia. But the polymarket has it.
17% seems unlikely and there is a website that is that people are signing.
It's like seanmeyer XYZ or something like that.
And yeah, anyways, if you want to kind of.
So Pavel says this is an interesting case study in media being clearly disconnected from the cultural pendulum. This is obviously not going to result in any sort of action from Sequoia. And if you had any pulse on things you would know that. And so Sean McGuire rides on and is continuing.
Yeah, I don't, I don't see anything happening here. This has happened a few times. He probably regrets the way in which he phrased the initial post, yet clarified it, made a 30 minute video explaining explaining it. I'm sure way more people saw the original post than saw the video.
Yep.
But yeah, hoping everybody can resolve their differences and yeah, I don't see him going anywhere. I think he, there was a similar.
Call to push him out of Sequoia during the Trump stuff because he was like the first major Sequoia partner to really be vocal about supporting Trump and there was some backlash to that. There's some questions about whether or not Sequoia would have his back there, and they did and so we'll see how it goes. But seems like he's, he's, he's, I, I think he's enjoying the fight online. He likes a good, he likes a good fight online. He enjoys it. Wrestling.
Wrestling in the mud.
He's wrestling in the mud for sure. Anyway, let me tell you about Linear. Linear is a purpose built tool for planning and building products. Meet the system for modern software development, streamline issues, projects and product roadmaps. In other news, Replit and Microsoft are bringing Vibe coding to Azure. This is an exciting deal bringing it.
To the enterprise low code platform. Replit will integrate directly with Azure Container Apps and Neon Serverless Postgres, letting enterprise employees build tests and ship software by simply describing it in natural language. Microsoft customers can soon buy Replit seats through the Azure marketplace and deploy and deploy it to managed Azure infrastructure, marrying Relet's agent, a code generator with Microsoft's compliance and governance stack. Replit already counts 500,000 business owners and says the deal makes anyone in the org a developer. So Amjad has been on an absolute tear. Tons of bottoms up adoption from business owners and I'm excited to see this.
He's also been really good at working with the hyperscalers like he had an early Google deal, now he's got a Microsoft deal. Somehow he's made both of those deals work, which is interesting because typically they're like if we go with you, we want you to stick with us only. But Microsoft's taken this very agnostic approach to they serve everything from OpenAI to Grok and then they'll also serve Llama and they'll also serve deepseek. Microsoft really wants to provide every possible model, every possible endpoint for AI so that you stay on Azure and you can use everything you want. So very excited for Amjad to be teaming up with Microsoft. Very exciting stuff.
Meanwhile, Jack Dorsey back in the mix. He is launching a new Bit Chat, a Bluetooth only WhatsApp rival block CEO and Twitter co founder Jack Dorsey spent the holiday weekend coding Bitchat and an open source messenger that passes end to end encrypted text over Bluetooth or local wi fi, no SIM card data plan or account required. The test flight beta and white paper promise off grid comms at concerts, protests or disaster zones. And cites Apple's airdrop bans as motivation. Although early the peer to peer architecture taps the same NOSTR protocol Dorsey champions in its decentralized social vision. So good to see one of the all time greats back coding on a weekend. Shipping people are also. You can, I'm sure would guess this, but leveraging bitch. Bitcoin support. Sending bitcoin back and forth so makes a lot of sense.
Yeah. Excited to have him back in the game. Well, if you're selling stuff on the Internet, you got to head over numerohq.com sales tax on autopilot, spend less than five minutes per month on sales tax compliance. In other news, TSA has begun letting travelers keep their shoes on. You had a good take on this, Jordy.
Yeah. What is the ROI on PreCheck now?
I think it's very low.
The shoes line. Yeah, there was the shoe line and then there was the note.
It was you didn't have to take off your shoes, but you waited in a longer line like 80% of the time because so many people had pre checked that the line would often be.
It felt like that's only been in the last like year and a half, two years.
But now it's like you pay, you wait in a longer line and you still take off your. You still leave your shoes on just like every other line. So. But it's crazy. It's been 18 years. I remember when they rolled this out, it was not because of 911 actually, it was because post 911 there was a shoe bomber who tried guy who put explosives in his shoes and tried to light it on a plane. And then everyone got freaked out about shoes. But it was like this freak thing. It didn't actually result. He wasn't successful, but it was still very like visceral and scary and crazy. It was like, oh, if there could be a bomb in your shoe, you have to check the shoes. But throwing the shoes through the X ray doesn't really detect I think like the bomb material. I think that's why they swipe now. So the residue is actually more important. But they're replacing this with a bunch of CT scanners, AI threat detection software. There's a whole bunch of stuff that can screen foot well with footwear without unclasping it. So it's a phase policy. It's already live at airports from LAX to Baltimore. Should reach all standard lines nationwide. Nationwide by late July.
I'm excited to see TSAs go to market on this. I want to see their efficiency. It's great to see they're already in testing in some markets.
Here is the value prop for Precheck. Precheck is now pushing a more like Clear experience where it's touchless and you don't need your ID or your boarding pass, so you just scan your biometrics, either your fingerprint or your eyes, Something like that.
Interesting.
I'm sure the schizophrenics will love that.
Yeah. The CEO of Clear is on invest like the best as of, I think this morning. And cool story. I guess she bought it out of bankruptcy for like $6 million. I haven't had a chance to listen to the full episode yet and turned it into a monster.
That's very cool. Well, let me tell you about Adeo Customer Relationship Magic. Adeo is the AI native CRM that builds scales and grows your company to the next level. You can get started for free. In other news, Unilever partnered with Nvidia to generate a ton of AI ads and it actually worked pretty well. They got 3.5 billion social impressions and they did some sort of collab with Crumbl Cookies and then they generated a ton of images.
Crumble flavored soap.
I don't know, it's scented. But basically they were able to send every influencer that they work with a ton of different variations. Just say like, here's. Here's a thousand assets and a thousand taglines. Pick whatever you want. None of these overlap with anyone else.
Interesting.
And so, and so for someone like Unilever, that's just like trying to just like blanket TikTok with stuff.
It like work really, really hard. I had a funny incident with a portfolio company once where I was. I said they were doing a launch or fundraise announcement. I was like, try drafting some text for your investors so they know how to position the product and things like that. Right. Because oftentimes one of the things that will just hold an investor back from sharing is not knowing just like the work to draft the post makes sense. And they made like 20 or 30 different lines, but they gave them to everybody. So everyone used to use the same line so bad.
Rookie mistakes. Yeah, rookie mistake.
Don't do it.
And in other news, SWIX has a story here that the new version of Chrome, Google Chrome, they're on version 138. You gotta ring the gong for the Chrome team. They are now shipping Gemini Nano for every user. So you'll have a local LLM in the web browser that in theory web apps will be able to interact with. So you'll be able to go to a web app and if they're in Chrome, they will assume that there's an LLM available, which is very interesting.
Have you noticed AI mode on Google that's rolling out?
Yeah, I haven't used it in the US but it's going to be the default in India. So Sundar announced that today. Basically huge. I think that's because the, the, the ARPU in India is lower and so that they can risk, yeah, they can risk losing a little bit of AdSense revenue while they adjust and figure out how they monetize the AI search, how.
To put some ads in that bad boy.
They're going to put a lot of ads.
They already are.
But I'm sure that the, I'm sure that the monetization rate is lower in AI search mode, in AI mode. And so they need to roll it out to a really big audience. India has a billion people, right. Tons of them use Google.
It's an interesting ad unit. I don't think this will be something that's permanent, but you can imagine in a free LLM it takes a little bit longer to inference the query and you could imagine that's an ad. Just put an ad basically.
While you're waiting.
While you're waiting.
That's not a bad idea. Yeah.
And then it's like, hey, if you want to, if you want it faster or instantly, you can, you can pay effectively.
But well, speaking of AI fin, AI is the number one AI agent for customer service. Number one in performance benchmarks, number one in competitive bake offs, number one ranking on G2. You can start a free trip.
We said this before, we'll say it again. Don't get, don't even try to get in a bake off. Don't do it. It's a disaster for you.
Well, our first guest is here. We've been keeping him waiting because of our technical difficulties, but we have Alex Roy, someone that I'm very excited to talk to. Welcome to the stream. How are you doing?
What's going on? Welcome.
Thanks for having me. Huge fan of the show. You brought me howls of delight when you booked SOHAN Last week, 24 hours after the story broke. I'm like, you guys are next gen media. It's fantastic.
That's amazing. That's amazing. For those who don't know you, can you introduce yourself a little bit of your history? But also I want to talk about what you're up to now and then I'm sure we'll have a bunch of questions.
I'm excited for this intro too because we had another friend of the show last week who said I had a non traditional background for venture. And then he goes on to say that he was like a product manager.
Product manager at Big Tech.
Which is fairly traditional, certainly more traditional.
But I feel like you did have a non traditional background. So take us through it.
20 years ago I was really into street racing and I was obsessed with the history of the Cannonball Run, this illegal race cross country from the 70s and so but as a startup guy, I decided if I was going to break the record, I had to approach like a startup. So I wrote like a business plan as if it was a startup. The competitive landscape was 18,700 law enforcement agencies and the terrain and the road conditions, the SWOT analysis, but you're actually worried about squat. So ended up using everything off the shelf one could use in 05 for mapping. So you know, Google Maps didn't exist. Google Earth had just gone public. So I was using Google Earth and its back end tools to figure out, you know, how to plan the drive. Drove cross country a couple times in a practice, had a thermal camera I salvaged from, from an old Cadillac and you know, laser jammers, radar detectors, everything one could have. And then created a map of the police locations, dropped it into a Garmin and then inserted into Google Earth to plan the drive. Because in Google Earth it's always daytime, but in the practice run it's two thirds at night. And so plotted latitude, longitude of every potential police trap and spotter plane, triple redundant communications, the whole thing broke the record. Wrote a book about it and turns out, and I didn't know this, the DARPA challenges were happening at that time. And so I got when as soon as Wired magazine did a story about our Cannonball Drive, I got outreach from all these DARPA teams and it was all the same thing as oh Roy, you're using radar, lidar jamming, mapping everything we use. You're just selfish and only not trying to help anyone but yourself. And we're trying to make roads safer. And so I became out of that an angel investor in like autonomy and mapping and then watch a podcast about AVs and EVs because that's, that's the future. And eventually I got recruited to join Argo AI, which was Ford and VW's self driving car unit where I was an executive for four years. And so little did I know that.
Everything I was doing to try to.
Drive cross country as fast as possible was directly relevant to investing in the.
Future of road safety.
Makes no sense.
Yeah.
That's amazing.
Yeah. For those who might not be super familiar with the Cannonball Run, can you explain Red Ball Garage, Portofino, the history and kind of where it went now and kind of the current status of the. I mean, it's still illegal, but people are doing it. There's these, there's these like delays. I see Vinwicke videos like, oh, yeah, a year ago I did this run. Like, what's the current culture and like status of the Cannonball?
So, you know, the unknown, forgotten history of the CannonBall is that 100 years ago, car companies didn't how to market, you know, internal combustion cars. So they hired this guy named Irwin Baker, who was a motorcycle racer to drive New York to California as fast as possible. No gas stations, no highways. And every few months a car would come out with a slightly larger gas tank or slightly better fuel economy, and he would go again. And so he went from the record. His first record, I think was weeks and eventually cut it down to.
You.
Know, something, I don't even remember this. He did 141 records cross country. But that's how you marketed internal combustion back then, to prove that it was more efficient than trains and cleaner than horses. And as the interstate highway system was built out and, you know, speed limits arrived, the Campbell culture just evaporated it. But what is largely unknown is that it was Coca Cola that underwrote the creation of local and eventually regional gas stations and which unlocked iterations in better speeds for Baker. And so if you go forward 100 years, you could kind of look at electric cars today, as more Tesla superchargers were built, it became. It unlocked electric Cannonball records. And in the future we'll have other records. Today there are maybe 100 or so people in a, in a Facebook group, the Cannonballers Group, myself, Ed Bolian from Vinwiki, who broke my record, Arnie Thoman, who said it during COVID And you know, people go on solo runs and occasionally there's a multi car run. But the modern history is almost entirely based on the legacy of Brock Yates. He was a very libertarian editor for Car and driver. And from 71 to 79 he went, he operated a full on illegal multi car race. When it got covered in time magazine in 75 or 76, it got a lot of attention. And the Last 1 in 79 then led to the movie and that was the end of Cannonballing in an organized fashion. It lives on today in the Cannonballers group.
Yeah. What do you think's next for Cannonball? Do you think people are going to try and get creative with the EV challenges? Because that feels like almost an entirely new dynamic. But then also, if you could explain kind of the, the idea of like a Covid run and how that might make it kind of impossible to ever reset the record. I don't know what your take is on that, but just giving a little bit of that history because it is fascinating.
So, you know, the traditionalist group is, you know, multiple cars, internal combustion, as fast as you can. And I did it in 31 hours in, in 2006. And then I was the. Our team was the first team to break it like 25 years. And then eight years later, Ed Boldian did it in 2850. And then Arnie Toman arrived during COVID with his team and did it in like 2530 ish.
Yeah.
So, you know, it is since the COVID record.
What. What were they driving?
