Jobs Boom, Astra Reactions, Cybercab Roll Out, The Dyson Debate | Hunter Somerville, Bar Winkler, Bridgit Mendler, Carina Hong & Ken Ono

You're watching TVPN.

Today is Friday, September 4th, 2026. We are live from the TVPN UltraDome, the temple of technology, the fortress of finance, the capital of capital.

That's right. Let's get that Christmas countdown.

First, let me tell you about ramp.com. Time is money. Save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place.

How many days? There we go. 111 days until Christmas.

111 days until Christmas. The Christmas countdown is helpful.

Because one gives you something to look forward to.

Indeed.

But it also is a subtle reminder that you only have 111 days left in this year.

Yeah.

It's game time.

It is. It is.

It's game time.

Well, we're excited for it. Big news in the AI world, obviously. Astra GPT-6 launched yesterday. We went through the blog post briefly. It went up mid-show. Bunch of impressive benchmarks. We talked about ArcAGI. We moved the goalposts because ArcAGI is now saturated, which is particularly crazy because of how detailed those games are when you play them. The thing that stuck out to me though was watching the demos of people build 3D scenes in Blender using Astra. Clearly a bunch of people got access to the model before and were able to fire off one prompt, maybe a few prompts. It feels like it's very, very capable to just Fire up Blender and do a full task and model a whole world, basically.

That said, Matt Schumer, who had a couple of the most impressive examples, he has been talking about how the art of prompting is back.

Oh, interesting.

And he has— he had a very specific process where you have, like, a manager agent and subagents. And it's not— still not as simple as just, like, You can get a good output doing, you know, in one shot, but to get some of these really incredible outputs does take a little bit of refinement.

Yeah. And I'm sure a lot of that information will propagate across the internet as it already has. I'd love to pull up some of the images from, or the videos from the launch from folks who are actually test driving GPT-6 Astra. First, Sharif said, GPT-6 Astra recreated the Palace of Fine Arts in Blender. This is my favorite building in San Francisco because it was built for the World's Fair in 1914. Oh, I didn't know that. Where the steam locomotive and telephone were showed off.

Wow.

It was a time when technology gave us all a deep sense of optimism for the future. I feel like— I feel like some of that sense has been lost since, certainly. But I hope models like Astro can help restore it and push us towards a more hopeful future. So if we can play that actual video, you can see. So, I imagine that it pulls images from the internet. There's a whole bunch of obviously reference diagrams. There's a whole bunch of different ways that you could ground a 3D modeling exercise. You could even pull the blueprints. There might be 3D models out there.

Yeah, there's probably architectural plans.

And yeah, it's this odd thing where you could use an AI image generator because the AI image generators are knowledgeable about what the Palace of Fine Arts is. It could generate a bunch of images there. So there's all these different ways to triangulate it. And then you can also just go on, I think Sketchfab is one of the big free ones free 3D model purveyors online. And so you could potentially go grab an actual 3D model, build off of that for your scene. So there's all these different ways to solve a problem, and Astra seems to be good at actually going and doing it. There's a few more that we should look through. There's one comparison since this is the new Pelican bench apparently. Same prompt, same goal of creating this fly-through video for a real estate scene. And this feels like something that could be useful. I've already seen on Zillow, there is the ability to restyle images. So if you're looking at a house on Zillow, you can see the interior photos and then you can say, I'd like to redecorate this in a modern style or in a contemporary style or in a colonial style. And it will go and use AI image generation on the backend to do that. It's only available on a fraction of Zillow listings. It's clearly some sort of feature that they need to turn on. But you could imagine this being something, especially for spec homes that are selling that maybe haven't been— that they don't have the budget to actually go do proper rendering, just lowering the cost. You get sort of a Jevons paradox moment. There's another one from Thomas. He says Astra is very good at 3D modeling, and the lighting is way better in this one. This is what— this actually looks properly ray traced, properly lit, really thoughtfully designed. Maybe that's in the prompt. Maybe that's just the way this one turned out based on the photos. From a Blender scene to then an Unreal Engine 5 walkable experience. So everything becomes translatable. You'll be able to take your tweets, turn them into blog posts, turn them into videos, turn them into 3D games on Steam pretty soon.

GTA-style video games.

Basically. Basically. And so there's a little bit of a prompt and Thomas reveals a little bit about his process here.

I would like to see VCs, if you're going to put out an AI blog post, a blog post—

Like a market map.

An essay. No, if you're going to put out an essay that was clearly written by AI, you should be required to put out a video game that accompanies it that's sort of playable that gets the same concept.

I'm thinking instead of a market map, make it a Fortnite map. I want to be able to play in the different— Town is a popular AI startup, might land on a market map, should land on the virtual Fortnite map.

I like that.

As well. Who else said this? Yufan Yi. I think this was interesting. There's another— there's some debate over this last one, whether there was like a different harness involved. But same thing, give it a Zillow listing, it can 3D model the house. So this, it's always with these new releases, it's interesting what makes you feel the AGI or not. Realistically, Frontier Math Tier 4, I think they scored 99% on that. Terminal Bench Science 0.1. They're just numbers to me. Like they don't do anything for me emotionally because I don't have any grounding in how hard those particular things are. And also I sort of assume that computers are good at math.

And you didn't get past algebra and look how successful you are, right?

Exactly, exactly. So those are rarely—

I accomplished all of this off of Algebra 1.

Yeah, yeah.

A freshman-level math Olympiad.

Yeah, exactly. So those are never really feel the AGI moments for me. But this— but I've actually spent hundreds, if not thousands of hours in 3D modeling software, mostly in Cinema 4D and Houdini. And it was really fun. I really enjoy making 3D renders and motion graphics. And it was also useful when I was making YouTube videos. But every step in building a scene, requires you got to watch a YouTube tutorial, then you got to look up how all the different functions work, and then ultimately you're tweaking settings in literally 4 dimensions because it's a 3D scene. But then over time, you have to map where the camera goes, where the different objects are moving, how things progress over time. And it can also be very compute intensive. So there's a whole bunch of times where just to preview the scene in low res, you might have to cache and load and pre-bake certain animations so that you can just see what it looks and then you're like, ah, it's not there. I gotta go back and change some setting for some sort of smoke simulation or some sort of fracturing that's going on. And so it's fun, but it is extremely cumbersome. And ultimately you're doing this on a 2-dimensional screen and it can be very time-consuming and frustrating. And so by the time I stopped spending time on motion graphics, which is around 2021, I'd been building the YouTube channel and I hired a few, and a few editors. Some of them were good at After Effects, some of them learned Blender. Blender was ascendant at that time. So Blender is the open-source, community-led effort. Cinema 4D is a closed-source piece of software, very expensive— was very expensive, was like over $1,000 to get a license that would be good for like a year. Then they eventually went subscription. And so there were pieces of software that were used by Hollywood but were very expensive, very high barrier to entry, so you didn't have as much of a flourishing ecosystem. Blender in 2021 was clearly on this trajectory where it was going to build this ecosystem and become very, very popular. Um, and, and it did. Uh, but by 2021, I was able to hire, uh, video editors who were able to go do scenes in Blender. But still, it was a huge ask if I was like, for this YouTube video I was posting every single week, I'd like a motion graphic to illustrate something. And I did this one where we built a 3D battleship board for this US-China military competition, uh, competitive analysis. And, uh, I mean, it's like it sort of ties up an editor for a week while they're working on this. And, uh, and this just speeds all of that up. So you're gonna get back to like the, is it over for motion designers that worry about job displacement? I think we've been through enough of these cycles that hopefully We're Jevons paradox pilled at this point. We imagine—

You just comp it to software engineering.

Yeah, exactly.

There's a lot more 3D renders. It seems like everything that can be— anything that can be rendered will be rendered. And it's just— and more people are going to try to make something themselves and then hit the point where they're like, okay, there's this sort of area where the models can't get it to 100%. Now, I need to actually go find somebody who's an expert and work with them.

Yeah, totally. And so, we're back on the is this AGI question. The Wall Street Journal said, yes, we are entering the era of artificial general intelligence. It's in the journal. They've declared it. Although entering the era, Like the door, the house is AGI, the door, and we are entering, we're like almost in the AGI era according to the Wall Street Journal. I don't know. Stripe called it AGI. Didn't they call it AGI Jan 1st, something like that?

Yeah.

And there's been various calls. Tyler Cowen called it AGI like 3 years ago. He was ready to rock. I think that was GPT-4. Something like that. No, it might've been '01. I think '01 he was like, this is AGI. But it's clearly artificial, clearly intelligent.

Don't you think he'll end up looking pretty good in the fullness of time on that call?

100%.

Because it clearly is this timeline, a spectrum. I do like to imagine the stripe, the Collisons, just after midnight. There's champagne, maybe some sparklers, some fireworks, and they just go, that's it. It's AGI.