Arnie was in an Audi RS6 disguised as a Ford Taurus police car. Ed Bolling was in a Mercedes CL55. I think I was in a BMW M5 disguised as a storm chaser Tornado research vehicle.
Yeah, it's a lot of S63s, E63s, that type of car.
I mean, there aren't many American cars that could do it. Today would be like a Cadillac, you know, CT5 Black Wing.
Oh, that's what John drives.
I have one great, great car. Fantastic. Perfect.
If they could just lift the speed restriction on the street.
Are you in New York right now?
I'm in Scots. I'm in Scottsdale.
Oh, okay. Okay. Well, I'll see you in like two hours then. You could.
So the.
So, you know, I looked when I looked at the history of Cannibal as an investor, I was studying the history of the Cannonball because understanding how gas station networks are financed and built and, and monetized and eventually made profitable. I was like looking at EV charging network business, I'm like, well, how does that. Have they Learned anything from 100 years ago? And the answer is no, they didn't. And so, so the lack of infrastructure and the lack of reliable infrastructure with uptime was a huge Achilles heel for the EV industry. And only Tesla figured this out early. And so looking at Cannonball Baker, I'm like, well, why couldn't I be the CannonBall Baker of EVs and AVS eventually? So in 2014, 15 started doing electric records. Tesla was the only vehicle at the time capable of setting these records because of the charging network. And recently I began looking at, you know, lucids and Porsche Taycans, and those are the only vehicles that could break the Tesla record. And recently a Porsche Taycan second generation became the first car to break the Tesla electric record in 39 something.
And what goes into a great EV for breaking a Cannonball record is it just total capacity of the battery or ability to sustain high speeds for a long time?
I mean, battery swapping.
Battery swapping seems like logical, but none of the cars have that stock, so I don't know how you do that. Like a fuel cell is common in a, in a gas car, but is there an equivalent in. Yeah.
I'm also curious, what other tech would you, would you use today? Would you use a network of drones? I can imagine.
Oh, yeah, that's interesting.
You know, potentially have thousands of drones spread out.
Well, this is where cannonball history and record breaking overlaps with investing in defense. Because if you understand the history of like radar detection and laser speed measurement and laser jamming, it's all analogous and they are literally parallel tech in defense. And so the electric records up until recently, because the power density of the batteries is less than, you know, a fuel tank full of gas, your speeds were basically limited to the optimal cruising speed of an EV, which is in the 70s. And, and then you have the issue of the charging networks. So there's nothing you can really do to a pure EV today. There's not much you can do to increase their efficiency. And you certainly can't increase your speed without eating into the, into your battery state of charge. But everything comes down to the charging networks. So when Tesla, you know, when their network was 150 kilowatt across the your maximum, your minimum time cross country with X, when they went to 250, you cut a little time off that. What's so fascinating about the recent Porsche Taycan record is the Taycan's EPA battery. EPA range is less than a Model S long range, which is what I drive. However, the Electrify America network, which was a terrible charge network for years, went through upgrades recently. And if you buy the cheapest Porsche Taycan with a big battery option in its lightest weight configuration, rear wheel drive only, and you can charge at Electrify America at speeds that are faster than any Tesla, the cliff in the Porsche battery will let you go, I think max charge rate slightly beyond 50%, whereas a Tesla, the first cliff is at 20%. So that's how the Taycan did it. Less battery range, faster charging, deeper into the battery. That's the trick. A lucid today is the longest range EV you can buy, but it does not. Its battery chemistry is not designed for what the Porsche is. And so for the time being, Porsche will hold the record. I think the next generation Tesla or Lucid could take it.
Talk about the state of EV charging networks today. Coverage, you know, are Networks holding back EVs at all? We live in California, there seems to be plenty of EV chargers around. Doesn't seem to be any type of bottleneck. But I'm curious nationally.
Well, I think that the non Tesla charging network, regardless of its ubiquity, has a lot of issues with handshaking and seamless payments and just hook up to your car. Tesla figured this out, you know, out of the gate. Like it has to be frictionless or people are not going to want to do it. And so the American EV charge industry still has a lot of work to do. And, and even if they do the non Tesla's on the road, I mean we have Rivian Lucid and the Porsche Taycan, those are all great, great models. But other than that there aren't, there isn't like an amazing EV product from any other American oem. I mean some of the Honda, some of the Hyundai's are good and so you need both half the equation. It's where the Chinese are so strong because the vehicle hardware and software and infrastructure is all very good, dangerously good.
Can you talk about. It's kind of surprising to meet someone who's taken, you know, traditional combustion engine car so seriously to the limit and then be interested in EVs? I feel like a lot of the traditional car guys are like I would never.
Purists.
So what excites you about EVs broadly. And then what do you think about the current state, the change to the EV incentives that are going on right now? How does the outlook look with Elon? The Tesla sales are kind of dropping, but that's for a variety of reasons from everything from political to exports to China to their late refresh cycle. There's so many different things going on. So what excites you about EVs and where do you see it going?
Well, I'm, you know, I'm, I think I'm very American in that, like I'm totally agnostic on technology. Like what is the coolest, best, newest thing? Are we making it? Let's try it. And if someone else is making it, we should make it like that's just, that's it. This country became great by ex. Innovating and exporting. So I don't care how something works as long as it's really great product.
Yeah.
So you know, the EV tax credit expiring, I think It's September.
Yep, September 30th.
It's gonna, it's not, it's not awesome if you're an American car company selling EVs and it's gonna certainly slow rate of adoption because prices will go up. But the real question is, you know, how do we depoliticize technologies from our internal American discourse such that the United States long term can compete in export market markets? I don't really care like about the political opinions of who develops a certain technology because we need to create jobs here and be exporting stuff outside the US and the Chinese are going to kill us long term in the EV market if we don't get on board. So the EV tax credit expiring sucks. But it's a minor piece in the holistic system that we need to be debating and executing on. And I recently met a woman, Elaine Dasinski from I think the foundation for American Democracy. She said we need a Pentagon for industrial policy. There doesn't need to be, you know, verticalized, monolithic, you know, command economy, but it needs to be a little less chaotic and more focused than we've been. And I think that's really essential. Really, really essential.
How do we solve the depreciation problem with EVs? I think it's for, for the consumer frustration of like, let's say you buy a Model Y, drive it for nine months and suddenly it's worth half of, of what you paid.
Well, I don't know if there is a solution to that. Probably the best solution is we need to be investing in EVs and supply chain such that we can bring costs down, such that there isn't such a huge delta between the new cost and the depreciation. And all cars appreciate to some extent EVs more. So I mean the great news for a lot of Americans is that used EVs are really cheap. A lot of them are great. I mean used Teslas are fantastic. I mean assuming you, well don't buy an old Model S, but everything else is fantastic.
Yeah. Where are you seeing opportunities to invest in or alongside the EV trend? There's so many different parts in the supply chain. There's so many different adjacent products that an EV buyer, if you're not going and funding like the next Rivian or the Next lucid. Where do you see opportunities for founders?
Well, you know, I, you know automotive is a very small part of my fund. New industry venture capital. I mean a lot of the, the biggest bets mostly had been taken and they're very expensive. You know the, I think most of the opportunities are in, in manufacturing and in energy. However, there is, I mean we only do deep tech, but outside deep tech there are a lot of opportunities adjacent to transportation and EVs and especially aviation. When people talk about job loss, you know, due to driverless vehicles, I think they are missing the big long term picture.
The.
There is a universe of startups that don't exist yet, not even barely in conversation over dinner for providing services for passengers in autonomous vehicles. You know, my mother, you know, she's older and she has some issues with shopping. She would love to have an attendant or someone just to go with her on trips to go shopping. Like where, where is the Uber for elder elder care that would, it's a startup that's adjacent to Uber and Waymo that figures out how to connect them and has a supply of people who can do that. Most people who are driving Ubers probably shouldn't be just from a safety standpoint, but they'd probably be very good at helping people and there's a much better long term opportunity in an employment opportunity for that role than there was for driving cars.
Yeah.
Now I'd also add that the, the, the fear and concern and the politicization of driverless tech is, doesn't track with the history of automation and other, other verticals. You know, we used to have 100% penetration of elevator operators in this country and we went from 100% in 1946 to 1% in like about 1970. But the total number of people employed by the elevator industry is bigger today than it was ever interesting because elevators will always be a business. But there's other roles around the elevator industry that's true of transportation. So I just don't buy any of the panic and doom that companies new tech at all.
What about go to markets? The race between Waymo and Tesla? Obviously very different approaches in terms of fleet development, but I'm curious how you're seeing that race.
So the race there, I mean, I used to like to say there isn't a race in AV because there'll be multiple winners. There's no one can name an elevator company other than Otis, but there are the big five elevator companies on Earth. They're all big. So in av, Waymo and Tesla are both going to have products and the people who claim one will win those destroyed. It's absurd. But the real race is, you know, can Waymo develop a business model outside of robo taxis before Tesla can generate a business at scale in robotaxis? Because Waymo's not in the personal ownership business and Tesla is barely in robo taxi at all. So Waymo needs to get into the licensing business and probably miniaturize their hardware such that it can conform to and become invisible on passenger cars. If Ford or anyone else had waymotech inside, if they could do that now, I mean, Tesla's in a tight spot, but Tesla's got the opposite problem. They've got to deploy a, I would say perceptibly safe system because no one knows safety means and get to the robotaxi business at scale with all the friction unique to their business model before Waymo can invade their space. But ultimately you're going to have three, four, five companies competing in AV tech. You know, we're, we, we just seen growing pains in both companies executing on different business models and that's going to play out. They'll both have their place.
Last question for me at least I'm interested in you compared the Cannonball run to building a startup business plan. What do you think about the actual activity of racing in a cannonball? That idea of risk taking of the stress on your body and founding a company. Have you found any analogies between the two?
Well, you know, the can't when you actually years of planning going to a Cannonball drive and then it's over in, you know, sub 30 hours. I don't know of anything in except maybe like an emergency product deployment like that, as stressful, physically stressful, but every other component of it is similar. And in that way the Cannonball is more similar to say a deployment of a prototype in a defense scenario for the first time. And you have, you got to deploy. It may not be ready.
You don't know.
You spend years developing it. It's go time. That's the closest analog. And I spent a lot of time with law enforcement and defense folks today and I'm constantly amazed that they find what I did so interesting because to me it's nowhere near as interesting as what they do. But then when you start talking about laser jamming and communications and mapping is spoofing, you know, aircraft transponders to mask your location and then I'm like, yes, conceptually it is all the same thing. I would also remark that there's something profoundly American about the Cannonball Which I think points to like the spirit of innovation. It is unique to the United States, but that drives American dynamism, which is that whatever the rules are, how far beyond them can we go successfully without hurting anyone? Let's go, let's find out. And that is the opposite mindset of a lot of founders from other countries and sadly Europe.
Yep. No, that makes no sense.
Makes sense. Last question from me. The CEO of Ford was recently making some comments. One basically talking about white collar work broadly, which felt more like he was just parroting, you know, maybe Dario or something like that. But. But on the other side, he's been very outspoken about how innovative Chinese EV manufacturers have been at multiple different kind of touch points. Do you think that American car manufacturers outside of Tesla have the will to start to innovate at the level of some of these Chinese EVs?
Well, I know Farley, to his credit, after I set my last electric cannonball record in a Tesla, he invited me to dinner. He sat next to me and grilled me on everything about Tesla Model 3. So he's good, he's on it. And the Mach E reflects some of those lessons.
Yeah, the Mach E, my buddy, my neighbor has a Mach E rally and it's a great wild car.
Good look. The Chinese car market has been there for anyone to see who wanted to go over there and do competitive intelligence for years. And the same thing's happening now has happened in the 70s when the Japanese arrived in the United States. People weren't paying attention because like, oh, they can't, you know, these foreign, they can't innovate, they don't know what they're doing X, Y and Z. Well, guess what? They did. And the cars are great. And friends of mine, Kevin Williamson's a journalist who went to China and has driven everything. He's like, the cars are better than anything, soldiers, even Tesla, because they're built better than Teslas. So we've got a problem. The only companies that are really have an opportunity in the near to midterm are, you know, Lucid and Rivian and Tesla from the United States. The other, the legacy OEMs have individual models that are, okay, Mach E, you know, okay, I don't know what's coming, what GMs got, but the real question is what could be exported to any other country to compete with a Chinese car in that market? And the answer right now is other than Tesla, basically nothing. Lucid stands alone. It's a luxury model, it's really good. And Rivians are very cool. Because the products are great. But we are in a tight spot and you know, we need to. Whatever your politics are, the United States needs to have national champions that are exporting. We shouldn't be talking about Tesla versus Waymo. We should be talking about Tesla and Waymo and exporting the hell of anything we can that employs Americans here that are great products that then support the American soft power narrative for the rest of the century.
Yeah, that's great. Great answer. I completely agree. Thanks again for stopping by. We got to do this again soon. This is super fun.
Anytime, guys.
Love you. Thanks so much for hopping.
Cheers, Alex.
Well, if you have an opinion about a public car company, get on public.com investing for those who take it seriously. They got multi asset investing, industry leading yields.
Remember the day when figure was allegedly getting value north of Ford, Ford Motors?
Well, yeah. If you want to rotate into Ford, you can do it on public. If you want to rotate out of Ford, you can also do that on public. This isn't financial advice here on the show. We just never knew the facts. And the next fact is that Ryan Peterson from Flexport is joining the stream in any minute now. Ryan, how you doing?
You just threw me in here.
I'm eating food.
I didn't realize it was on. Yeah, it's like a surprise, surprise. What are you having for lunch?
We've been under attack today. So the stream's been late.
It's like Chinese food, but it's a fancy salad.
Ooh, nice.
Is that Blue Barn? Yes. Oh, Blue Barn's fantastic. Great spot. Anyway, what is new in your world? Can you give us the update? We lived through the tariffs. We checked in with you a few times. Give us the temperature check on where we are with tariffs. Global trade broadly in your business.
There's no strategy with the tariffs. They're purely to torture Ryan Peterson.
It's all caps.
We're learning how to read and write in all caps all over again.
Someone said the art of letter writing was dead and then Trump just proved us all wrong. You know, he's writing letters to national to countries with really amazing capitalization standards. But he's dictating.
Just call you Mr. Flexport is.
The best part is that he's dictating because he doesn't type himself. So he's telling them no, like capitalize tariff and capitalized country and all caps.
For trade certainly has a certain aesthetic to it.
Yeah. Whether there's a strategy to it or not is a really good open question because we got new tariffs that are above on Our allies Japan and Japan and 25% tariff. I don't know. I couldn't tell you what the plan is there, but it's everyday new things, basically.
Yeah. What about just the general mood of customers, the flow of shipments? I mean, initially it was like the boats are sitting at the port not moving because people are just terrified. We heard about iPhones going in the bellies of 747s on rush flights to get out of the the airport like minutes before the tariffs went into effect. Has there been any more clarity? Are we in some sort of return to normalcy where at least CEOs can feel like they can plan ahead for what's coming? Or is it just as much uncertainty?
Today's the day. So the terrorists are. The 90 day reciprocal tariffs came out and they were paused for 90 days. 90 days ago today. So they are going unless something is announced, which he announced six or eight countries yesterday with their new tariff rate. But everybody else that's happening today, they'll go back to the 90 day to the original rates from April 2nd. That freaked everybody out tonight at midnight. Now then I believe he tweeted earlier today or Truth Social that It'll be on August 1st that those new duties are due. So there probably will be a bit of a race here for the next three weeks to try to get stuff in at the 10% rate instead of the higher rates. That's coming back. The thing that freaked everybody out 90 days ago was more the China rate, which was 145% and that's come back down to 30 and people are kind of like, okay with it. I don't know. We'll take the temperature here and we'll see what happens with freight bookings, but things seem kind of normal with the customer base. But we haven't got enough data yet. I mean, it's brand new information. The new rates that came out yesterday and that are going to go into effect in the next just a couple.
Weeks, just as a little extra data. Right now the S and P and NASDAQ are basically flat, but copper prices have hit all time high on this planned 50% tariff. I am interested in China. I remember when you were explaining the nature of the tariff, there were a few different elements. You were talking about the, the energy prices into China basically being passed through. I believe that was a tariff on oil that then was getting passed through to the cost of goods in some way. How are you thinking about the. You said 35%. Is that all in or is that just a piece of the Overall tariff picture, if you're considering doing business in.
China going forward, it'll depend on your product. Tariffs are driven by three things. One is the country of origin. And that is something that actually Trump is changing the rules around in real time with this idea of transshipment that he said goods that are trans shipped through Vietnam will pay a higher duty. But very odd. I mean, those of us in the industry think of transshipments has no effect on duty rate whatsoever because it's still the original country. If you ship from goods from China to Vietnam and then from Vietnam to the US Unless you did what's called substantial transformation, you change them into a product of Vietnam, they still pay the Chinese duty rate. Whereas he's kind of changing those rules in real time and have not yet published something in the code of federal regulations that we can actually understand what he means. So remind everybody that true social is not how laws get and regulations get published. We're waiting to see what really comes out.
Yeah.
So, okay, so you have the country of origin, you have the classification, which is, okay, what type of goods are they that drives, and do they have the components of steel and aluminum that sets a different duty rate? Or what is this thing actually classified as? And then the third one is valuation. How are they valued? And those are the three kind of key drivers of what your duty rate is. And right now, there's just a lot of uncertainty that we're waiting to see what this transshipment thing means. The world will find ways to minimize their duty rate. Hopefully illegal ways. Don't make sure you work with an attorney and don't just go do this. But it'll be very interesting to see. Can you actually save duty by shipping from China to Vietnam somehow in this world? I don't know what the definition of trans shipment is going to be. The other interesting thing here is that from a trade perspective, logistics and customs companies are kind of happy about this stuff because it adds complexity. You need partnership from experts more, and you're doubling the amount of trade because there's going to be, if there's incentive now to move goods from one country to another, transform them, and ship them onward. That's two shipments, one from China to Vietnam and one from Vietnam to the US or one of the sad ones that came about here is we have a longtime customer that makes bicycles in the United States, and they import components and assemble them in the US and sell them. Well, bicycles got an exemption. They're duty free, but the components did not.
Oh, interesting.
So these guys are Actually better off shipping their manufacturing offshore now, and that's what is going to happen.
That's very odd. Have there been. Yeah, you feel like. It feels like there's a little bit of this, like, game theoretic dynamic where Trump puts these blanket tariffs all over the place, but there's potentially an incentive for some country to just jump headfirst and say, hey, we're playing ball. Give us the lowest rate possible because we want to grab as much manufacturing capacity over here and be as friendly as possible. Have there been any, like. I mean, I know that the situation with Japan has been more difficult than expected. Have there been any countries that have popped out as like, new favorite trading partners of American companies, companies in your feeling?
Well, Japan, companies wise, is different from the government, probably.
Sure, yeah, yeah, the government.
It looks like the UK so far.
And.
But the UK's got a 10 duty rate and some nice exemptions that they got on auto, which I guess is like Land Rovers and Bentleys, maybe Aston Martin. Yeah, they don't make a lot of those. But Jaguar, is that.
Is that how they.
I don't think they've been making many cars.
No, no, Jaguar. Jaguar actually took a year off. They skipped a year producing cars.
They were taking a year off business.
We're just gonna sell down our inventory.
Oh, okay. It wasn't just because the commercials were that bad.
No, no. Then they launched the rebrand after they had announced that, hey, we're not producing any new cars, which was like, your.
Job belong in a Jaguar commercial.
Look at.
You belong in the north of England.
I didn't understand that. I didn't understand the hate on the actual silhouette of the new Jaguar.
It was beautiful.
I thought it was cool.
It was very cool. I mean, it was one of those prototype cars that didn't have, like, mirrors or anything. So they clearly needed a lot to do to make it a real one. Yeah, but, yeah, I mean, a lot of.
But. So come back to your question. The UK is interesting because they actually have a trade. We have a trade surplus with the UK and they still got a 10% duty rate. So that told everyone. Oh, the best you could possibly do is 10%, even if you have a trade. Even if we maintain a trade surplus. Vietnam is probably the best example of someone that tried to play ball and they said, hey, we're going to do everything duty free on American products being imported. And they ended up with, what was it, 20%. And then this weird transshipment thing that I already got to that we don't understand. Yet. So that's probably the thing that Trump would like to get, is that trans shipment. What he would like to have is that these countries put duties on China and Chinese goods coming into Vietnam. They want them to get hit with high duties so that you're not effectively laundering, washing Chinese goods through that country, doing some minimal transformation, and then shipping them to the US they want to make it so, like it's a, you know, like create this coalition of the willing against China. That's my, that's my read on it.
Yeah. Yeah.
It's kind of hard for them.
Yeah. There's kind of two narratives. One's like Trump likes tariffs. He wants to throw tariffs all over the place, make us a high tariff country. On the flip side, there's like, this is about a trade war with a rising near peer competitor. And the entire narrative is really just US versus China and everything else is in service of that. Is that a reasonable narrative that you think people should.
It's one of those two, probably, yeah. Maybe some balance. And I'm not sure they know. And there might be different factions within the US Government that want different ones of those. And it's, it's pretty hard to get a read on it. But from a company perspective, they're just looking for whoever makes the cheapest stuff that's high quality that they could sell. I mean, I don't think these countries, there's not that many companies. They might on the surface claim to be patriotic or have some views or something, but that's not what companies are meant to do. They're meant to make money. And so they're going to route, they're going to wait till the rules get set and then they're going to water flows downstream, they're going to downhill, they're.
Going to go figure out where the.
Best place to make stuff is.
Yeah, we saw this with Nvidia, where there were chip bands and Nvidia, hey, the speed limit's 65. I'm going 64 miles an hour. And so I'm going to make a specific chip that is as good as possible while not breaking that regulation. That's not the spirit of the law. And it's like, yeah, I'm a $3 trillion company. My job's not to adhere to the spirit of the law. I got to adhere to the letter of the law. I got to make as much money as possible.
And so that's what, that is what.
Companies are supposed to do.
So government needs to set the rules and then we all figure out how to make money, given those rules. The problem right now is that the rules change so you can't make any decisions in the next. I advise people wait, yeah, least tomorrow. I mean, we'll see what, what new trade deals drop tonight. But I think every few weeks now.
If any, if history is a guide, as soon as you leave this zoom, there will be news.
Yep.
Someone check Twitter right now.
What, what, what? How cooked are Temu and Shein right now? There had been some reporting last week around people, us shoppers ditching them because of the loophole. I didn't know how real that was.
I haven't seen the latest data. I think they're both back live. You can buy stuff on their apps again. So I think they're set up in a way that lets them transact and just pay the customs duties and I think they're fine. I don't know how the consumer is responding to it all. They had stopped selling for a couple of weeks, so that probably hurts you pretty significantly.
But if it's a $30 couch and then you're paying $90 or whatever, it's.
Like, well, it's only 30% on their wholesale price, not on the price that you pay.
Oh, right, right, right.
So I think they may be fine. It's all. But I have heard that apparel companies who most directly competes with Shein, at least actually sales are up quite a bit despite the tariffs and having to raise prices. And that's probably in part because Shein's volumes have come down and the consumer is shifting their spend onto other brands. So we've seen that a little bit in our customer data in our fulfillment business where customers are exceeding their forecasts. The world is not all doom and gloom from an E comm perspective, which it's a big departure from 90 days ago where I was screaming everyone's going to go bankrupt. And companies were screaming that. To me. I think that relaxing the Chinese tariffs made a big difference. But again, we're going to figure out what happens the next 24 hours with these new duties and how people respond to that over the next few weeks.
It's going to be exciting. What about some of the smaller players in the market? Just the broad drop shipping economy. Jordi and I were talking about this this morning, how it feels like. And Sean Frank from Ridge was talking about how like the default road to entrepreneurship a few years ago was like, spin up a Shopify store, white label some stuff, get a business up and running. And then it shifted to courses and influencer stuff and AI and info Products, info, products. And now it's wop and clulee and these other things. They were kind of talking about the modern first time entrepreneur track. Are these, you know, small, lean, dropshipping operations still going? Do you think, do you, do you ever run into any of these people? Do you ever see them kind of like you used to hear about, like, oh, someone just found this really interesting niche on, on ads and they scaled this like very simple business to like tens of millions of dollars. Are you still seeing that or do you think that's kind of like.
I think you mostly see it in China and Vietnam, but selling into the U.S. yeah.
Okay. They're entrepreneurs.
Their entrepreneurs have figured out how to do marketing and branding. And the more that that was something I predicted would happen 20 years ago, that it would take, but that it would take a generation because they have to speak English. They have to know enough about, they have to have enough humility to go, I speak English, but not that well. I should probably hire an American to do the writing instead of like, oh, I'll do the writing myself and have it be English, but terrible.
On Sunday, I wanted to get a new hose. Not a spigot, but one of those things that you can like control the nozzle or whatever. So I go on Amazon, I search hose nozzle and I went five pages without finding a legacy US brand that would sell me one. And they're all these names that are basically some five, six letter combination of English letters that doesn't even read.
Well, Amazon's a real problem in that regard.
Yeah. And I would, would happily like, I want to check a box that's just only buy from brands that have been around for 50 years. And I would have, I would just happily like, those are the products that I want to buy. Yeah, I don't want to buy the six dollar product that's going to break in a year and I got to buy another one. I just want to buy the thing that, that, you know, it's not been.
The thing to watch on this front. Those companies don't. 60% of all the sellers on Amazon don't even have a US entity. The goods are actually imported from abroad.
Interesting.
And there's a bipartisan movement in D.C. to block this policy, to block this law, to make it so that you have to have a U.S. entity. You can't just import as a Chinese company or whatever country company. You can't just import stuff into the US without someone here who can be held accountable to following the customs law to whatever other child Safety laws, other stuff that's downstream that can happen. And I think Amazon has business exposure there that people like you are going to. At some point I started gravitating towards Shopify and like the Shop app where you're like, I don't know, there's something higher quality about people trying to build a brand on Shopify today anyways. And then they have political exposure. I mean, what did Trump call them? Treasonous.
I forget what that for passing through the tariffs.
Yeah, you called them a treasonous.