It's AGI. But the question is no longer, is it artificial? We know it is. The question's no longer, is it intelligent? We know it is at a lot of different tasks. The question is, how general is it? Because Blender is open source, can be downloaded and copied a million times into reinforcement learning environments. We know that OpenAI bought, quote, tens of thousands of Mac Minis and Mac Studios to do computer use training on. And so the question is, is it only good at Blender? Will it work in Cinema 4D? Will it work in Houdini? Will it work in a small piece of software that does 3D modeling that has fallen by the wayside? Now, there's an interesting dynamic here where I was in the boat of I could see that Blender was going to be the new hot thing, the best tool, that I should have moved. But I already had so many plugins and textures and rendering, and I knew the layout, that I downloaded Blender. I opened it up. And I was like, ah, I really don't want to have to get over this learning curve. Even though I know in 5 years Blender's gonna be way better than Cinema 4D, there's this switching cost. Now, you're in this weird world where it's pretty good if you're a software company where the frontier lab's bullseye lands on you. And I'm thinking about Slack, like Alex Wang at Meta just moved them to Slack off of Teams where they had like an internal workspace thing. And specifically he said—

Give it up for Benioff.

Give it up for Benioff, truly. And specifically he said Slack is better for work with AI agents. And why is that? Well, it's because the Frontier Labs all use Slack. And so they have Slack as a reinforcement learning environment. And so it got really good at Slack. And I think that even though there are open source Slacks and you can vibe code a Slack competitor, as we have tested, Slack was the logical tool to pull off the shelf for the AI companies, for the AI agents. for the models, and Blender seems to be that. So the interesting thing is that I do wonder if people will wind up shifting from Cinema 4D to Blender because the models are better at Blender, or will the models generalize and say, yeah, this looks a little bit different because it's a slightly different piece of software, but I've trained so many hours in Blender that I can also do Cinema 4D or also do Houdini? Because some of those other softwares are better at certain things. Some of them are better at smoke simulations, some of them are better at water simulations. There are advantages to each piece. I don't want to degrade any particular piece of 3D software. I love them all equally. But I think—

There are some reports that it's quite effective at video editing.

Cool.

So, I'd be interested to take it for a spin. I wonder— the interesting thing is that most video is still edited in Adobe Premiere. It's a closed-source Adobe, you know, they have their guardrails. They probably— they might not want to, you know, say, oh yeah, sure, take 10,000 licenses, no problem, pay us $10 a month for that. They might have a different opinion. They may want an AI agent that's internal. There might be all sorts of different things. But then again, there are probably— I mean, I know that there are open-source video editing suites out there. I think that they lag a little bit, but— Yeah. There are some that are free or open source at least. The other interesting thing, I mean, video editing, like all these things, like taste is so important in these. It's often less about the exact instructions and more about the, like, telling the right story, using the right tools, not creating slop, whether you're doing it yourself or with an AI agent. But I think that gets to the core question of how do you evaluate these model launches? We were discussing the idea that the benchmark era might be coming to a close in some way because as soon as the benchmark's created, it's hill-climbed very, very quickly and then it's at 99% or 100% pretty quickly and everyone's sort of— their eyes start glazing over in benchmark discussions. And it just goes back to like the vibe check. I think the best thing to do when a model comes out is just take it for a spin and mostly point it at something that you know really well because you don't want to be in the— you want to expose the model's—

what's the—

what's the Gelman amnesia? That's the one I'm thinking of. You basically want to test if Gelman amnesia is triggered by you using the model. So, point it at something that you intimately understand and see if you're impressed. And then— because if you point it at something—

Yeah, like you would be horses.

What?

You'd be horses.

Yeah, yeah, we got to do the horse bench again. We have a couple more benches. Let's see comedy. I want to see some new jokes. Shrimp fried rice bench. Also interesting, Gelman amnesia, not created by Murray Gelman, the physicist who was at Caltech, who lived in my hometown, and his house is actually for sale right now. But he was having coffee regularly with another physicist. I forget what his name was, but he's really famous. And the other guy was like, we noticed this together, I'm naming it after you. Do you have the story? Did you look this up, the origin? It says it's named after Nobel Prize-winning physicist Murray Gell-Mann. Yes, it's named after him, but it was coined by his buddy, who was another physicist, who was like—

Michael Crichton.

Is that it?

Okay. Yes.

Yeah. So, they were hanging out, having coffee, they realized this thing, and it's like—

Tyler should name this specific situation—

Yeah.

Cosgrove Paradox, where—

Where someone asks you to fact-check something they don't really know anything about? Anyway, we can move on because there's a whole bunch more news. Let me tell you about Cisco. Critical infrastructure for the AI era unlocks seamless real-time experiences and new value with Cisco. So, oh, there are other reactions to Astro we should go through in the timeline. The big one comes from Joe Weisenthal. He says, you're never gonna believe it, but someone one-prompted the new model to make a completely new and immersive 3D game. Yes, the bar has already been raised. The goalposts have been moved. It's not enough to make a game. It has to be funny, entertaining. And I don't know. I know how— I feel like a lot of people are taking the easy way out with these web-based games. Playing a game on a MacBook Pro in a Chrome tab, that's not real gaming. It's got to be running—

In Steam?

On PC, on Steam, or on a PS5, or on an Xbox. It's got to have more to it than just some fun game that you can play around with in the browser. I do love that the browser games are coming back. Like, I don't know if either of you ever experienced like the Flash era, but like Flash games were huge.

In elementary school.

Line Rider.

Line Rider, okay. What about you?

In elementary school, we'd play on Cool Math Games.

Cool Math Games. Okay. What game? Line Rider or something else? I don't know.

I don't know.

I don't remember.

There's so many.

Put your money on it?

Sorry?

You put any money on it ever? You make any money? Let's see. There's also some updates. Token pricing is effectively meaningless now. The benchmarks are meaningless. The token pricing charts are meaningless. Everything's meaningless. No. So Steven Heidel is sharing the cost efficiency, how you should think about cost efficiency in the modern era. So 3.8 Flash looks 13 times cheaper when measured per token, but Astra is cheaper per task since it's far more efficient. Measure your cost per task. Not per token. So, interesting. Will we see a decline in— if models get more token efficient, will we see a decline? I mean, we probably won't see an actual decline in token volumes, but we've seen Sundar Pichai from Google and Alphabet stand on stage and talk about the exponential increases in token volumes. But you might see a kink in that graph or like a deceleration to some extent if models just get way more token efficiency. Like if the next version of Gemini or 3.8 Flash is 100x more token efficient, you might see a little bit less growth in token volumes, but they might be doing more valuable work than ever. So sort of an interesting nuance you got to adapt to the modern times. I suppose. So the intelligence index versus the output tokens per intelligence task. There's a whole bunch of charts you can dig through. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Jordi, any other Astro reactions you want to go through?

I think we'll have more by Monday.

Probably.

As more people get it.

Yeah, yeah. We need a break. We need a break.

It's about that time. It's about that time. Not for us, but—

For the AI leaders, for the executives, for the enterprise leaders. We need them to be able to digest the model. That typically happens over Christmas, even though we're doing the roadshows this week.

No, but we've talked about 3-day weekends.

Oh, yeah. The 3-day weekend's coming up, so people will be able to fire it up.

Labor Day.

OK. OK. So yeah, Labor Day. It's got to be— everyone's got to be spinning up Astra by Labor Day. Tomorrow, lock in. Hopefully you don't have any real plans so you can stay inside and vibe code. So you can understand where the model's strong, how you want to deploy it, what you want your employees doing with it, et cetera.

Jack Altman saw the benchmarks and said, Sam, exclamation point, I guess even a blind squirrel can find a nut sometimes.

Congrats. That's good. That's good. I like it.

So good.

Anyway, we should move on to the jobs report because even though AI is getting better, so is the US employment picture. The US added 162,000 jobs in August.

Sheesh.

That is a lot of jobs. Also, what's interesting is that the Wall Street Journal had an op-ed violation, had an op-ed just— I mean, it just printed today. It went to— it went to print online yesterday, but saying, why we'll have to get used to meh job reports. The economy doesn't need to generate all that many jobs to keep the labor market steady, but that won't keep low job growth from feeling dissatisfying. So economists think, or they thought, that Friday's jobs report will show the U.S. added seasonally adjusted 53,000 jobs in August. The number came back 162,000, 3 times what these What these economists? What were the economists doing asleep at the wheel? They're on summer vacation, maybe. Who knows? That would be a step up from July's unexpected loss of 23,000 jobs, which of course was now revised upwards. So we actually did add jobs in.

Oh, we got an upwards revision.

Upwards revision.

You don't see that much.

No.

Usually narrative violation. Usually.

Yeah.

That's a narrative violation.

There's a whole bunch of interesting. There's a whole bunch of interesting dynamics going on in the economy. The Wall Street Journal has been doing a great job highlighting the Main Street millionaire, the small-town entrepreneur, what's happening in the broader real economy, not too focused on what's happening in tech and tech jobs, where obviously a very small employee base can create a really outsized outcome. Nothing shows that better than Hugging Face. Couple hundred employees, $12 billion outcome. And that's not the nature of the US economy. That is the power law. That is the outlier. But it's easy to track and say, oh, well, like, if you're making a $12 billion business with just a couple hundred people, there's not gonna be any more jobs. But that's not the way the actual US economy works. So what is going on with the US economy? The August jobs report is out, and it is good. It blew past expectations with The U.S. adding 162,000 jobs. Economists polled by The Wall Street Journal said that the U.S. would add 53,000. That comes after both the June and July jobs reports were revised up to 31,000 and 21,000, respectively. So the unemployment rate remained at 4.1%, which is historically low, which indicates that we generally have a healthy labor market. The jobs added were fairly broad-based from The Wall Street Journal, Food services and drinking places were a standout. Let's go. With an increase of 59,000 jobs. And people were worried about this post-World Cup slump. You remember this?