For me as a consumer, it's not like I want my hose nozzle. It's not like I need an American made hose nozzle. It's just like this, this commitment to like this, this relationship that you have when you're buying a product, which is I want to trust that this company put in a real effort to design a product that is going to be effective and durable. And I'm happy to pay four or five times what the, what the entry ticket price is. So Lindyism.
I like it.
Yeah. I mean people are clamoring for this on Pinterest. Like they want to search Pinterest from like any image that's posted on Pinterest before 2023. Like before the AI images came. Because you search for anything and it's just a lot of like sneaky AI stuff stuck in there. If you're looking for something very specific, you're like, I don't necessarily want an AI image and there's no real good filter other than just like, like is it in the Pinterest database with date created before 2023? Like that is the provenance of that image. And it's kind of the same thing like on Amazon being able to filter for did this company exist, you know, 10 years ago?
What was the postmortem on the conflict in the Middle east from a trade standpoint? I remember there was one night, well, it looked like two tankers had been blown up and I was seeing that, that like open source intel on it being like, oh, like I just was feeling bad for you. I didn't send you a message or anything, but I was like, oh, this is the last, last thing that, that the, that global trade needs right now. But we seem to get out of it.
Yeah, it's the Strait of Hormuz is incorrect. Much more important for the wider economy than just for. Certainly not for containerized train it trade. It's pretty small. Like only about 2% of containers go in and out of the interesting Strait of Hormuzzi. And that's basically going to Dubai. Jebel Ali Port, it's called. It's a top 10 port in the world. But it's mostly a transshipment port, meaning cargo comes in there, gets unloaded. Think of it like a packet switcher. The cargo comes on and gets rerouted. It's not that important of a port in the wider scheme of things. The big thing obviously is oil.
It's like 20 something.
5% of the oil in the world flows out of there. It's fascinating. If you look at the oil prices, look at the price of diesel, there's no sign that anything ever happened in the Middle east at all. You don't even see a blip upward.
Wow.
And when Iran threatened to close the Strait of Hormuz, the oil price went down 6% instead of spiking.
Yeah. Weird. Why?
My read is no one thinks that's a credible threat because if they close it, China gets all their oil out of there and China is kind of necessary to keep. If Iran stops, if Iran screws with China, what's left? You going to be against China and the US like this is probably bad strategy. Yeah, I thought that would maybe if. When I saw that they might do that, I was like, this is kind of some 40 chess by Trump. If he gets Iran to shut down the Strait of Hormuz, we don't need it in the United States we got, we produce our own oil.
Basically.
It's an overstatement. But when the oil price goes up, America does better now.
Right.
We produce. We're the number one oil producer in the world. Like that might be good for us as a country. That is a crazy story.
Like the idea of us, I mean I grew up with, with the idea of like energy independence and the idea of America needs oil and we got to go to these wars to get the oil and oil drives everything and now, you know, we're a net exporter.
Yeah, there is. It's not like purely that we produce it. We can just keep it here and be self sufficient. Right. Isn't it that we're producing it? Like it has to be.
There's different grades of oil in our refinery. Capacity is like different. Yeah, sure, there's some of that, but that's nonetheless. A higher price of oil would benefit American companies. Yeah. Like normal people, you and me, our energy bills would go up and it'd be bad. But on that for the country, probably income inequality goes up or something. But high price of oil is good for the number one oil producer in the world.
How do you think retails broadly are thinking about interest rates in the next 12 months? There's certainly pressure on that front. On Powell, it would be retailers have had to get used to high rates. And if you're financing inventory and things like that, it is meaningful cost. I've heard rumors of aggregators shutting down that were heavily debt fueled. Probably can't sustain in this type of environment.
So I assume, like everybody else, I don't think they're. That's. I don't think we're that sophisticated in our business. But cheaper interest rates would definitely help consumers buy more stuff and drive a lot of growth. And I think most of us would realize, like, hey, we kind of took it for granted for too long. We probably should have borrowed more money when it was 2% and bought more stuff and done capital investments. And next time around I think it'll be a massive spending splurge or investment splurge. If they drop interest rates back down, people won't take it for granted.
Totally.
Europe and sf. There's been a bunch of massive trade deals and SF AI engineers going to meta. What's been the mood on the ground? Do you have any advice for the folks picking up nine figure checks?
First of all, I love asking people because. Because if you didn't get an offer, what's wrong with you?
No. And if you're not, if people aren't poached, if there's not like whispers that people from your company are being poached, you're probably not at the right lab or lab.
Oh man. Maybe I should make some posts about this. Getting angry at Zuck for taking my best people.
Yeah, the psyops need to continue.
The rumors are true. He's been personally dialing the entire engineering team.
The advice would be it's.
You should.
Go read the most important business book ever written. It's called the Prince by Machiavelli.
Oh, sure.
And Machiavelli's rule number one is never hire mercenaries. He says it over and over again. It's the whole theme of the book. Mercenaries are useless. They will run away when the time. They'll take your money and then run away when it comes. When it actually comes.
Fight.
And that is really dangerous out here in this world of AI poaching people paying this much money. I'd be very worried if I was. How do you build a culture out of that? It's going to be challenging. They should all go read the book and figure out what are they going.
To do about that. How do you think about the sports analogy. A lot of people are comparing the current trade deals to sports. No one thinks about Ohtani as a mercenary. You know.
Yeah, it's interesting question. Sports, you have a fixed size of your team, so you can only have so many players. So maybe there's something to be said for that. Of like, okay, you've got. Therefore it bids up the price of the.
Also, if you look at, if you look at the. You know, we know nothing about sports, but growing up in the Bay Area, watching the warriors go on this very dominant run, I'm sure at any point many of those players could have of increase their.
Well, not at any point because they have contracts.
Sure.
Years.
And so, you know, the ultimate is.
Tom Brady, because Tom Brady, his secret was marrying a supermodel wife who made so much money that he could work below market price and they could build out a great team. So maybe you need to hire engineers who have rich spouses.
There we go. There we go.
Yeah. Yeah. That's going to be the next. The next bull signal.
Yeah. I don't, I don't do it for the money party.
Okay. You're getting poached. You're going to the top team. The holiday party was popping, but it's.
Yeah, I don't know. I'm fascinated by seeing how this plays out. And it is, you know, at the end of the day, you got to hire the best talent. And Zuck's kind of smart. He's got the war chest. And OpenAI did this before. I mean, they were hiring everybody. I love seeing them whine about it because all my friends have had their engineers poached by OpenAI who are overpaying by 2x. So.
Yeah. Yeah.
A front end engineer was making 350k just to, you know, HTML and CSS.
Yeah, it's kind of.
Yeah. Three hundred and fifty and then, you know, tens of millions of dollars and probably stock options.
Hopefully not for the front end engineers.
This isn't even people.
Hey, hey, no slander for front end.
I mean, it depends on when they got in. Right. If they got in a couple years ago, it could have 10x'd in terms of, you know, net value.
It's, you know, it's one of those. It's like trying to be the toughest fighter or something. At some point, you know, you can beat up everyone on your block, but you eventually run into Mike Tyson and you're going to lose.
So that's exactly what's happened. Yeah. I mean, it's maybe better not to.
Try to fight yeah, it's so interesting.
Because like the last, the last, like big, like, I mean we've had it, we've had a few like big hyper scale, like really fast growing, super hot, like ordained as like, like winner companies like SpaceX, Tesla, Uber. And those companies were not seen as a direct threat to Google or Microsoft or Apple. And so it's not like there was ever, I mean, Google was kind of thinking about Waymo as a counter to Uber, but it was never like our core business might disappear, so we have to spend, you know, unlimited money to protect that. You know what we have, people are.
People are starting to talk about how the competitive dynamic between Meta and OpenAI being that if you're spending three hours a day talking with chat, that's three hours a day that you're not scrolling, putting comments on your friends, Instagram posts or hitting reels or things like that. So there's, you know, we all have 24 hours of the day. We're making more humans. But, you know, yeah, yeah, I want.
To see what ideas they have. Because, because the way to compete with Google was not make a better search engine like 10 years ago, 20 years.
Ago.
A lot of people tried that. You couldn't do it. So I don't know that the way to compete with OpenAI is to make a better chatbot.
I completely agree.
They gotta come up with some new ideas here and I don't know what they are, I don't know what they have in mind.
I want a new thing at the bottom of the Instagram app that's just dopamine and it's just a perpetual loop. You don't have to scroll. Yeah, something there.
The whole app might be that eventually.
Just surfacing content you want to see would be a win.
Yeah. Anyway, this has been fantastic. Ryan, thanks so much for hopping on.
Always good to send us a note when new Truce hit the timeline. We would love your reaction.
I don't think you guys can handle the speed at which this is going to happen.
We might need you back on the show tomorrow, who knows?
Yeah, yeah, I don't think you need that. All right, take care guys.
We'll talk to you.
Great to see you, Ryan.
Cheers, bye. Let me tell you about adquick.com out of home advertising made easy and measurable. Say goodbye to the headaches of out of home advertising. Only Ad Quick combines technology out of home expertise and data to enable efficient, seamless ad buying across the globe. It's Billboard time and we have our next guest in the studio already Jonathan From Air Garage talking about parking. How you doing, Jonathan?
It's time to talk about parking.
Everybody's favorite topic, parking. I'm just glad I finished my lunch before I got let in on like. So I'm a little more prepared than he is.
It's a bit of a random stream today.
Great hearth, great hearth. Is that a tv? That looks like a tv.
Yeah, I wish it was a real hearth. I, you know, it's San Francisco, it's June or it's July now, I guess, but it's pretty much always cold so we've got to try to stay warm however you can.
I think the real hearth comes after the next round.
That's right.
Once there's some major liquidity in the business, maybe then you can upgrade. But give us the news. What's the latest from you?
Yeah, very exciting news today. We just announced that we raised our Series B from and was led by headline growth and participation from some of our existing investors and customers and all that jazz. So very exciting. $23 million series B and hit that gong.
Let's go.
Great contact, great contact, great contact.
Sounded very full from this end.
Yes, it was fantastic. So give us an update on where the business is today. Kind of introduce the business and yourself and then we'll kind of dig into some questions.
Yeah, definitely. So Air Garage for people that haven't heard of us before, Basically we work with real estate owners that own parking lots and parking garages and we help them manage and monetize their parking assets. So you know, if you've ever parked your car in the United States or really just anywhere, you know that it's a pretty terrible experience. There's signs that are confusing. It's like a five paragraph essay on a sign. The machines are broken, they eat your credit card, they eat your ticket, all that sort of stuff. And you know, I think a lot of people when they first experience that or if they've thought about it at all, it's kind of like one of those schlep blindness problems that most people, you experience it but you just kind of ignore it when it happens in your daily life, if you've thought about it, you're kind of like thinking, okay, maybe it's like this because, you know, that's how real estate owners like it. They want to scam me, they want to like rob me blind. A lot of people, you know, they have it out for real estate owners, but it's not that way at all. Real estate owners hate it too. And so, so we work with those real Estate owners. We've built this basically vertically integrated operating system for parking real estate. It combines everything involved in taking it from being a piece of asphalt with stripes on it to a full fledged parking operation that generates income using technology. So it's the technology you would expect in any other industry, but it's just completely new to the parking industry. So we do payments, we do advertising enforcement, we build hardware, we build software. Basically what we're trying to do is bring as much data and visibility to these assets as possible to generate more income for real estate owners, to create a better experience for driver at the same time. And so, you know, the net result is we can increase net operating income by 20 to 50% for real estate owners. And for a real estate owner, that's huge because your asset is valued on and why and you want to make as much as, as much as possible. So that is how we've managed to grow as much as we have. So we've grown 10x since we raised our series A. We're actually cash flow positive as business, which is awesome.
Congratulations.
You know, just trying to get as many locations across the country as possible. So now we have 300 plus locations in 38 states across the United States.
Fantastic. I have two funny stories about parking. One once was I started, I stopped carrying a wallet because I was like, Apple pay is so good, I'm fine, I don't need a wallet. And I got locked in a parking garage that didn't take Apple pay once. And I had to talk my way out of it and hit the security guard and be like, sorry, I literally have no money of any kind of dying. I die, I can't do anything. And then the other time I was parking in a, in a garage, going to a restaurant and, and like, it's very confusing how to pay for this garage because there's no like, there's no like gate. So you just go in and then there's a ticket attendant. And the guy comes up to me in like a jack, in like a, like a, you know, security vest or whatever and says like, here, yeah, you pay me. And I pay him and I go into the restaurant and the restaurant owner is like, yeah, that guy doesn't work there. And it was just like, like an enterprising homeless man, I think. And so, you know, good on him, but not, not the best user experience. So, yeah, I have a ton of questions I want to know. I mean, so hardware, like that seems actually pretty hard to solve. I mean, taking cash, credit cards, Apple pay, like the demands are Pretty wild. Do you. Do you white label stuff? Do you actually develop everything? Do you have manufacturers for this stuff? Is it commoditized? Like, what is that? What is the shape of that side of your business?