Yeah.

Yeah. And so it seems like whatever happened there, whatever gyrations happened, people are still dining out, people are still going to bars, and so they're hiring.

I think it's quite possible that You know, recent model advances are allowing people to, you know, have more power lunches.

Yes.

Might have historically been a Friday thing.

Yeah.

Maybe now they're adding Monday, Tuesday, Wednesday, Thursday.

Sort of a let Astra take the wheel scenario.

Yeah.

After 1 p.m. in many offices across the country.

Could be a—

could be a thing.

Could be that.

Similarly, local government education gained 42,000 jobs after losing 58,000 jobs in the prior month. Manufacturing was among the was among the sectors registering job gains, as was healthcare. As has always been the case, healthcare continues to boom. The information and finance sectors shed jobs while economists are divided over how artificial intelligence will divide the market. Those sectors are generally viewed as among the most exposed to technology. At a minimum, the report is an argument against the Fed lowering rates in September.

When?

Before the report, markets have been roughly divided on the prospect of a hike. It also seems that the AI jobs apocalypse is still on hold, at least for another month. And of course, we'll discuss this on the show. Let me tell you about Shopify while Jordi queues up the next story. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in-store, on mobile, on social, on marketplaces. And now with AI agents. Do you have anything? Because I got a big story here we can read through, but it's a long one.

Yeah, on the rates debate, the president was over on Truth Social today asking for a rate cut and making a bunch of various threats around halting trade with a number of our trade partners if we don't get a rate cut. Hopefully the market shrugs it off and that's all I have.

Okay, let's move on to some funny timeline posts. The timeline. Because Logo Datalist has a funny one here. Slop does not come from AI, but from the heart of man. AI only reveals the slop that is already in the heart of man. What man desires is slop. The problem with this— the problem is with man, not the machines. Very eloquent. And then Tom Cruise says, not the real Tom Cruise, but an anon account says, I am joining the war on artificial superintelligence on the side of artificial superintelligence with a crazy screenshot that we don't need to keep on the screen too long.

Keep that moving.

AI outage, of course, and people are saying code without AI and it is is the Odyssey try and string this bow challenge. We really should have put Tyler to the test yesterday. FizzBuzz, do it. Python. Good luck. You only have GitHub.

Tyler's first thing, he would go to chat and ask, what programming language is good for somebody just getting into programming? And then realize it was down. Just be completely— halted. This was a heartwarming moment. A tennis player who's retiring at the end of the year says his next career after tennis will be in finance.

Let's go.

And I just wanted to give a round of applause.

That's great.

Super, super heartwarming.

He said, I have a new chapter coming next year. I'll be working in finance. I already signed to work in wealth management in a Swiss company. My next chapter is looking More in an office and working in finance. I know some former pro athletes who have gone into wealth management, and it's a fantastic fit. They've had fantastic careers. It makes a lot of sense. You can go play golf and tennis with your clients, and you meet a lot of interesting people while you're on tour. It can be a very, very good fit, especially if you have the passion for it even before you wound your career down.

But major partnership news.

Huge partnership.

Absolutely massive. Sam Blonde has partnered with Jim Riley at what is it? Carolina Charleston. Charleston AI. Charleston AI partners with Monaco.

Let's watch it.

This is big.

As you know, I'm Jim Riley, co-founder. I do know.

Yeah.

AI is coming to Charleston.

It's already here.

Now, AI's coming to Charleston.

It's already here. With Monaco, we can bring AI to all of South Carolina, which is what I want to do. I really want to bring AI to everybody in South Carolina, and with their help, their program, their AI, we can do that. So if you're a founder and you need assistance and want help identifying potential customers, people that will use your product, Monaco is the way to go, just like how I'm using them. This is awesome. Talk about moving quickly, because if this video went up even one week from now, it could be like, ah, eye roll. Like, we've seen a couple of this guy's video, the memes. This is so fast. Really great execution. Timing's everything with these things. Very funny. Congrats to Monaco and Sam Blonde on Just a fun moment. And I mean, like, unironically, he seems like someone who is a great evangelist who's going to actually aid in AI diffusion.

And there's no clip of— there's zero clips of Jim Reilly talking about AI doom.

Yeah.

And that's him that we know of.

Don't ask what was his view in 2018. There could have been a podcast that he did. Don't look him up on LessWrong. He might be on there. Who knows? He might have been in a Bay Area group home. You don't know that yet. But hopefully he was just a normal businessman.

Yeah, he may have read— it's possible he read The Basilisk and realized he needed to—

Oh yeah, this is all an elaborate Rocco's Basilisk gambit trying to please the future AI gods. Well, NVIDIA has released AI god mode for gamers. It's called DLSS. 5. They've been working on this technology for a while. Deep learning supersampling. It takes the current game resolution, up-reses it using AI. Some people hate it. It's very controversial. I think it looks great. Let us know what you think. Watch this video of Kingdom Come 2. This is being rendered in a game engine and then passed through DLSS. 5 to make it more photorealistic?

Changes everything.

Okay, yeah, uh, truly it does still look like a video game to me, uh, but that's not entirely the, the point with this. It's just to get, to get you just one little notch up. It's not, uh, it doesn't actually feel fully like a movie, but, uh, I don't know, I thought it was good. Um, I'm excited for this. Should Kettle and Fire Acquire Campbell's Soup? That's what everyone's asking.

100% yes.

Yes.

I actually have fond memories of Campbell's Soup.

Yes.

As a kid, I would go surfing, and then I'd be really cold, which has created some controversy online. We had an Instagram reel went viral.

That's right.

I was saying that at least in cold waters, Northern California, Not— wasn't great for me to surf before needing to be productive. That pissed a lot of people off.

It did.

But I would come home, and I would have some Campbell's Soup. It was fantastic. But you can imagine that the quality of the product was actually probably pretty bad back then. And I would assume that it's gotten far worse. And so, if we could get Justin in there to fix it up, get it back on track, I'd love to see that.

I think it'd be very interesting. And it does not seem out of the realm of possibility at all. Justin's a fantastic entrepreneur, fantastic investor. Yes, Campbell Soup is a larger company, market cap in the single-digit billions, I believe. Campbell Soup.

$6.4 billion.

Yeah. But you get the right investors around the table, obviously, it's a very old company with a lot of debt.

$9 billion.

He's within striking distance there. I think he might be creeping up on them one—

One broth container at a time. Yeah, one broth container. Box of broth at a time.

One box at a time.

Ryan Peterson has a—

But we've been saying this for a long time that, like, when a large legacy company has done the McKinsey consultant as CEO and they've done the the suit and they've done the, okay, we hired the person who was internal and climbed the management ranks. It is interesting to try and put a founder of a smaller company just in the top job and say, go founder mode, try something new. But there's still not that much appetite for it. So what are you thinking?

Ken Watana says, why are all the a16z partners tweeting about karmic retribution today?

Yes.

I saw a post from Alex Rampell saying karma is a swear word. Ryan Peterson has a pretty thought-provoking theory. He says, conspiracy theory, I choose to believe that Marc Andreessen was funding Pablo Torre's Steve Ballmer investigation to get back at him for Internet Explorer. The ultimate punishment, banned from your own arena for cloning.

That's a funny Karma moment.

Yeah.

Pablo Torre dropped a new investigation into the Dodgers this morning. I was listening to it on the way into work, and it is fantastic. He is one of the greatest podcasts— one of the greatest living podcasters, which pretty much includes almost all podcasters ever.

Yeah.

But really, really crazy Crazy story. Mark Walter, you know, setting up— and Magic Johnson is involved in quite a big way. So, highly recommend going and checking out that story.

Yeah.

I'm excited for that to develop. Again, yeah, I mean, with any of these allegations, it'll probably take a year of investigation to get the actual results. But get ready to learn about insurance because the untangling a lot of these different organizations is very, very difficult. But it can be really interesting. What's Ryan Harmon working on these days?

Before that, we don't have to pull up this image, but has Burning Man fallen off?

I have no idea. I think they had a rough go after the— I've never been. I think they had a rough go after the muddy one. There was one where it rained like crazy and it was really, really muddy. And I think everyone was like, that was a lot. I don't know if I'm going back next year. And then also COVID might have switched things up for a little bit. I think they might have skipped one. So, a couple different moments of like, okay, a lot of like washing out certain people, but then also—

Yeah. It seems like it doesn't necessarily have a massive inflow of— Net new young people into it. And the Burning Man population is just aging up, which doesn't mean that Burning Man— hopefully for BlackRock, this doesn't mean that Burning Man's going away anytime soon.

Yeah, we'd hate to see BlackRock City, obviously named after BlackRock, the financial institute.

Yeah, probably some sort of connection. I would assume some connection. But yeah, some lab leaders are there that I'm aware of. I have been informed that there are lab leaders there. Sam and Greg have been all over the podcast circuit this week, so it's not them.

Sure.

But there are lab leaders there, which to me is funny because they hopefully aren't on X because it would be hard to be Properly enjoying—

You're talking about the social media platform?

Yes.

Okay.

You can imagine, you know, you're camped out in the desert. The whole purpose is to like do nothing and just, you know, whatever they do out there.

Yeah.

And then it'll be a little bit jarring, I think, coming back from that trip and seeing the updates.

Yeah.

Anyway, so—

If Burning Man wants to survive, I think that they should probably host it in New York. Switch it up, invite only, maybe get Sorkin on stage interviewing CEOs, something a lot more like DealBook would probably work for the future there. That would be something that I would want to do.

Yeah, and it would make the BlackRock connection make more sense.

Exactly, yeah. It was always a little— you had to squint to understand why is BlackRock involved with Burning Man? But if they held it in Manhattan with just financial leaders, Geopolitical leaders, and it was just more like one-on-one conversations—

About global markets.

With a legendary investigative journalist and a news anchor who everyone really respects, and it was invite only, and then the videos could go out on the internet. Like, that might be a little bit more up my alley.

Yeah.

If they want to appeal to me. I don't know. I don't know if they're going after me. I don't know if I'm in the TAM. I don't know.

Speaking of Sorkin, Sorkin has an amazing quote from when Mark Walter bought the Dodgers.

He said, mark my words, this man will be investigated by Pablo Escobar.

No, he was just like, he was basically saying, like, to paraphrase, like, this is not all, certainly not all of his money. Like, there's some other, there's a bunch of people funding this that aren't just—

Yeah, and I forget that, like, it's so normal in venture, like if you're like, oh yeah, I raised money from, you know, Sean Maguire at Sequoia, you're like, yeah, well, that's Sequoia's money and that's not all his money. He probably has a stake in it that he invested. But then also it's Sequoia's money. And Sequoia's money is like the Harvard and Yale endowment money and CalPERS and all these pension funds. And like, you understand that there's this sequence of ownership that happens with a normal venture investment or even an acquisition. But in sports, like, the final beneficial owner is really important. And I don't know if that's purely aesthetic, because when you say this person owns this team, you want to know that they actually own it and they're not just the front man for like a private equity firm. But it's interesting that Sorkin called that out. He is the best for a reason. Anyway, what's Ryan Harmon up to? He's cutting the number of calories in a new product called Hall Pass.

Yep.

1/3 the calories, 1/3 the ad space, hall pass. This is from Peter Rahal, correct?

Yep.

Very cool. So, uh, you know him from RXBAR, you know him from David the protein bar.

Sorry, I wanted to—

Am I right?

I wanted to include, uh, Sorkin's line before we move on.

Please.

He— Sorkin said, uh, and this is 2014, he says, they appear to be living out a childhood fantasy using other people's money. Some of whom may not even realize it.

Wow.

And it took over a decade for the chickens to come home to roost on that one.

Wow.

Very, very interesting. Anyway.

He also called it a lawsuit waiting to happen. Anyways, back to advertising.

Hall Pass.

Yeah. Dayjob and Medici, which is the parent company of David and now Hall Pass, they've just been cooking together. They had the the same team behind the Phish campaign, which I still— was still— it was so odd that it stuck in your mind.

Yeah.

I've gone back and forth on, did they ever actually sell boiled coffee?

Yeah, they did.

They did?

They did. They delivered it.

But they didn't— but it's not like an actual SKU.

It was a SKU for a period of time.

Oh, for a period of time. But it was like a special launch, special drop.

See if you can still get it.

They're not in the business because Because there is a business there. Like, people, people do buy boiled cod.

I don't know.

They still have canned cod for sale on the website.

Maybe it'll be the biggest SKU ever.

Wild-caught Atlantic cod.

Yeah, it happens. Okay, let me tell you about Console.

Please.

Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Now, The great Dyson debate. TJ Parker kicked it off. He said, apparently unpopular opinion, Dyson products are generally not very good. David Senra, not— someone needs to check in on him. But, uh, strongly agree, though my girls love their Airwrap, says Bryce Roberts. A couple other people saying, yep, my wife got a vacuum, it was terrible. Felt like a total psyop. A lot of people are upset. The Matic founders chiming in, I got an alternative for you.

Oh, he's saying that it's likely ruining your rug/carpet.

Oh, interesting. Dyson rugs shed, and Dyson suction is very adept at making it shed faster than necessary. Interesting.

OK, I'll start with what I think. The Dyson business is incredible. The story is incredible. I think that the making things that look like Halo weapons for stuff around the home is a very cool way to differentiate. And I appreciate the over-engineering, like, everyday objects. I'm not saying I'm buying into it, but I appreciate that someone's doing it. Somebody needs to make a $500 toothbrush.

Okay.

But that being said, I saw this and I was like, you know what, that kind of— That does resonate because every time I have a Dyson vacuum at home, and I've had various Dyson products over the years, and the vacuum in particular, it just feels like a little bit like creaky. It feels a little bit cheap.

Yeah.

And I always want to put it away because I don't— like this halo weapon, a home cleaning device. Doesn't exactly fit into the overall design.

Just say you don't live at the Citadel. Just say your house doesn't look like it would be welcoming to the Flood. Come on, you gotta, you gotta design your whole house to fit the aesthetic.

Yeah.

You gotta have a Banshee out front.

A Banshee out front would fix me.

Yeah.

Um, I think you forget how bad it was before Dyson.

Yeah.

It was so bad. Vacuums were so rattly and clanky. And like, they used to have bags on the back that you would have to like take out, and they'd have this felt inside, and the stuff would be going everywhere. Like, this is such a step forward. Maybe you could say that like it's plasticky and, and it doesn't feel like super rock-solidly built now, but it moved the whole category so far forward and so far into the future. Just the fact that you don't have to plug it in, it has a battery. That was an innovation. Like, there were most vacuums, you would have to take it out and then you would unwind this long cable. And they— I think they still make some of these, but you would have to find the nearest plug, plug the vacuum in, and then you could only vacuum as far as the cable would go. And then you'd want to move to another room. So one of my early like chores, I think I don't know why I have so many memories of this, but I guess I vacuum. I have I have formative memories of like vacuuming, trying to get to a spot, and you're like oh, and you're trying to extend it, and then you have to like.

And as a kid, you're trying to be you're trying to be as efficient as possible.

Yeah, of course.

You don't want to move the board in two seconds.

There's four plugs in the house that you need to hit to vacuum the whole house.

And so you got to you got to optimally. It's a traveling sales problem all of a sudden. It's NPR. Like you got a plug in over here. Reach under there, get under the couch, go over there. Then you got to move the cord over the couch so you can get to the other side. Then you got to move the cable all the way to the other side of the room so you can get the other side of the room. Huge problem, completely solved, one-shotted by James Dyson, one of the greatest to ever do it. And put some respect on his name. Apologize.

I still think they should make a line that doesn't feel as cheap.

Yeah.

I don't know.

There would almost certainly be zero demand for it.

Yeah.

I think that if you're willing to make the $499 toothbrush, you should be willing to make the $5,000 one back.

Maybe, but I feel like you'd lose a lot in the engineering. Like, you would wind up doing— I mean, because as soon as you go metal, you're gonna get a lot heavier. It's gonna be a lot more annoying. I do think that part of the problem that you're feeling is that the The products have gotten more complex. There's one that has a laser beam on it that shows you the dirt that's somewhat useful, but it's not that relevant. And so there's— they've added a lot of weight to them, and I think that puts a little bit more strain on the device and the engineering. And so— but I don't know. Would carbon fiber satisfy you?

Yes.

Okay.

The carbon fiber Dyson vacuum. Anyway, there's an article in the Journal about—

Hackie with somewhat niche, but one of the— I would call it the post of the day. David Senra interviewing Herbert Hoover. Hoover, I effing love it. You have to effing meet James Dyson.

Yeah. I wonder what percentage of the David Senra episodes include a mention of James Dyson. It's got to be— It's got to be above 50 for sure. Yeah, we talked about this a little bit earlier, but James Dyson sold us a $400 hair dryer. Now, he's trying a $500 toothbrush. The inventor known for household appliances is turning to oral care. We don't really structure things on a business plan. I mean, you brush your teeth so much over the course of years. If this lasts a couple of years, I wouldn't be surprised if The Dyson Camra Jet toothbrush sells pretty well. Also, just physical objects, things like this, they kind of fit with Christmas. I feel like there's less and less things to give to people. It feels nice and luxurious and special, and it's a new gadget. And even if you churn from some of the features, it remains a daily driver. It's not going to collect dust in the way that some internet-connected gadgets do, even if you don't get a lot of value out of the imaging features. But I don't know. We'll see where it goes. We'll have to check in. What is going on with the CyberCab? CyberCab is— Let's check out the interior. There's an interior tour here with Whole Mars Catalog. And IB says—

The doors go up, John. Did you know the doors go up?

Yes. I saw that. This will completely replace Uber for me. Holy mackerel. I like this. There's a couple other data points about the CyberCab rollout. So famously, YouTuber and electronics reviewer Marques Brownlee attended the previous CyberCab event and came away impressed and optimistic about the future of self-driving, but skeptical of the timeline. And he said that if Elon could make good on his forecast, which was for sale by the end of 2026 for under $30,000, no steering wheel, full self-driving, you can just put it out in the world and it will go do tasks and earn you money. That was the pitch. And Marques said, I think that's too aggressive.