Yeah, so the hardware that we do is all basically cameras and sensors. So it's mostly license plate reading cameras that sit at the entrance and exit of the garage, and they track every vehicle that goes in and goes out. That gives us real time occupancy data, gives us enforcement data, but then it also can actually enable these magical experiences. Where I park down near our headquarters in downtown San Francisco, we have a garage right near there. My account, because I'm already registered with their garage, when I drive into that garage, it automatically starts a session for me. When I drive out, it automatically ends my session. I never have to touch my phone. So to your point, you wouldn't have to carry a wallet anymore. And it bills me for exactly the time I use. Right. It's like that's when you think about it, like, that's just how parking should be. Yeah, it's not that way. Right. It's like so old school. And that's because basically real estate owners, they hire a parking operator, which is basically a property manager, to run their property. Those parking operators, they're 50 to 100 years old. They've been doing things the same way since the 1980s. They use cash and machines and attendance and. Yeah, they all use basically parking machines in their locations. And it turns out the parking machines are just kind of the wrong way to run a parking facility. You have those gate arms, people, more often than you would think, people drive through those gate arms. A lot of owners, when they're switching to Air Garage, they're telling us about, you know, 15 times a month, somebody breaks their gate arm, and it's like $4,000 to fix it every single time. And that's on top of you spent 100, $200,000 to buy the gate arm in the first place. And then you just need an attendant to sit there and like babysit it. So one of the things that we do differently at Air Garage is we're in entirely a gateless system, which means you remove that entire huge sort of headache and expense of maintaining and keeping these gate arms in your facility. And so when you're a driver, that's also a better experience. You drive in, you drive out, the license for reading cameras, track your car. We get real time data, you get a better experience, and you have a lot less expense and sort of, you know, maintenance issues. To deal with the gate arm because I think a lot of people can relate to the experience as well. Of like, especially after a big, you know, sporting event or event that you've been to, you're trying to get out of a parking garage, you're sitting there for an hour because people are taking so long to go through that gate one at a time. You get rid of those, you entirely get rid of the bottleneck. And it's a way better driver experience. So our hardware, the way we build hardware, we build it all actually in house, which is very unique for any parking company. Most parking companies, they don't even build their software in house. We build software and hardware in house to give us the best control and visibility. But the critical thing is you don't make the hardware a dependency. Right. If a camera goes down for some reason, people can still drive in, still drive out, they can still pay. We still have a functioning system through the software. So you remove that dependency and it makes it a way better experience. And also you just get a better experience as a driver and as an owner.
Okay, I got to ask two VC coded questions. Yeah, I'm sure you got in the raise, but I'm curious. So we had the CEO of Metropolis on. Yeah, yeah, I'm sure you guys were competing in a, maybe at some point in a more intense way they've taken the more roll up approach in terms of asset heavy. At very asset heavy. They raised 1.7 billion to buy.
And not technically asset heavy because they're not like purchasing the assets themselves, but they got it basically acquire operators. So they just took to your point. They took an inorganic growth strategy which. Yeah, I certainly have lots of thoughts on, but. Finish your question. Sorry.
Yeah, no, and I'm just curious, like I'm sure every investor meeting leading to this B people were like, have you thought about raising a billion and doing this strategy? And you did not. So I'm curious how you evaluate those two approaches.
Yeah, I think that there's a lot of different ways to build a company. And basically the thesis that they're working on is actually the same thesis that there's another company called Reef Parking. I don't know if you've heard of them, but they took the same approach and what they did is they raised about $1 billion from SoftBank in 2018, 2019. That was right as we were raising our seed round. So a lot of people were looking at that saying like, oh, they beat you to it. Sorry, you're out of luck. But the thesis is basically okay, we have this technology, we have this platform, we're going to raise a bunch of money, we're going to acquire this old school operator that has all of these locations and contracts already. And of course we're going to, you know, it's going to look beautiful in a spreadsheet. We're going to replace all of this GNA or this overhead or this cost of goods sold of human labor, or we're going to replace it with this technology and our margins are going to look great and there's going to be a bunch of upside. It's kind of like classic private equity playbook. You know, it's a very like investor obsessed, spreadsheet driven way of building a business. And the critical ingredient that has to be true in that is like we have the technology. And that's what wasn't true really for reef parking. And that's why that whole thing sort of went sideways and they're no longer really a business. I mean, they're working through restructuring. Mubadala has like taken over that business and is trying to sell it, last I heard. And you know, that's the question for Metropolis is like, do they have that technology? And we're taking a very. At Air Garage from day one, we started by taking over one parking lot, working from first principles, figuring out how do you run this parking lot using the technology that we have available to us. We didn't know what a parking company was, we didn't know what a parking operator was. We knew nothing. We were just completely making it up from scratch, which is a huge advantage compared to doing things the way they've always been done. And so we've built the platform as we've gone along to actually truly do what what we say it does. Right. And that's the question for them is as they're going through their portfolio and trying to transition people, are those people actually going to transition to the platform? Does the platform actually work? And the thing is, real estate owners have been burned so many times in the last 20, 30 years by vaporware software and people saying like, oh, we have technology and you real estate people are so old school and you don't get it. And like we're going to change your life and make things, everything great with this technology and it hasn't played out many times. And so it's very hard to earn that trust with real estate owners. And so if you start taking over locations and you don't have that technology and it doesn't work out, then you sort of quickly fall out of favor. And so they're making that bet. And that's the question is, do they have it remains to be seen. I, you know, I have an opinion on that. Of course, I won't comment directly on that. But yeah, it's just a different vision for how to build that end state. And then, you know, do we go and do roll ups in the future as well? Potentially, because we do actually have the technology.
Jordan, second question, second VC coded question. We had Alex Roy on Famous Cannonball Driver, wrote the driver and we were talking a lot about autonomous vehicles. I'm curious how you think AVs will impact the parking industry. Interesting. People have had this sense of, oh, there's so much parking in cities and when we have AVs, we won't need them. That's never fully tracked for me because gotta charge, gotta charge them, you have to maintain them, you have to store them. There's way more demand during rush hour than there is, you know, during, you know, and I don't want to send my av, if I have an AV that I personally own, I don't want to put a bunch of miles on it, driving it way out of the city to then circle back. So it's never fully tracked to me.
That some people just like parking lots more than parks. Like public parks. You can't really do a lot of car spotting in them. That's right around hunting for the Lamborghini.
Can't do cars and coffees in a public park. You got to have a parking lot.
No way.
That would be fun though.
You know what's funny is, is this was a question I got all the time in our series, like our seed rays. And that was like at the peak of the autonomous vehicle hype wave, like the classic hype wave that autonomous vehicles has followed. And this was every VC in the seed round was of course asking about this. And that was actually like six years behind. What did you say?
I'm six years behind.
You're six years behind. Well, and yeah, the funny thing to me was that during this, this most recent fundraise, basically nobody asked us about that. And that's while at the same time I'm in San Francisco riding in Waymos and cruise cars constantly. And so it's like actually real now versus six years ago. It wasn't actually real. You know, I think I have a lot of thoughts on this. One of the, one of the reasons we chose Floodgate actually to lead our seed round at the time was because they saw without us, they independently came to the same conclusion that we did that. You know, basically parking is going to be this really interesting asset class over the next 10 to 15 years at the time, based on what they were seeing, because. And the partner at Floodgate is on the board of Lyft or was on the board of Lyft at the time. And so she was seeing the changes even before autonomous vehicles with just the Uber and Lyft trend over the last 10 years. And so what they saw was somebody's going to need to basically figure out what to do with all this space in our cities. Like 26% of the median urban core buy land area. Median urban core over 500,000 people, 26% by land area is dedicated to parking. So that doesn't even account for garages under a building that is dedicated surface area in the urban core of our cities. Not looking at the suburbs, which is even more parking. That's a massive amount of space in our cities. There's eight parking spaces for every car in the United States. I think the stat is there's enough parking surface area in the United States and eight is the low end. The estimate is actually eight to 11 parking spaces for every car. There's, I think, one stat that I saw that basically there's enough parking space in the United States to cover the entire state of West Virginia with asphalt. And so regardless of how you look at it, there's going to be some repurposing that needs to happen. And in order to do that over time, you first need to like, understand how that space is being used today. And that's the thing is like, that just doesn't exist. Like the parking operators that are running these facilities right now, they're so old school, they're doing things manually with paper and like they don't have any real time visibility. About like how many spaces are actually filled right now. And like, what's our busiest day of the week? These like basic insights that, that, you know, if you were a real estate owner and you called your hotel general manager and said, hey, what's our busiest day this month? And your hotel general manager couldn't tell you the answer to that question. Like you would fire them on the spot. Because that's like basic information you need to run an asset. But that's the state of the parking industry is like literally we took over an asset actually last week. Massive garage, very large asset in a sort of tier 2 market. But you know, the way they were tracking validations is literally a paper spreadsheet in a binder, you know, tracking who gets a Validation for going to the retail shops on site. That's the state of the parking industry. And so our mindset has always been, you know, if someone's going to bring this online, we're going to bring it online and be able to more flexibly use this space in our cities. Maybe that's autonomous vehicles in the future where they're parking in our facilities and they need to be able to pay for their parking in an API driven way that doesn't exist right now. There is no default online way to purchase parking, or no operating system, or no source of system of record for where the space is used in our cities. And then the other thing that's really interesting from our perspective too is we've taken a lot of things in the parking industry that traditionally were fixed cost parts of your business and turned them into variable costs. Because we're a software company, we're a technology company instead of a, basically a staffing company, which is what these traditional companies are. So if like parking revenue starts to trend downward, that actually squeezes all of our competitors in a really helpful way for us where like, our cost of goods sold is entirely marginal instead of being this fixed cost, which then means that we can survive and thrive in an environment where parking supply continues to decrease because parking demand is decreasing. And actually parking supply being constrained is great for our business because that's the time when you actually can charge for parking.
Yeah, I mean, I can see how this becomes a major catalyst for you guys. If I drive into a city center in an autonomous vehicle, it drops me off and then I want my vehicle to park nearby. Imagining the interaction right now between the average city center parking garage and autonomous vehicle. Just not going to happen.
It wouldn't happen.
Yeah. Anyway, so great, great company.
Congratulations again. Thank you so much for stopping progress.
Looking forward to the next 10x.
Yeah, get on it, get on it. We'll talk to you.
Thank you both.
Awesome.
Have a good one.
Cheers.
Bye. Next up we have Jonathan from Grok coming out.
We actually are going to reschedule that one so we can jump into some timeline.
Okay, cool. Yeah. Okay. We're pushing him to one. Great. Cool. Yeah, let's do some timeline. Well, Eliano over at Palantir received an airdrop of Matayina, my favorite energy drink that I drink four of every show. Basically. Delicious. Congratulations.
That is 480 milligrams of caffeine.
It's an adult amount.
It's an adult amount. It's a horse amount.
It's great.
Love to see it.
Shout out to rob for dropping those off at Palantir. Wonderful. And I think he's enjoying them. Let's tell you about Wander.
Find your happy place. Find your happy place.
Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning and 247 concierge service. It's a vacation home, but better and wander.com to go check out.
Everybody thought our ad was AI and I think that's a great compliment to wandering it the the property that we were at and the Wanderer property we were at in Malibu. It looks almost too good to be true, but it's very real and you can go stay there today.
Yeah.
Let's see here. Ahmad had the post that I was referencing earlier. He said, why is meta scared of OpenAI? This graph says ChatGPT is not just a productivity app. People are spending hours on it to talk to AI, write poetry, etc. Meta needs attention to drive ads. If this trend line continues, AI will eat social app time.
So, so this is mobile minutes per daily active users per day and ChatGPT is at 29 minutes. Now the question is like we're not seeing a drop off from any of the other apps. So it might be that ChatGPT is cannibalizing something else in people's life, maybe like talking to other people. But. But it could also be books. It could be, you know, time spent reading the newspaper, researching or doing anything else. It's not exactly showing up in the Instagram data there. That would be really, really. If you like, when you see, you know, chatgpt going up and Instagram going down, then you get really worried. But still, ChatGPT is on a generational run pretty crazy, going from 5 minutes, what, a year ago to 29 minutes today. So 6x growth in a year. People, people love it. I spend a lot of time on it. I'm always doing stuff. It's fun. You know, it's just like super interactive. Just dictate something, get the response, read it out. It's just like so much more efficient way to learn for so many different things.
Totally. I would love to see what, what these other like actual proper AI companion apps like Replica. Yeah, what their usage per dau looks like. Because I would expect that at least for paid users it's significantly higher than that.
Well, let's tell you about getbezel.com, your Bezel Concierge is available now to source you any watch on the planet. Seriously, any watch. Go check them out.
We love missed opportunity to ask Alex Roy about The right watch for the. For the cannonball.
Oh yeah.
He'd probably say an Apple watch today.
Yeah, he's very technology pelled, very modern answer.
We have a trade deal. Trade deal. Summer U. She says today is my first day at Meta Superintelligence Labs. I'll be focusing on alignment and safety, building on my time at Scale Research and Seal. Grateful to keep working with Alexander Wang. No one more committed, clear eyed or mission driven. Excited for what's ahead. Congratulations to Summer. I would imagine this was a maxed out contract.
Yeah.
Good to keep the team together.
It's interesting because Alex was moved over to Meta, but Scale is still an independent entity. But it seems like some folks from Scale are now moving over to Meta and Meta's probably looking through, seeing who makes sense where and the businesses will kind of diverge. It feels like the true like super intelligence research and the hardcore AI research of training new models. That llama V5 I guess is going to be happening over at Meta, but Scale obviously still has a business in data generation and so if you were kind of on the research side at Scale, you're probably going to be moving over at some point. But this is probably the first of a few of announcements. We'll totally.
So she was the VP of Research at Scale AI prior to that was in research at Google DeepMind for almost six years. So an absolute killer.
Yep. Who do we have? We have a few people in the waiting room.
Fun.
I don't know Jacob, to figure that out.
People might just be joining. The link might have leaked.
The link is leaked. This is the thing we use.
We shouldn't leak. We can highlight. So WAP announced. Oh yeah, they had an ad yesterday.
Can we pull this ad up? Is that.
Yeah, let's pull it up.
Let's play.
It's in the M XNT says first off, banger. Second off. This video means as soon as VO3/ insert new AI video generation model, drop WAP immediately assign him to diving headfirst into it, seeing what they can do. This is the biggest bowl signal for new companies having the human resources capital to be able to actively chase staying on the bleeding edge of new tech and innovations the second it drops. Massive, massive, massive. You simply cannot beat a team that is hell bent on being first. So I thought this ad was very cool.
Yeah, we should watch it. It's really crazy because they're really doing a great job of solving all the key problems with generative AI, which is like character consistency. It's not fully photo real and so very clearly, in the prompt they said, make it look like Grand Theft Auto. And then all of a sudden, it's like a Ghibli moment. And I feel like this is kind of the first video that we've seen that's really dominated, like, a new style that I think you'll see a lot more people be making them.
Well, it's perfect too, because WAP can make a GTA themed ad, whereas, like, very few brands could do that.
Yep. But, yeah, I mean, I imagine that this will be as simple as giblify this image pretty soon. Just, hey, take the trailer for F1 and make it GTA style, and it'll look amazing. And people will do that for all sorts of videos. It'll be very slow and expensive first, but then it'll be very cool. Let's play the WAP ad. This is pretty cool.
Our parents told us to get real jobs, get off our phones, go to college.
I took a completely different route.
There's something different about kids who make money online. They don't give a. They taking a foot.
Sounds like heaven, bro.
There's a huge influx in kids dropping out of school. Shot mosi study, bro.
Dude, we got to grow faster.
Just use w, bro.
Yo, we're printing the super trim here in my garage.
College enrollment numbers are on the decline.
See that? Making money online, everything. A bunch of great cameos in here.
And once you taste that freedom, there's no going back.
Was that Blake Anderson there? It looked kind of.
I think that's another guy, but it does look like Blake. The.
How much of that do you think was AI?
I think it felt close to 100%, I think.
I think some of it might have been actual game footage. I'm not exactly sure, but interesting. Some of that could have been just generated in GTA 5. Yeah, it's possible just to actually go.
Shoot some of those. Yeah. The interesting thing here is it feels like WOP is 10 years ago. We talked about this earlier with Ryan. 10 years ago, the online entrepreneur meta was drop shipping and Shopify. I had kind of a crazy flywheel of different types of influencers online, promising mostly young men that they could make millions of dollars selling products online. And by the time people, I think, were making a bunch of courses about it, I think it had become incredibly difficult to actually arb and buy a product from a traditional US brand and sell it on Amazon and make any kind of money. And so that whole industry, like, effectively has shifted over the last. Over the last. I want to call it, five years into what, what are now like info products. And so WAP has this effectively an economy of people selling info products and then armies of young people that, that basically do like bounty style work for them. So clipping for a musician or clipping for a live streamer, things like that. It is a cool way for somebody in high school to make their first money online. And I think that is like a formative moment to have real dollars hit your stripe account. But you know, this, this, you know that kind of meta is constantly evolving. I think they have over a billion dollars in transaction volume. So we got to have the founder on a guest. He had some interesting takes. Are we ready for our next guest or should we keep doing.
Extremely tapped in. We have Jake, who's at. We're flipping the schedule a little bit. Let's bring in Jake and then we'll go to Craig. Nice to meet you, Jake. How you doing, Jake, Welcome.
Thanks for having us on, lads.
Yeah, thanks so much.
Great to have you.
We're dealing with a bunch of different scheduling stuff today. Would you mind kicking us off an introduction on yourself and your company? Me? Sure, yeah.
So my name is Jacob Glanville. I'm the founder and CEO of Cinevax, previously at Pfizer. I also founded a company called Distributed Bio. At Cinevax, we focus on universal immunity technologies. That's a universal flu shot that we recently closed a Series A to be in humans in eight months. And then some of you may also have seen some of the news around our universal antivenom program. That's a single anti venom that can treat all snakes.
That's amazing. I hate snakes. I actually am like, it's the one thing I'm terrified of.
John always checks his boots for snakes.
I do.
Indiana Jones, you race a Series A and you're gonna have a drug in market in eight months. Is that how much did you have to raise prior to get to that point? You know, usually we hear about bio companies having to raise pharma companies having to raise a billion dollars to even have a shot at getting something in market trials.
We're going to be in humans in eight months. So that's the first human trial. So there'll be more to spend. We've raised. Prior to this Series A, we brought in about $50 million. Half of that was non dilutive from like the Gates foundation, cepi, the Navy, the Army, the Naval Medical Research Command and Rare and a number of other places. Because what we work on is a. It's a bio threats problem. It's also a global collective goods problem. If you think about how much money has been spent on $90 billion of productivity loss just on flu alone in this country every year. And so in addition to that, we had NFX and the Global Health Investment Corporation that were our backers. And then now we're raising, we've just raised another 45 million. So nearly 100 million raised gets us into humans, which is a really big deal because we show it's safe, but then we also show it's effective. And so we can establish the first universal vaccine and a platform technology that we can make more universal vaccines out of.
I definitely feel the productivity loss every time, every time I feel myself getting sick, I'm like, this time I'm going to power through it. I'm going to be just as effective. I'm not going to let it like let my email pile up because I just feel terrible and it's very difficult to.
How different is this path from the traditional biotech path? I feel like a lot of biotech companies, they go public really early. It's just a completely different path from what we typically see in venture. Are you taking more of like a Silicon Valley venture path or do you see yourself as the same kind of strategy as the biotech companies that we've seen like the other companies that have just kind of gone through the normal IPO early, then the stocks trading based on the FDA results and that type of stuff.
I historically have always tried to ignore the normal path and go what's the best path for this company? So my last company actually never raised any venture capital. I built a series of profitable verticals and we be exited to Charles River Laboratories without taking venture here. I need venture because we're going to go through these very expensive phase trials and even there we have accelerated approval and some other tactics that make it much less expensive than some of the ones you described. So what's going to happen for us is human trials. After that there are sources of potential non diluted funding that could give us really big checks. Once you're in phase one to potentially cover the phase two and phase three or we do a series B and there's maybe a series C to, to get the first product out. And the whole idea is to prime the pump. Once you're profitable, then it's a rounding error to run your other programs which is like so torturously expensive for the first product. Like once you, once you've got something commercialized then you can go build the other ones. You know, it's also possible that we get kind of some FOMO going because the current $7 billion flu market, there's four companies that are all competing over that and all of their vaccines are like, like 10 to 60% effective. Like 1 in 3 people who got a flu shot last, last season actually benefited from it. And so I'm imagining those groups are going to be like, I want to work with these guys, otherwise I might lose market share. And so there might be an opportunity there for an acquisition or partnership. Earlier.
Yeah. On the non dilutive funding side, what has been your take on the NIH discussion publicly? We had Andrew Huberman on the show and he was talking about some of the risks of cutting some of that early stage funding. The number that was being thrown around was around 40% cuts. I don't know if that affects you directly because you're kind of already running the company, but what's your take on the NIH situation?
Well, I'm in biotech in general. I think we've benefited and we're able to bring this technology forward because of non dilutive sources out there. So more is better. Our antivenom program has been supported by an NIH award. That said, there's. I'm working on a fundamentally new technology and sometimes we've had a hard time convincing grant agencies to go in. At the NIH in particular, I think some people just didn't get it because it was too different and new. That changes as we have more data. But also I think a little shakeup sometimes is helpful to be able to have them invest and try new things rather than just reinvesting in the past because the past technologies just weren't working. I think we need new stuff. So obviously I think the United States needs to maintain its preeminence in biotechnology. It's one of the the 21st century sectors where we have dominance and you don't want to lose that because, you know, there's China and other organizations out there that are countries that are very happy to try to fill those shoes. And once you lose that, the brain drain goes the opposite direction. And that's part of the reason why we maintain the preeminence is that we have the institutions that are propped up partially by government money to be able to go make all these breakthroughs and have your Genentechs and so forth. So I think for those reasons, I think the United States should remain the global epicenter of biotechnology because of macroeconomic reasons.
Yeah, Genentech, the original VC what? Hundred Bagger or something like that. Fantastic. Return.
What are your kind of goals on the efficacy side, obviously the goal would be 100%. But if the typical flu vaccine is 10 to 60% effective, what are you aiming for?
Yeah, we're looking at, like, we'd like to see somewhere between 80 and 95%. It's hard to get that last 5% for any vaccine because you have people that have, have significant, you know, immunosuppression or some sort of deficiency. But what we're seeing in the animals and all the work so far, I think probably 85, 90%, maybe 95 is sort of what we're aiming for. Like, and that's a big deal, right? That's. That suddenly changes the. Not just like, you get a, you get a shot. And that Pavlovian thing, I mean, that's a big problem with vaccine skepticism is people like, I got the shot and I still got sick. Right. And so you can solve that. People, I took the shot, I didn't get sick for two years. That's a, that's a better outcome. But the more profound thing that happens is that when you start distributing a vaccine like this around the world, the pandemic era is over. Because you don't go, oh, God, there's an H5N1. It's getting in cows, it's getting in people. There's just like, hey, guys, there's a reminder we're seeing some H5N1 in people. If you haven't had your Senti flu shot in the last two years, go get a booster, and then you don't have a pandemic anymore. And so that's a crazy transition compared to how humanity's been operating since the beginning of time.
What is the key unlock for the universal antivenom?
Yeah, so it's the same biology. It's the idea that there are these little Achilles heels on these proteins that like the viruses that can change just about anywhere, but they can't mutate. A couple key sites. Snake venoms. There's 650 species of venomous snakes around the world. They all have different venom, but they share certain toxins, and those toxins have conserved sites. The origins of this anti venom was I found a guy named Tim Friedy who had, for 18 years, he had injected himself with escalating.
Yeah, I was about to, I was about to ask. We covered, we covered that story and it was just like such an example of, dudes rock. This guy. What a hero.
Yeah, yeah, yeah. Tell us the full story. Because it's like, it's an Incredible story.
This guy just, he decided he was going to make himself immune to snake venom, but like all of them. And so he started collecting a menagerie of snakes in his basement, like all the gnarliest snakes around the world. And he started off with tiny little doses. I don't know if you guys saw Princess Bride, but it was like the iocaine powder of Princess Bride starts with tiny little doses. And he would slowly increase the dose until he was. He'd be able to tolerate a full dose. So normally it would kill a person easily. And then he started letting the snakes bite him. And so he had 202 bites and 654 immunizations over 18 years of 16 of the most deadly species from around the world. And so I heard about him and I'm working on universal flu vaccines. And I was like, if anybody has beating through their veins right now universal antivenom, it's going to be this guy. And so I contacted him, we started doing a research project together, and it was phenomenally successful. So it just speaks again to the power of universal immunity to be able to completely change the face of 21st century medicine.
What is the, what is the distribution of that antivenom gonna look like? I imagine we should put it everywhere, right? It's always first aid kit. Growing up, I mean, growing up in California, like being in the country a lot, it was always in the back of my mind as a kid of this like, fear because I'd come across snakes all the time. You know, this fear of, you know, you don't necessarily know what bit you and then you don't necessarily.
You.
First of all, is it venomous or not? Hopefully you're not getting bit by snakes a lot. But then if a person is getting treatment, are they even going to have the right. So I'm curious what that kind of go to market looks like. How do you get this distributed as widely as possible?
So I mean, the way you described it. That's right. And it's worse in the developing world. So there's a lot more people who die and get permanently messed up by snakes than you might think. It's 140,000 people who die every year and another, another 300 to 400,000 that lose a limb, which really sucks if you're in a village because then your family's having to take care of an invalid when really they need everybody to be able bodied. And so, yeah, right now there's 650 venomous species of snake and they make one antivenom for each Snake or sometimes a couple snakes in one bottle. But so you get bit and first off they have to bring you probably multiple hours sometimes to get to a hospital. Has IVs and then they go, oh, I'm sorry, bro. Were you bit by like a super venomous snake? Did you happen to have the presence of mind while you were being bit to ruffle through the grass and grab the snake and bring it in doggy bag so we can take a look at it so we can check?
Yeah.
Did you bring the killer snake that just bit you? Did you bring it with you?