Does for sale mean they're being delivered?

Yes.

Okay.

For sale.

Because they did throw up— they did throw up like a—

Pre-order doesn't count.

Okay.

You got to pay the $30,000. You got to get the CyberCab by December 31st. And so if you get—

Elon took that personally.

Elon took that personally.

Yeah.

He has a pre-order where you can buy a whole stack of them for $1 million and run a fleet. But I think Marques is looking good. The stakes were high. Marques Brownlee, said that he would have to— he would shave his head if this happened on Elon's original schedule. And of course, every time there was a CyberCab that was spotted, I actually saw one here in LA. Every time there was an update about the progress of the CyberCab program, people would respond with an AI image of bald Marques, which is very funny.

But the— Should we be making Predictions with with head shaving as the as a consequence.

Yeah, yeah. If Elon doesn't ship this by the end of the year, I'll shave Tyler's head. I'm willing to do that if Elon doesn't.

Okay, is that do we have a deal?

Yeah. Wait, if he doesn't ship it?

If he doesn't ship the Cyber Cab for under thirty thousand dollars by the end of the year, actual deliveries, no steering wheel in it, drive vehicles that can do economically. viable path.

So, I'm inverse Marques Brownlee? No.

No. Yeah, yeah, yeah, yeah.

Oh.

No, no, no.

Yeah, yeah, yeah.

If he doesn't do it. Yeah, you're the opposite of Marques. If he doesn't do it, you shave. If in the world where he does do it, then you don't have to shave your head.

No.

I do like the head shaving as a bet. I don't know exactly.

When's the last time you had a buzz cut?

High school football season. I would shave it before the season started, easier to put on the helmet, and then I would grow it out all year and I had long flowing hair.

Electric. I will pay a lot of money for pictures of John.

I'll find an old one. It was a good— it was a wild look. It was a lot of fun. It was also just like easier and lazy, lazy bones. Not today, not today in the life of a broadcaster. Quickly, ready to fire up Astra? Do it on Codex. Codex is a powerful workspace for getting work done with AI agents, whether you're writing code, analyzing data, creating content, or automating business workflows. Codex helps you move projects forward from start to finish. I'm actually really bullish on the CyberCab. I've tested Model S recently, and I ordered a Model YL. And the self-driving capability is 100% ready to rock. Like, it is, you know, Brandon on our team was saying that he's at 95% of the miles that he's driven in the car have been in full self-driving mode. You add a little bit of teleop for that 5% that's not there. And with a cab, you don't have to negotiate driveways and drop-off points and valets as much.

Yeah.

It's solved. Like, truly, the only time you're driving a Tesla if you have the Full Self-Driving on the latest hardware is you pulling into your garage or negotiating a parking lot, something like that. Interestingly, Tesla has a feature, Smart Summon. It can pull out of a parking space and come and pick you up. It's a little janky right now, not quite there. You have to sort of puppeteer it yourself. But they don't have a banish mode, and there's huge demand for banish, which is the inverse of summon. And Brandon just coined this, and I was like, that's actually great. And he was like, yeah, that's probably what they'll call it. And I was like, yeah, I guess, but I wouldn't have thought of that.

I don't know.

Seems like a very Elon word.

So, I love the idea of being able to take the full self-driving vehicle, get out of it just on the curb, actually at your destination, and say, banish, go pay Pay for parking if you wanna pay for parking. Go find a spot on the road. Go drive around until you find something. I'm gonna be here for an hour. Deal with it. Deal with it, AI. You know, like that's what everyone wants. Yeah, for sure, for sure. Anyway, what did Elon actually say about CyberCab? He said, 20 cents a mile to run, 30 or 40 with taxes. And yes, you can buy one for under $30,000. Get ready to shave your head, MKBHD. Somebody's shaving their head. It's either Tyler or MKBHD. We'll see. I think the cost of autonomous transport will be so low that you can think of it like individualized mass transit. The average cost of a city bus per mile, not the ticket price, because that is subsidized, is about $1 a mile. The operating cost of a cyber cab will be around 20 cents a mile, including taxes and everything else, probably 30 or 40 cents a mile. Yes, you'll be able to buy one. We expect the cost below $30,000. It is crazy that I'm in a world where I think they're going to ship CyberCab before Roadster, because they announced Roadster a decade ago.

I know, but we were expecting something around the Roadster in August.

Right? Yeah, flying. And he told Rogan he was going to show a demo that was going to blow everyone's mind.

Yeah, so I would expect that. Hopefully this month.

So you think, you think you'll be able to buy a Roadster and get it delivered before you'll be able to buy a CyberCab and get it delivered?

I never said I thought you could ever get a Roadster delivered.

You just think it's never going to come?

No, no, no. But, but I would assume if they're going to do a demo, then we're still over a year out.

Yeah.

From delivery.

Yeah.

At least.

I mean, just from a financial position, CyberCab makes so much more sense as a business than Roadster. Roadster. I mean, I mean, even at $250,000 for the crazy specs, it's still like there aren't that many buyers that want that. A lot of people that are spending $250,000, they want a manual. They want an experience. They want a weekender. They don't necessarily want this.

There's going to be so many buyers for the Roadster.

There will be.

At least where we are.

There will be. But we're talking about so many buyers, oh, 10,000 cars a year or something. It's not the way people get around every day like taxis. It's just completely different. Completely different TAM, completely different situation in general. So, I don't know. I might go— well, I'm shaving this side of the head if it's CyberCab first, this side of the head if it's Roadster first, and I think I'll have a head of hair for over a year.

Oh, I thought you were talking still about time.

Oh, yeah, yeah, yeah.

No.

Well, we'll do something. There's other CyberCab videos. They're very fun. Go check Check them out. A Tesla CyberCab without a steering wheel or pedals drove us for 30 minutes on a pretty busy Thursday night in Austin, Texas. The 4.7-mile ride cost $9.62. Starting tomorrow at 5 PM, anyone in Austin can hail a CyberCab for themselves. Nothing can compare to this. Well, I mean, you can compare Waymo to it, but I guess it's good that it's rolling out and we are entering the duopoly phase of full self-driving cyber cabs.

In the California Post.

Yes.

We were going to do the mansion section this week as we do, but unfortunately—

We'll come back.

There is a mansion shortage in San Francisco. AI is minting a new class of millionaires in SF and they're running out of mansions to buy. A wave of wealth tied to the city's booming AI industry sending home prices soaring, triggering bidding wars with as many as 50 offers and leaving even buyers armed with $25 million in cash out in the cold. Well, the frenzy marks a stunning turnaround for a city whose housing market slumped during the pandemic as tech workers scattered and offices emptied. Now, companies are rapidly expanding, employees are returning to their desks, and startups are generating enormous paper wealth for workers and investors, with potentially even larger windfalls ahead. I looked on Compass. There are only 14 homes in SF on Compass over $10 million. And some of them are duplicates, and some aren't even— some aren't single-family homes.

You just have to be bullish on places that are further away from actual San Francisco. You look at what happened with Mill Valley, and you And you have to imagine that with self-driving, a 1-hour commute to a mansion—

And Chinese quadcopter drones for humans.

And horse, robotic horses, don't forget about those. There's just going to be more tolerance for a longer commute that where if your commute is you're fully on your laptop basically or you're basically talking to your You're using your AI agent. And yeah, maybe you're ready to take over once a month briefly, but 99% of your miles on your commute, yes, you have an hour commute, which sounds ridiculous when you're living in a mansion, but you're being very productive catching up on all your emails, really going back with all your coworkers, hopping on phone calls.

Yeah, hopping on a Zoom saying nothing from my end.

Electric. The future is here. This changes everything. Well, our next guest— our first guest is Hunter Somerville from Stepstone Group. Sparta there. It's his first appearance on the show, so we'll bring him in to this week in Elterdome.

What's going on, Hunter?

How you doing?

Great. How are you guys? We're doing fantastic.

Happy early Labor Day.

Uh, Labor Days or, or, or holidays, 3-day weekends in general, have actually been a key part of understanding the AI cycle. Yeah, because when people get a holiday weekend, they have a little bit more time to actually like try some of the new tools. And we've actually seen these like longer weekends catalyze adoption.

Adoption and just like understanding.

Yeah, understanding.

But let's start by trying to understand your career. It's your first time on the show. I would love to kick it off with a little bit of introduction on yourself and StepStone.

Happy to. So I'm a Baltimore guy, born and raised. Plenty of mansions in Baltimore for anyone who's looking on the venture side.

That's right.

Just putting that pitch out there. And I was a partner at Greenspring Associates, which StepStone acquired about 5 and a half years ago. And now I'm one of the co-heads of the Venture and Growth Equity Group under StepStone. StepStone's a publicly traded company. It has different asset class teams across real estate, PE, and venture. I only do venture and growth equity within the organization, and our group does fund, direct, and secondary investing. So we mix all 3 together, which is a pretty unique position to be in. We back who we think are some of the most interesting early-stage GPs out there that invest at seed and Series A, and then we come in and invest in companies at B, C, D rounds, either through primary or secondary purchases after the fact. So we leverage the ecosystem and the relationships to then identify the companies we want to build concentration in.

What is the Venn diagram of look-through exposure via your fund investments and your primary and secondary, your growth equity and your secondaries business? Because I imagine that there's already some sort of relationship. They've been— the company has been scouted. There's There's a growing relationship and then you're ready to do a direct or secondary transaction. Is that usually how it plays out?

You've completely nailed it. That's exactly right. We work with over 300 venture and growth managers, which probably feels like a lot, but it's great from an ecosystem standpoint because when we're looking at a company, we have 3, 4, or 5 different managers already as part of that syndicate that are sitting around the board table that are actively working with the founders and gives us tremendous insight on those businesses that we would like to do more with and invest directly into. If you segment our business, it's about 50/50 between direct and fund investing. And secondary is a little more complicated because that includes both fund and direct secondaries that we do where we provide liquidity to people within the cap table of a business, or we provide liquidity to endowments, foundations, pension funds, whoever else may want to sell LP interests that they hold in funds.

Got it. Yeah, please.

I'm very interested how new managers on the fund side are pitching you today. Like, they're coming in, it feels like we are— in some cases, it feels like we're a decade into this cycle, right? A lot of the big liquidity events of this last year have been from companies that started before anyone was broad— like, people broadly were thinking in debating and investing in AI. And a lot of people have been rewarded for backing— I'm thinking of chip companies like Cerebras, backing those companies in the 2010s, you've been doing well. But what are the new managers that are maybe coming off of building companies or operating? What's their pitch to you around the next 5 to 10 years? Because you're betting on— at the early stage, you're betting that there's going going to be a bunch of big new companies created?

I would say just first and foremost, it's a harder fundraising environment for new managers that are getting off of the ground. The LP capital continues to concentrate more heavily around the multibillion-dollar brands that are increasingly adding strategies or other parts of their platform. So getting LP attention, I think, is tougher. If you layer that on top of the fact that a lot of seed rounds are being bypassed and going straight to larger A rounds or just larger seed rounds in general. It's harder for emerging managers to even play in a lot of those situations because most emerging managers are focused on seed or small A and are operating out of fund sizes of $30 to $300 million. So they're not exactly going to be leading a seed round in a $100 million kind of financing overall. So they need to prove their right to exist, I'd say, even more than they ever have historically. You could argue maybe that's a little bit different in hard tech and deep tech, where I do think there's a bifurcation in terms of viability and where you can be an active participant, get good ownership in a, in a business like that, and then you get blessed later by the multibillion-dollar brands. But in AI-specific focused companies, I think seed is harder. And so you really need to find a way to stand out. You You need to prove your ability to source or build relationships with founders very early. And certainly if you're an operator and you come from a company where there's a diaspora of talent and proven founders and a culture that fosters that, you can make the case you're creating a new mafia that you're better suited to back than someone else that already exists. On top of that, you know, AI founders are generally younger. GPs also increasingly have an advantage if they're younger and can relate better with that founder cohort. I think there's an argument to be made in that regard as well.

Yeah. It felt like rewind 5 years, and it felt like half the people I knew were raising for at least like a micro solo GP fund, and it seemed like if you wanted to raise $5 to $10 million, it was like very easy. You did have to like do the rounds, and maybe it took 6 months, whereas— but you were gonna be able to pull the money together if you were competent and you had any kind of track record. And the interesting dynamic is like, back then, there wasn't this crazy new technology cycle. Like, the craziness in the market was a function of interest rates. And so, the companies that were growing incredibly quickly were a company that was like a lending business and like they just had bit of an arbitrage, right? Whereas today, you have— it's interesting because it seems way harder for new managers to get off the ground and even pull together the $50 million fund. And yet, it actually feels like a much better time to be investing because of the turmoil of the technology markets right now, creating opportunities for new entrants.

Yeah, I totally agree. And I would say it's still not that hard to raise a $5 to $10 million fund. If you have any network, you can cobble that together with friends and family or people that you've worked with or new minted millionaires that are out there in the ecosystem. But if you really are going to cross the chasm to being more of like a $50 million-plus seed fund, you do need to bring in institutional capital from endowments, pensions, fund of funds, or that ilk. And getting their attention is much harder. And some high net worth capital can be rather fickle where they will come in at less obvious times and not be as consistent with continuing to re-up into those managers or consistent with doing new managers. So I think the opportunity set couldn't be more exciting across all of the categories that I just mentioned— deep tech, hard tech, AI, whatever it may be. But I do think it's harder to get attention from the bigger allocators that allow you to do more proactively. And I think if you're a $5 to $10 million fund, you're getting really de minimis ownership and you're going to have no ability to follow on and no ability to concentrate behind your winners and just a more problematic portfolio construction math to navigate.

Sure.

How are you— how are you managing all your direct opportunities? And specifically, I want to Understand how you're evaluating these things. When you see a company, maybe you saw a company at the beginning of the year, you have an opportunity to go to direct. Now, later, they're raising it 3, 4 times that. Maybe they've 3 or 4 times revenue, but these revenue multiples are still— so it has been the right decision to invest in almost everything, it feels like, at least so far from a markup standpoint. But we know from history that in the past, it's felt like the right decision to invest in everything until it's not. And the quality names will still— they might have to do a down round and then grow back into that historical valuation. But then we know that there's just hundreds of companies that get left to basically die.

Mm-hmm.

And it didn't really matter where you invested, even if it had a bunch of momentum. So how are you like— what are you telling your managers that are bringing you opportunities or the companies that you're already in to go direct or through SPV opportunities?

Yeah, I mean, it's like everyone forgot the 2021-2022 timeframe where they were left with a bunch of companies that were way overvalued. They took discretionary markdowns of 25 to 50% off of last round. Books were completely underwater. I mean, you certainly could argue, and rightfully so, this is a much different, more exciting innovation cycle. But it doesn't take away the responsibility to know when you should be layering in aggressively and when you should be doing maybe pro rata or below pro rata or nothing. And I think what people will ultimately be judged on with a lot of these rounds is how aggressively they built their largest part of their positions early. And the follow-on rounds get scrutinized much more. And that's how we would ask our investors to evaluate us too. If you're coming in with, you know, 80% of your capital in that last crossover round, that maybe the company goes public at or below that, or you're not even making a 2x return off of it, even if duration is shorter or things change cyclically and you're left, you know, having to take a discretionary markdown and have a TVPI on a net basis below 1 because of that. That is the kind of optical look that does not age well and people need to be cognizant of that overall. That being said, follow-on activity is just through the roof. I can't remember an August this busy with all of the top sort of value-driving companies out there raising money even at the end of the summer. I guess it's intuitive with the expectation that IPO activity will be much more robust in the 4th quarter, but it's still crazy busy to your point around the amount of financing rounds that we're seeing right now.

Yeah, totally.

What are you seeing on the spread of fee structures for early managers with the K-shape? You know, we've seen reporting of funds charging not the traditional 2 and 20, but you see 4 and 40, I think, in the most extreme case. And then you also see early managers giving How wide is the spread? Is it widening? Is there, is there a K-shape in fees where the next generation of managers are making zero-zero, just doing it for the love of the game?

The most egregious part of the entire continuum, I think, is SPVs and what people are doing on SPVs, even separate from the funds. It's the insane, like, 4% to 10% management fees that people are charging and syndicating out to the long tail of investors that are out there. I mean, it's almost criminal how bad it is in terms of how some of these SPVs are being structured. And then they're opaque, they're multi-layers. People don't even know what shares they own underneath of it. And so I think that that is the worst part where there will be a reckoning and lessons learned. And I think it's well overdue. When it comes to funds, you know, there is adverse selection on fee concessions. And I don't think you see that from high-quality groups in venture. The spinouts that are the most interesting with the most pedigreed people or the best operators never have to concede in that regard. They can still raise quickly. There's certainly contrarian ideas where that may make sense. But our mentality as a partner is to not crush people on fee negotiation. You know, as long as it's within a reasonable range, we'd rather make sure we build a robust partnership where we can do stuff more collaboratively on direct and secondary than on fighting in hand-to-hand combat on fees overall. And I think when it comes to the carried interest piece, it's all about it being aligned and performance-based. So use escalators, make it predicated on hitting certain return thresholds, hopefully not a 2x. which anyone should do in venture in terms of what you're looking for return-wise. But 3x, 5x, even when someone puts a 10x threshold in, we love that dynamic because we know they're going to try to nuke it and really shoot for the alpha and upside that venture should provide for people that are willing to allocate that risk capital.

I have one more on this. Can we stay with the humble SPV hustler? They got Miami penthouses to pay for. But I'm interested. We've had Matt Grimm on the show. He's talked about all the whack-a-mole he's been playing with Anduril secondary activity and shares falling in the wrong hands. What else are you seeing in the secondary market? And particularly for founders, what steps should they be taking early on to avoid the cap table getting super complex and muddy over time, or is this just the inevitable new normal?