Otherwise we don't know if, if they. We stalk the anti. If it even exists and we don't want to give it to you because it's made out of horse antibodies and it can cause a whole bunch of like gnarly side side effects. And so that's, that's the world that we're trying to replace. What this is now is a single. We're halfway through, we've got half the 300 species of neurotoxic snakes and we're working on the other half. That was what the cell paper was. But the idea is to have a single product. It's fully human antibodies, so they don't have all the side effects of the horse antibodies. You can lyophilize them, freeze dry them. And so the idea is like an EpiPen for snake anti venom. You have some solution and you have some dry powder. You just, you get bit, you turn a little crank, shooky, shooky. And then you stab it into your leg and you get the antivenom immediately. And that means it can be in backpack, it can be out in villages outside of refrigeration. There's no multiple hours of waiting and they don't need to know what kind of snake bit you. And that will change a lot of people's lives.
That's amazing.
Very cool. I'm glad you guys found each other.
Yeah. Thank you so much for stopping by. We have a slam schedule today, but this is fantastic. And thank you for fighting the good fight against snakes. There's nothing I truly hate more. So I really appreciate it. Making the world one man, one man versus 600 snakes. My literal only phobia. So thank you. I have no problem with spiders, no problem with anything else. I hate snakes. So thank you so much for doing what you do.
There's a part of your brain that actually is networked, so if you're shown a bunch of pictures, you'll spot the snake picture first. So there's Something deep in our brains that has taught us recognize and fear snakes from primordial times. So what you're experiencing is not unusual. We've been built to tell these things. And I'll pass on your thanks to Tim because he's the one who really.
Just credit for this. Yeah, I'm also Irish, so there's probably something about the Pied Piper and wanting to rid the world of snakes. It's in my blood, but we'll get to that. The snake eradication project. Second, thank you for making the world safer.
Congrats on the raise.
And congrats on the raise. Good luck. We'll talk to you soon.
Cheers.
Bye. Next up, we have Craig Piggott from Halter. I was very excited for this company. They raised money.
This is the most excited you've ever been about a startup.
Well, it's an extremely bullish story and we'll hear it from him.
It's about running your ranch the way you've always wanted.
Yes. Thank you so much for joining. Craig, how are you doing?
I'm doing good. How are you guys doing?
You're overseas, correct?
I'm in New Zealand, yes.
Awesome. Thank you. I think it's early your time, I guess. Is that right? Awesome.
Like 8:30.
It's not too bad. Okay. Not too bad. Well, thank you so much for joining. Why don't you kick us off with an introduction on yourself and the company? Cool. Yeah.
Craig, founder and CEO of a company called Halter. Originally from a farm engineered by training. I did a stunt at Rocket Lab working for Peter Beck, which was kind of my intro to startups. And then, yeah, founded Halter. We effectively are an operating system for ag or for farmers and ranchers. We built a collar that goes on a cow. That collar is solar powered and runs a bunch of ML on the edge to train the animal to respond to cues. So we can then virtually fence and shift these animals from a phone or from the farmer's phone, which means we can run more productive farms. Ag covers about half the planet's habitable land mass. So that's like the. The goal, I guess, is to lift the productivity of the. That of that land.
Interesting. So, yeah, what was the state of the art beforehand?
Fences.
Just cowboys. Yeah. Barbed wire, literally.
Barbed wire fences for hundreds of years have lifted the productivity of land, but obviously they're pretty expensive and hard to move and a lot of shortcomings.
And so how do you actually steer animals with just a column collar?
The collar has cues on it, like sound cues out of each Side if you imagine maybe like parking your car and as you like back into a wall, it beeps. And so it gives you feedback on how close you are to a virtual fence. So we use sound and vibration and then we train the animals with a low energy Pulse. It's like 100 times less than an electric fence that say most cows in New Zealand would be used to. And yeah, the collar does all of that training and kind of animal guidance.
What has been the story around cowboy employment in the context of adoption of tech? I imagine this is the kind of thing where I'm using cowboy sort of as a joke. But I imagine by introducing this technology you can free up ranch workers to do more important things, like for example, taking care of a sick cow or something like that. Has that been the case?
Yeah, like there's been massive labor shortages on farms and ranches for a long, long time. It's obviously not glamorous work. And so I think our tool like Holter does automate a bunch of work, but really it also enables you to just be more precise and kind of run a, like a more productive operation.
And then you can.
And any extra time you do have, you spend on the never ending kind of to do list of running a ranch or a farm.
Yeah.
Can you give us the state of the business? I mean.
Yeah, I want to know about scale. How many cows are we tracking?
I think people don't understand the scale of your business. So I just want to hear from you.
Yeah. We're live in New Zealand, which is where we started. Australia and the U.S. i think we're across like 18ish states in the U.S. run over a thousand farms, hundreds of thousands of cows which we fence and shift and track and monitor their health every day. And growing over like 2x year on year at the moment.
So amazing.
It is like. And these, well, you got a lot.
Of room to run. I looked it up. There's one and a half billion cows in the world according to some survey. And I'm sure some percentage of that are just relatively wild and not billion cows.
Wait, cows are expensive? I feel like that's like a trillion dollar market. Then it must be. Yeah. I had no idea.
The funny thing about the TAM is it's like it's very different to most businesses. It's kind of a function of land mass, not like population. So you do end up with, with like a very distributed TAM through South America and Europe, like everywhere. So. But the market is huge. So we're trying to remain focused on nz, Australia and The US like sure. Even within that is just a multi billion dollar opportunity for us.
So would you guys, would you guys ever do health tracking for the animals? Is that something that you already do? I imagine farmers would, would love to understand like metrics around stuff like heart rate, hrv, things like that, steps, et cetera.
Yeah. To do the guidance, well you need to understand each cow's like behavior and how they respond to the cues and how best to train them. And so kind of a byproduct, maybe underselling it, but as a function of that you then can also do health and fertility and things like that very well as well.
So.
So that's already part of the product whether a cow's yet sick or not healthy for any reason or if they're fertile, which is important for breeding and things like that. So yeah, if you've got the hardware there that is just a over there update to the collar and then we kind of turn that on as like an additional SaaS product.
How SaaS beautiful.
Is the business model mostly like subscription SaaS structured or is there a hardware component and a fixed cost as well to kind of onboard?
It is.
The hardware is bundled in the software.
So it's pure.
It is like very SaaS. Like there is tiers you can start on core and upgrade to Pro or unlimited. And there's a small kind of upfront installation fee. We have to put up towers on these farms to run the network to talk to the collars. But predominantly it's just an ongoing subscription. You don't buy the collars or anything like that.
Yeah. Take us through the raise and kind of what the plan is going forward.
Yeah, so we just closed a hundred mil Series D from led by Bond.
Very meeker.
There we go.
Awesome, awesome.
Firm and probably two things really, like more expansion in terms of new markets and even just pushing harder into the markets that we're currently in. And then a huge amount of product and R and D work. Like we have kind of for the first time taken a very analog operation and converted it into this digital farm. Like farmers spend two hours a day in the product and they're doing everything from animal health to the fencing and shifting to like measuring their grass and all that. And there's just so much kind of room to run on new product and shipping new products. So that's the other kind of half of the raise is just doing more kind of building our product and engineering teams and doing more R and D on that side.
Will we see cow bcis in the next 20 years bcis.
Good question.
Don't know.
Don't know. Yeah, maybe unnecessary, but you know, if you can do something, you know, a collar is potentially more elegant solution than.
Probably pretty costly to implant the brain chip, but who knows? Anything's possible.
If anybody's going to do it, it'd be you guys.
Yeah, well, thanks for stopping by.
There's definitely some benefits to being on the, on the outside.
Yeah, for sure. It's a little bit easier to change it out and clean it off if you need to.
Awesome. Well, congrats on the raise and yeah, come back on when you have more news.
Yeah, thanks so much for stopping on.
Cheers.
Up next, we have Jonathan from Grok coming into the studio talking about chips. Hopefully we can get him in. We've been juggling things. How'd you sleep last night, Jordyn? Go to 8sleep.com?
Absolutely, terribly. My 8 Sleep couldn't save me.
My, you can get a pod 5@eightsleep.com tbpn.
Jonathan, we got somebody.
Yeah, we have Jonathan here. How's it going? It's good, how are you?
Welcome.
Thanks so much for stopping by. Sorry about the little mix up with the scheduling. Glad you could make it though. Great to meet you.
Thanks for having me.
Would you mind introducing the company and the current state of the products that you sell just for anyone that might not be familiar?
Sure. So the company is Grok Groq and we sell tokens as a service, so very much like you use OpenAI's API or Anthropic's API, except we serve mostly open source models and we've built our own AI chips that allow us to do this much faster than anyone else.
Yeah, I remember the initial demos. I believe it was on an early version of Llama and the inference rate was just so much, so much higher. What were the key decisions that you made to enable faster inference?
One of the key ones was we don't use any external memory. Every time you read from memory, the chip just has to wait for it and it's very slow and it's like drinking a drink through a martini straw. And so what we've uniquely been able to do is keep the cost down while also being very fast. You can actually get speed off of a gpu, but then it just becomes completely uneconomical to use because you have one user using that at a time.
So what are the trade offs of not having off chip memory? I imagine there's a big discussion over how important are large context windows. Is that an important factor to how LLMs are scaling. And ultimately, you're somewhat beholden to the decisions that are made by AI researchers at big labs, right?
Well, interestingly, we actually support very often the largest context or the largest context that's practically offered by anyone else. And this is a little counterintuitive to most people. The trade off we have is memory capacity. But GPUs have a memory bandwidth issue. So what the researchers do to improve the memory bandwidth issues, always improve the memory capacity issues for us. So we can run all the same models with the same context length. The trade off is actually something else. You could think of us a little more as a nuclear reactor instead of thinking of us as a Diesel Generator. Whereas GPUs, you can take just eight GPUs and use them for a model. We need a lot more of our LPU's, but when we're running that model very much like a nuclear reactor, we get much better cost and it's better for the environment because we, in our case, we use less energy.
Interesting. What does the shifting landscape around different custom silicon look like right now? I feel like every hyperscaler is taking it more seriously. We've heard stories of Microsoft wanting to pull back from Nvidia chips. It seems like Amazon's kind of doing the same thing with Trainium and Inferentia. Are we going to just be in a world where every hyperscaler has their own chip, or do you think that the market will fragment in kind of a different way? Way.
My background. I'm actually the person who came up with Google's TPU chip. So I started 20%. No way.
Wait, it actually started as a 20% project.
It did. It was unfunded.
That's amazing.
Yeah. So there are 20%. That's a real thing. But what most people don't realize is that there were actually three different AI chip efforts at Google. And that was the only one that actually survived and was competitive with GPUs. The other two ended up getting canceled.
Canceled.
And so when you start looking at the broader and broader world of chips being built, a very small number of them are actually going to be competitive with GPUs. So most of those are going to be canceled. But I think where people get really confused is they keep talking about the chips. And so it's not about the chips, it's about the software. There are all these little features like prefix caching, speculative decode, and all sorts of things like that. If you didn't have one of those features, you wouldn't actually be able to be competitive with the latest gpu, no matter how good your chip was.
Sure.
So the thing is, you have to first catch up in software and only then do you get to compete on the chip. And so one of the things that we did that was very unusual is we spent the first six months working on a compiler which makes our software work very easily so we can actually compete on the software level. With Nvidia, we have the same models running, and only then do we get to compete on the chip.
So are you talking about at Google that was like the push to break out of the CUDA ecosystem, essentially?
Well, actually this was at Groq. So at Google we had these handwritten kernels which are these sort of assembly routines. And it's the same as with GPUs.
Okay.
GPUs are actually more similar than what we've built here at Grok.
Yeah. How, how do you think about the, the current AI paradigm and how long like the current algorithmic machine regime will last or where, how it will change? Because I feel like the narrative around companies that bet on a particular, you know, okay, we're going really long on the transformation or something like that, then there's always this question of like, will the transformer stick around? Well, it's been around for a long time. It seems really good. Maybe it'll be around forever, maybe it won't. How do you think about those? Is that the right question to be asking? Yeah.
I remember when we were at Google, we hired this really senior person right before we were done designing the tpu. And he came in and he said, you cannot build a generally programmable chip that will out compete GPUs. It's impossible. Instead, we need to stop what we're doing and just create a chip for a particular model.
Sure.
And we had to walk him through the whiteboard and show him, but we actually ended up proving to him we made a faster chip by making it general. And the counterintuitive part was if you're actually able to optimize some of the circuitry in the chip to be reused for many different things, but you're only designing a couple of those circuits, you can make them much better than if you have a whole bunch of, of different stuff that's not as optimized. And when you look at one of these models, you can't fit one of these models on a single chip anyway. You have to use multiple chips. So are you going to design a different chip for each part of the model? And it starts to become ridiculous and doesn't make any sense. There's actually been people recently who've been trying to revive that idea. But in the time when they started, since they started, MOEs became a thing mixture of experts. So if they had already started designing their chip, the chip would be obsolete. And then now we're seeing other sorts of things emerge that are very interesting. And if any of those win, those chips will be obsolete. So we always focus on as general of a programmable chip as we can.
Switching gears a little bit. I believe you're in Europe right now. It looks dark outside and it looks like you're.
Thank you for staying up.
Yeah, thanks for staying up to hang with us. Talk about. About the decision to put the first European data center in Helsinki and kind of everything that went into that.