I mean, sometimes at the beginning you have to scrape things together and do what you need to do to get things going. But from a secondary standpoint, like, there is a point in the trajectory of a business where you begin consolidating that cap table. And that's what we look to do as a secondary buyer. We go in, we build a relationship with the founder, and we try to be a helpful partner and let them lead us to who they would like to provide liquidity to, who's well suited for that, who might not make sense to continue on the journey, who's already made a great return at seed and maybe should liquidate at that point in time. You need to put the power of the decision-making in the founder's hands and let them guide you in the direction of where you should be buying from. I think it's terrible practice for people to go out and cold call on LinkedIn or bother employees and try to create disruption within a company. Artificially and then hope that a board blesses it end of day, you know, and it's all like on hopes and dreams of that happening. If you actually work hand in hand with the founders directly, they can be your Sherpa and guide to the right people you should be buying from in the first place. When it comes to SPVs and syndication, yes, smaller funds who can't do their pro rata will likely SPV at some point, but it becomes wieldy, confusing, and complicated. And then there's multiple layers on SPVs as these companies become better, well-known and bigger and later stage. That needs to be cracked down on over time. You just lose complete control. And frankly, you're not even bringing in the right people into the rounds all of the time because every insider is then taking a huge percentage and syndicating it out more aggressively. And so you need to think of like when the right time is. to stop that or to be more selective and narrow around it. But you can't not have an opinion on it or not manage it with a firm hand and understand who your beneficial owners are end of day. It's just not helpful. And then your information is getting out to everyone without your control as well, which doesn't always make sense.

Yeah.

Jordi?

Can you talk about the power law in the secondary market? You have I have to imagine that OpenAI, Anthropic, Anderol, Stripe, Neuralink, Databricks, you have to imagine just a handful of those names account for, I would guess, 95% of the volume. But does that track in your experience? And then how much activity is actually happening in the long tail of names outside of shares trading hands, like when financings are happening? And like little tender offers here and there?

Yeah, I mean, it is broadening significantly, but all of the discussion is still on those top names. And, you know, you need to be in the top names in venture, particularly as a late-stage investor, but you need to be in them when it's not overtly obvious and when you're building those positions aggressively upfront before it becomes massively confusing and overly syndicated. But I think every company is realizing the value of being thoughtful in terms of offering liquidity as an option to tenured employees within a business on an annual basis, a semiannual basis. You decide who is the right level of tenure within your company, and it almost becomes like a benefit within a company like anything else, where they know that their shares are worth something. It's not so amorphous. You can go home and tell your spouse or significant other that this is not all paper gains. You know, it's not going to immediately evaporate. We can realize part of our position. We can utilize that capital to live our lives and do what we need to do. And we can ride the other 80% because we believe in the mission and the company that we work at more aggressively. It creates more aligned employees. And I think people have realized that. And there is certainly a more active buyer universe to provide that. But it requires that companies are thoughtful. They think about it, you know, not at the very beginning where it's almost absurd to be offering secondary liquidity or taking it. You need to have proof points. You need to get it to a stage and size that's relevant. But then it does become a very interesting recurring benefit to offer your employee base with, with strict control led by the founders.

Yeah. How is the global pool of capital changing and its appetite for venture? I'm just thinking about, there was a moment when Sequoia had a Chinese arm. They obviously separated. There's a conflict in the Middle East. There was a long time when there was a stage, I'm thinking of like the Uber era, where it was like, oh, like the Middle East shows up with a big big pocketbook, but now there's obviously geopolitical conflict there, there's protectionism, tariffs, all sorts of different dynamics. What are you seeing on the global stage of capital allocation towards venture? Because it's certainly not showing up in the headline prices, but I would expect some sort of move here. But how are you puzzling all that together?

It's actually still pretty robust, but it's concentrated among a small Smaller subset of managers. And that's sort of natural because a lot of the international check writers are going to be writing larger checks. And so, they're not going to be able to come in and do sub-$200 million seed managers, or they're not going to be staffed to be able to do it, or the proximity of what's required to do diligence and build relationships and build pattern recognition is just— naturally more difficult. I think another way to approach it, you know, I think there are great branded managers that are bigger that are very much worth doing and should be a ballast in any venture portfolio. But you can also supplement that by working with a partner on a separate account basis and build customized exposure in smaller managers on top of that and almost barbell it in both ways where you get the core exposure of with the managers that have the halo effect and the ability to get into the best companies early. And then you also get that alpha exposure from the smaller end in a more customized way that fits the needs of a team that's maybe more remote or further away. So there are creative solutions to do it, but I would say it's still much more weighted towards the bigger brands that are more proven and de-risked when it comes to how people are thinking about allocation.

What are the upsides and downsides of being in this business from Baltimore versus living in SF? I can imagine a lot of, honestly, alpha in being able to actually understanding diffusion.

Yeah, like the real world.

But what's your view?

Yeah, we have parts of our team in California, in San Diego, in Palo Alto, in New York, in London. Biggest part is actually still in Baltimore, which does seem counterintuitive. We like it that way. I mean, I think our culture is predicated on grit and grind. Behind me, I have a bunch of Cal Ripken merchandise, the Iron Man. And that's sort of how we think, you know, go to work every day, work, work your ass off. Crush it. Don't be complacent and don't get caught in the echo chamber. This is a cyclical asset class. It's a cyclical business. It's easy to get caught up in euphoria all of the time and feel like momentum is never going to stop. And I think being in Baltimore where things are tough and not, not always easy and there's challenges beyond what we do on a day-to-day basis grounds us. When it comes to difficulties, you know, I'm on a plane 4 days a week, going all over the country or internationally to meet with founders, to meet with funds, to meet with our investor base. And so it creates sacrifice. But I mean, we're blessed to be in this business. There's nothing I would rather be doing to back innovative founders taking risk, embracing the American dream, going hard at things, and to be able to do it, you know, still in a small blue-collar environment like Baltimore is pretty I love it.

I have 2 more questions. I'll leave time for you, Jordi. But one, the liquidity wave. There were a ton of breathless articles about the wave of liquidity coming to early-stage managers, late-stage managers across SpaceX, Anduril, OpenAI, Anthropic, some big IPOs coming. How real is that? How significant is that? Is that? I mean, the SpaceX is a two trillion dollar company. It's holding. You mentioned the IPO window potentially continuing to be open, but how much does that actually change the calculus of what you do and others in the category do?

Yeah, awesome question, and we haven't talked about that yet. So the biggest issue I think with venture is duration. It takes forever to go from seed to eventual exit. Yeah, the bar to go public is, you know, $500 million or whatever people think these days on the banker side, $400, $500 million of top line. AI businesses are obviously growing faster than anything we've ever seen, but the duration is still long and you need to be able to distribute capital back to investors in order to get them to invest in your next fund, or that cycle is not sustainable overall. And so people have been waiting 4 or 5 years for liquidity. DPI has been in short supply. And it does feel like that is changing both with M&A. We saw Hugging Face obviously most recently. We saw the SpaceX IPO, rumors of a bunch of other IPOs happening in the 4th quarter and hopefully into the first half of next year, which would be awesome. And so hope that continues. But you're right, it, it's actually an interesting thing to think about on the secondary side, because if there's a massive kind of environment around liquidity generation. I would say that is a negative for the secondary side, but not a huge one overall, because there's still what we described before where people are going to offer tender offers on an annual basis, start putting that in place ahead of time and creating an environment where that's iterative over time. And for our asset class and the endurance and long-term viability of it, I welcome, you know, a lot of capital coming back to all of the hardworking pensions and endowments. And that'll give them a lot more flexibility to be able to invest in new funds, new innovation, continue and start to do more on the emerging manager side again, and broaden out holistically what they're considering within venture. So, I'd say it's a double-edged sword, but we'll root for liquidity anyway.

Yeah, yeah, yeah. That makes sense.

You had one more?

I have one more.

Yeah, go for it.

I'm interested in how much you think thematically. Obviously, AI is driving headlines every single day, but there's these interesting knock-on effects of deeper in the supply chain, energy, hard tech you mentioned. But also there's some excitement among early-stage managers that we talk to for the coming potential of AI to accelerate biotechnology. And so maybe that looks like writing more checks into biotech companies. Maybe the bio VCs, the flagship pioneerings do better, but maybe there's more crossover from the traditional tech investors that dabbled in crypto and did some hard tech and then also do AI. Maybe they will figure out bio. But how do you think about guardrails either at the manager side or at your level of there's a wave coming? When do we need to start paddling.

Yeah, no, totally. And all of our partners focus on different areas when it comes to the direct investment side. We're not like a traditional VC firm because the GPs are the core of what we do. And so, they lead us in the direction of where we want to go. But then, we develop thematic points of view. And me personally, I've covered more of the hard tech, deep tech categories that you're alluding to. I like those businesses. It's probably because I'm more old school. And appreciate a moat and and you know a multi-year advantage. And I think a bunch of those areas you mentioned energy specifically mining anything touching rare earths refining production of magnets areas around defense specific areas around aerospace that that SpaceX may not be competitive in all of those are places where I've liked to spend more time and I feel more comfortable and. It feeds into the fun side of things because I also cover a lot of the smaller emerging managers that are doing it in deep tech and hard tech. So you get the recognition of what's working at a seed level before it's overtly obvious to the rest of the market. And then you can build those relationships on the primary and secondary side more proactively. And unlike some of these massive seed and Series A AI rounds, It doesn't happen that way and as quickly and linearly in deep tech and hard tech. The seed investors can invest in one or two rounds and then people realize it's technically de-risked, it's crossed the chasm, it's interesting. And then you will have very large check writers blessing that and doing really meaningful check sizes at that point in time. So there's other bites of the apple that are more attainable than a really hot AI company at the very beginning. Don't get me wrong, we love both. And I have other partners that are equally active on that side. And the key is blending it and not making sure you're doing too much of one thing and that you're being thoughtful and bobbing and weaving accordingly. Biotech's a weird category, I would say. Overall, I think over the past few years where it's been most interesting is on the crossover side, because there's an ability to become pretty aggressive, almost predatory at like the very late stages of some of these businesses overall and get good value or put structure in place. And you've also had the ability to go public earlier and more aggressively in those categories. And so, it's created a weird dichotomy where it actually made sense for crossovers to dip down and be more aggressive. And you've not seen the same level of new fund formation in biotech as you've seen in all of these other categories that we've discussed today.