So we were at an event here, an AI event, and about a month ago we decided it would probably be great if we unveiled a European data center while at this event. And literally a month ago, we now have it up and running, actually running models. So I think that's a record for time to decide to deploy and getting it up and running. We actually.
Congratulations.
Well, we actually weren't expecting it to have it until the end of this week, which would have been insane in and of itself, but it's actually up and running now. So kudos to the team.
The team is probably sweating because they're like, okay, this is the bar. Now we actually have to be faster than this to.
The previous bar was 51 days when we deployed about 20,000 chips in Saudi Arabia. It was 51 days from signing before we set it up.
Oh, I remember that deal. Yeah.
Now it's like a month and three days or something.
So what makes that possible? Is it just the. Is it the energy efficiency? Is that a huge part of it?
Best friends with Morris Chang.
Use TSMC to make our chips. Okay, yeah, in our case, we've actually simplified the architecture pretty significantly. We don't use any external switches as part of our core interconnect, only to move data in and out, out from the system. It's pretty simple cabling, air cooled still, we're not liquid cooled. We like to stay a generation behind on all of the technology and just compete on the architecture and the software. And so that allows us to use these highly proven technologies where everyone's already debugged them and we're not debugging them as we deploy, but also because we don't use that external memory. So that external memory tends to lead to a lot of chips that fail in field. And so as you're bringing them up, a lot of them fail. You have to fix them. And there was a recent model that we ran that required 4,000 chips. The first time we ran that model on the 4,000 chips, it worked.
And we would have never been able.
To do that if it was 4,000 GPUs. The first time you run something at 4,000 GPUs, you have to replace a bunch of them.
Yeah. Because it's problems everywhere. There's been some rumors that, that you've operated the, like the, the token business, the API, at a loss. That feels like a completely rational strategy in many ways. Nothing wrong with that. But can you comment on that as a, as a strategy? Is that just a misconception? Is that deliberate? How do you think about just investing for adoption? Broadly? Because there's, like, it's really important as a business to grow, and there's a whole bunch of different strategies and tools in the tool chest. What do you think is valuable, What's. What's been the strategy to date and how's it evolving?
Let's get even more specific than at a loss. Because when you're running a service, you can claim you're not at a loss because you have an infinite amortization time on your capex and your chips, and that's what a lot of the Neo clouds are doing. So we actually agree, aim for a more aggressive payback period than anyone does with GPUs. And part of the way we can get away with this is, first of all, our electricity cost is about one third per token versus a GPU. So our OpEx is already lower. And that's sort of the flat you have to beat your opex, otherwise you are losing money, period. There's no accounting tricks you can do. And then on top of that, the question is, how quickly can you pay back? For us, everything we run is above our OPEX cost. So we're not losing money. But different models have different payback periods. Some of them as quick as two years on our V1 silicon. Some take, you know, four years. And it's a blend, right?
Okay.
And that overall blend is what we're happy with.
Makes sense. Yeah. On that note, there's one world where, you know, the models are getting better. Everyone just wants the best model for every single task, and they're just willing to pay for the latest and greatest cost of. There's this other interesting model where I feel like we get a new capability, whether it's just the ability to convert, you know, CSV to JSON or something really? Or like raw text to JSON or data extraction or, you know, censoring, you know, bad words and a whole huge transcript. And like the problem is completely solved by the LLM and you don't need to throw a more advanced model at it in the future, even if one comes down the pipe. You just need speed and reliability and eventually cost. And so do you see, is there an important part of your business where you can be the baseline provider for a current capability that will continue for decades, potentially even as a newer model comes online? Or do you want to try and shift the business towards being constantly on the motion aggressive side of the frontier? And is there actually a trade off there or can you just do both?
So a very common usage pattern we see is that people will develop their systems on the latest and greatest model.
Yeah.
And they will optimize it for two things. One is they'll optimize it for speed and the other is they'll optimize it for cost. Very often the most successful AI startup startups are spending almost as much per month on token as a service as their total revenue. Right. And they'll lose money on paying employees and things like that. But they're trying to keep those balanced. Yeah, we can bring the cost per token down. Interestingly enough, the general inclination is that they want to spend the same amount, they just want more tokens because those tokens can actually turn into more quality if you can iterate, if you can do more in parallel, the speed thing is super important because that's engaged engagement. So if you think about it, every time you increase the performance of a model by about or any service by about 100 milliseconds, your engagement rate or your conversion rate tends to improve about by 8%. That's on desktop, on mobile it's over 30%. And so no patience on mobile whatsoever. And so what we see is people get stuff working, they prove that it's possible on the most capable model and then they'll try and find one of the models that we run, all the open source models. And then when they're able to find one that works for them because they have to try a couple, then they're very happy and then they don't change things. And so there are a couple of models which are actually quite old, quite terrible, that we would love to deprecate, but there's a bunch of users because their system works and they don't want to touch it. The other Thing is, and this one's interesting, so we had one of our engineers try solving a bunch of math theorems and they were using this thing called Lean, which is a formal theorem prover, and they would use different LLMs in the background to solve, and then they would use Lean to test every single time. Claude Opus would actually solve it in the fewest iterations. However, Quin 332B running on our chip, because they're so fast, always solved the theorems faster. It may have more iterations, but it was able to go through them faster. And so you actually got the result faster with the less expensive model running on grok.
Interesting.
It's been a crazy year so far. We're about halfway through. What are you excited about in the next six months on the GROQ side? And then what are you excited about in the broader open source ecosystem?
Well, on the Grox side, we're continuing to scale up. We're going to add a bunch of new data centers and new continents. After this, I'm going somewhere else in the world. I'm not going to tell you where. Where we're working on getting another data center deal done.
Amazing.
Congrats.
But also there are some rumors that we might have some new improved systems or chips or something coming later this year.
And there's allegations, allegations of massive advancements, rumors of breakthrough.
A major breakthrough. Yeah, I love that. Can you give me your retrospective? But also your current take on just Moore's Law?
Broadly, yeah, I think Moore's Law is true, but I don't think it's the most important part and people miss this. So Moore's Law, for those who don't know, every 18 to 24 months, the number of transistors on a chip doubles. That translates to the economics get better, that translates to the chips get faster. And you can trade that off. What's really been happening in the last five to 10 years? And I don't know what to call this law, but like when I left to start Grok, one of the observations was every 18 to 24 months, not only did the transistors double, but the number of chips doubled. The transistors doubled and the number of chips doubled. And so when we started, we actually asked ourselves the question, what would happen if we assumed there were an infinite number of chips? How would you design the system differently? That's one of the reasons we decided to get rid of the memory altogether, because the memory just holds bits that are doing nothing, waiting for an active chip to do something with them. We're like, let's just make everything active and do everything all at once. And so that was a very important observation for us. And I think people are going to focus more on the number of chips doubling over the next couple of years than how much each chip improves.
Can you. We'll let you go. This is my last question, unless Jordy has something, but can you tell us a little bit about, broadly, what international leaders in other countries generally are thinking or what's driving their motivation to spin up token factories, these AI factories, new data centers, the sovereign AI efforts? And because I'm thinking back to, you know, the development of the original cloud computing revolution, and it felt like there wasn't as much of an incentive or narrative around, well, we can't possibly just let AWS come over here and set up shop. This feels different. What are the motivations? Is it economic? Is it, is it freedom of speech? Like, what, what, what are the. What are the shape of the conversations that you're having that get an international leader excited about setting up an AI factory, a token factory, a data center?
So it's a little bit less about fear and control, and it's actually more about they're not getting enough compute. So think of it this way. GPUs are expensive and there's only a finite number of them being built, not because they wouldn't build more GPUs, but because they rely on that external memory, which is a real bottleneck in the supply chain. And so Nvidia is going to build every single GPU that they can physically build this year. AMD is going to build every single GPU that they can physically build this year, and then they're going to sell every single one of them. If they could build more, they would sell more. That means that there's an allocation problem. Countries aren't getting all of the cheaper chips that they need for what they're doing internally. So, for example, when we built that data center in Saudi Arabia, there were a bunch of people who'd been waiting over a year to get GPU orders filled, and they had no idea when it was going to happen. And so they switched over to us because they could immediately get access to compute. So what you're seeing is a lot of these sovereign plays are more about making sure they get the compute. It's as if we're entering the industrial age, and all of a sudden everyone realizes, if I plug something into an electric socket, I make all of my workers more efficient. Well, if everyone else in the world is getting that or it feels like that's the case and you're not getting that. You're not getting the efficiency, you're not getting the compute to augment your workforce and make them more capable. You're at risk of falling behind. And so for them, it's as if all the power companies are focusing on some other really rich regions of the world and they want to get that power as well so that they can start running their AI models.
So is that more like the 5G rollout? The idea of just bringing mobile to your country versus Google search? I suppose. Is that the more accurate analogy?
Maybe that would be fair.
That makes a lot of sense. Jordy, do you have anything else.
Any expectations around OpenAI's alleged new open source model? Is that something that you guys are going to be integrating heavily with?
If OpenAI open sources their model, we will immediately launch it.
Amazing.
Awesome.
Well, that's great news. Well, thank you so much for taking the time. I appreciate you staying up late. We'll talk to you soon.
And congrats on the new data center, new record. Cheers.
Have a goodbye. Bye. And let's tell you about graphite.dev.graphite.dev code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster.
The best teams in the world, including TVPN of course use graphite.
What else?
We have some timeline.
We have some timeline we can go through. We got to get out of here soon. Yoni Rickman says he wants to get long mold. Did you see this?
Yeah. So I threw this in here. Request for startups. I want to get long mold. Basically he says mold is super hot right now. It's like the protein of environmental contaminants. It's in everything with no signs of going away. It's in the air. And this stood out to me because I once rented a house that ended up having mold in it. It was a huge disaster, took a lot of time, was making me sick. And I think a lot of people suffer from mold exposure or mold related illnesses and don't realize it at all. So there's a bunch of different ways that you can solve this. I don't necessarily know what they all look like. I mean one of the biggest.
Yeah, yeah, he mold testing, eight sleep of mold, smart air purifiers and testing. That makes a lot of sense. Then he says mold insurance. Maybe that's genius. I don't know. That's not what I would have thought of, but yeah. And then mold resistant building materials. That's kind of cool. Cool.
Yeah. So there's a bunch of different categories, so I wanted to throw this in here. If you are long, mold yourself and your business, go reach out to Yoni. I would love to see more solutions here.
In other news, tomorrow night there will be a Grock 4 release live stream. Wednesday, 8pm Pacific. They're staying up late.
I love how they always do their streams after. After everyone else has left the office and gone home. They're like, we're still here. We're gonna go into our tents, come out of our tents and do a stream. So I'm excited for this. I mean, this is going to say a lot about.
The new Memphis Data center since Grok 3. So, yeah, there will be a big question about where this hits. Also, we hit this pre training. While it feels like a lot of people are saying we need new algorithms, we need new ideas and I'm sure they'll be able to. The question is not like, can grok, can XAI build a big data center and get to the frontier? The question now is like, can they push past the frontier with some actual like completely unthought of innovation? So that'll be interesting to track, but either way, there's a ton of questions.
I'm interested to see what they're doing on the image side. Video side.
The video side would be very interesting. I mean, X has a lot of video data because X has been uploading video. You know, certainly not as much as YouTube. But a lot of YouTube videos get uploaded to axe info.
Yeah, imagine. Imagine if you can tag Grok under like a meme and say, animate this.
Yeah, yeah. Turn this into.
22Nd GTA.
That'd probably be very expensive on the inference side, but, you know, maybe worth it. We have a lot of GPUs. Should we close this Audi ad? Beautiful.
Let's do it.
Advantage Audi, beautiful ad.
I couldn't tell if this was a real ad.
10 out of 10. No, nothing. Notes.
I couldn't tell if this was actually live. If it is, it's four tennis balls.
If it's not in the shape of the Audi logo. Fantastic.
Fantastic.
Well, thank you so much for tuning in. Today was a little bit tomorrow, chaotic start 40 nature startup and they didn't.
Want us to podcast, but they couldn't stop us.
They couldn't stop us.
They'll never stop us.
So we will see you tomorrow.
Have a great afternoon. Wednesday evening. See you soon.
See you soon.
Cheers.
Bye.
More from John
Shifts In The Creator Economy, Kylie Jenner x Meta, GPT 5.6 Limited Release | Diet TBPN
Snap Specs, Taste, Midjourney Hardware | Eric Newcomer, Merrill Lutsky, Carter Reum, Swami Sivasubramanian, Thomas Suarez, Mark Gurman, Ryan Daniels, Isaiah Granet
Midjourney Medical, AI Talent Wars 2.0, Jake Paul Joins | Derek Thompson, Rene Haas, Robert Slaughter, Rob Reid, Thais Castello Branco, David Senra, Jake Paul & Geoffrey Woo
YC Demo Day Lightning Round, New Snap AR Glasses, SpaceX Rips | Garry Tan, Andrew Lee, Stamatios Floratos, Hugo Frisk, Efraín Torres, Russell Smith, Payton Case, Akshay Trikha, Diana Hu, Harj Taggar, Connor Hayes, Luke Burgis, Anda Gansca