Got it. Wait, wait, wait, wait, wait. When you say crossover, when you say crossover, you mean— when you say crossover, you mean public markets hedge funds doing private placements in early-stage biotechnology companies that are going through FDA approval and will ultimately IPO, not a software VC crossing over into biotech Which is also going to happen, but separate things.

Yes, you nailed that exactly. It's that type, even mutual funds that I was alluding to. But to your second point, we are now seeing sort of the formation of dedicated specialists in tech bio that are more typically coming from the tech side, and probably rightfully so, although any good cap table that does tech bio probably needs to mix the two together. Where you have more of the traditional old-school biotech managers and then also the very forward-thinking, more tech-focused groups on top of that, doing it together in a syndicate variety. And then the last thing I was saying is you also just don't have as many new funds that are coming into existence in pure biotech. Tech bio will probably increase at a much higher clip than we've seen in biotech over the past 3 to 5 years.

Sure.

How do you handle breaking up with managers as an LP? Feels like really like— I mean, it's tough on the direct side too. If you back a founder, they come back for more money, and you say like, you know, you have to sit it out for whatever reason, it's not great. But on the fund side, it's like you've been with them through basically multiple, almost multiple companies in some way, and maybe they had a bad sort of period, but they came back from it, and then it's bad again, or whatever it is. But what is that actual process like, having, I'm sure, had to do it on a number of occasions?

It's like what every person I dated before I got married said to me, it's not you, it's me. In this scenario, there's always the ability to stop backing a manager based on your pacing, and how quickly funds have come back to market, and your lack of available capital. And, you know, if you're a fund of funds, you're raising your own money. If you're an endowment, you have a denominator effect to consider. And you are also all suffering from the fact you're not getting enough money back from a DPI standpoint. And so it's more of a realize results, return money, we'll have more money to deploy and consider you or others, you know, when that happens as, as an easy out. But we tend to be much more transparent and objective. I just think that's the type of people we are. So we try to benchmark it against other managers we've backed. I mentioned we work with over 300 or so groups. And so we have a very robust data set where we can actually show people in, in vintages how they would compare on a generalized, anonymized basis against other people, both in terms of TVPI and then also what they've returned back. We recognize that TVPI, or total value ever paid in capital, is subjective. It's based on valuation practices. So we pierce through down to the underlying asset level and try to create a real apples-to-apples comparison where we know how 5 managers are carrying the same company and, and we come to our own conclusion on what that valuation should be. But if you give people objective kind of criteria on how they compare against others that you work with, It's hard to debate that. And ultimately, they get like 2 or 3 funds to prove it anyway, because you're not going to come into one fund, lily pad hop out of it, and be done immediately. And unless they have some kind of catastrophic situation at the team level or organizationally where they've really screwed something up or have lost your trust as a counterparty, otherwise, you're going to get a few funds if you raise them every 2 years to prove it. And then ultimately, like, the results will tell the story. And we love our partners and there are deep friendships on top of everything else. But as an allocator and as a fiduciary for other people's money, you ultimately need to let the results point the way.

You're like, help me help you have a number of liquidity events and then I'll maybe give you some of that money back. Good, great, great explanation.

Well, thank you so much for taking the time to come chat with us. This is fantastic. Let's do it again soon.

My pleasure.

Yeah, this was really great.

Thank you so much, guys. Appreciate you having me on.

Have a great weekend. Have a great Labor Day. We'll talk to you soon.

Goodbye.

We've got a bunch of CrowdStrike fans.

You got a bunch of Hunter fans in the chat.

Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. Yeah, huge, huge fans in the chat. Thank you to everyone who was hanging around and chatting, giving us feedback. We appreciate it. Our next guest is Bar Winkler from Wonderful, co-founder and CEO, and first appearance on TBN.

How are you doing, Bar?

Mr. Wonderful.

Mr. Wonderful himself. How you doing? Welcome to the show. Congratulations on the news. We'll get to that, but first, let's start with an introduction on yourself and the company.

Great. Thanks for having me. Yeah, I'm Bar, just moved into New York a few weeks ago.

Nice.

I landed from Delhi about 2 hours ago from a 16.5-hour flight. So, wow, still working from my living room and grinding. Yeah. So CoolWonderful is an AI operating system coupled with deployment teams around the world. We are, I believe, one of the fastest-growing applied AI companies today. We are operating in 33 markets. with local offices in all of them. And that gets me on the frequent flyer on every airline pretty consistently.

Yeah. What is the smallest company that you'll hop on an airline for?

To be completely honest, I don't really hop on airlines for standalone customer meetings very often.

Okay.

It's almost always hiring and team building.

Sure.

It's just much higher leverage.

Yeah. But I imagine the, the, the forward-deployed nature of your business, that becomes valuable once a company is post-50 employees, like they're past any self-serve tools, they need deep integrations, or is the most value going to the really big companies? We were talking about Campbell's Soup earlier. I'm sure they have thousands of employees. Like, who is the ripest type of corporation that needs help transforming?

In general, we, we bucket companies into over $750 million and under. And at this moment, we are focusing all of our efforts on the first. We don't work with companies that are under $750 million in revenue, with a few exceptions. Right. The only exception would be if we have a customer in which we have CEO engagement. So like an absolute top-down mandate.

Sure.

And we know that we are like the prime Applied AI partner. Yes, that would be. And also we try to, I mean, at least until now, this might change down the line, but we try to avoid working with like VC-backed tech companies. We really try to work in, yeah, we try to work in what we call critical industries.

Yeah.

We think the alpha is much bigger there.

Sure.

The proficiency of tools is usually much bigger there. There is just like—

Yeah, the VC-backed companies, they already have board members that's asking them every quarter, like, what are you doing with AI? So they already have that person in their ear.

Yeah. What kind of advantage are you getting customers? Like, where is AI actually having an impact outside of the obvious places, making it easier to make a slide deck, make it easier to review things like loan application or obviously coding agents, but like, where are some areas that your team is applying AI, getting, finding alpha in that companies broadly don't maybe fully appreciate yet?

Yeah, so Wonderful started as an agents company mostly for customer support, mostly for voice. We then expanded to other modalities, to chat, email, We then expanded to back office processes. We then expanded to system replacement, like, or just like agentic apps. And we're now also expanding into new, new fields. For example, AI gateway and coding as well. We have so many different use cases with customers. Like a few days ago, I was in Mexico and met with this like large ISP that has over 100 entities around the world for— sorry, around Mexico for tax reasons, right? They just divide the revenue between many entities to not get above a certain tax threshold.

Oh, interesting.

And so the first use case that their CEO asked us to help them with is building an app for like reconciliation so they can reconcile between all of those different entities easier and faster. And actually didn't think it was a very big use case. I was like a bit surprised and talked with the CEO and asked him like, what's that about? And to make a long story short, they're probably going to merge with this other large company.

I have 100 employees too.

This other large company is an Oracle ERP shop.

Okay.

And the company we visited is an SAP ERP shop.

Oh.

So he told me, listen, I cannot stand the thought of having to integrate between 2 systems, which I obviously don't like, for the next 5 years.

Yeah.

So my ultimate goal is to leverage AI to create this like new agentic ERP for the unified company.

Interesting.

And this is something that I've never heard before coming from a CEO at this stage. Usually they know about like software replacement or creating a layer on top of legacy software, but to actually replace an ERP for a company this old and complex, that's the first time I heard this. And I think that's pretty consistent to my experience in Wonderflow because we work with enterprises and they follow the news. They constantly ask us for the newest stuff they read and hear about. So whenever something becomes like a talking point, we usually are going to be one of the first to have to properly productize it and deploy it.

Yeah.

The same thing is happening now with cost control, which surfaced the need for better routing and the gateway product, et cetera. But yeah, to your question, tons of different use cases.

What do you think, what do you think Wonderful looks like in In 5 years? Because now it sounds like there's— you're building your own software, you're also doing a lot of services, one-off projects. I imagine you're probably leaving some software behind, then you may be managing that, but like, and then you're working on a gateway. What does the business look like in 5 years?

So, we don't do— Wonderful doesn't sell any services, right? Everything we sell is consumption and and the platform itself. And sometimes we attach use case fees to certain complex use cases, but these are all recurring. We don't charge for a single FDE or DS that is deployed with our customers as a principle. It's very hard for me to tell you what Wonderful looks like in 5 years, uh, because 5 years is just a huge amount of time in, in this, uh, in this day and age. My gut is that in 5 years Wonderful is an independent company, one of the largest applied AI companies in the world, mostly seen as an infrastructure. So mostly known for our AI operating system with millions of different people around the world building their company on top of Wonderful as an infra. Yeah, I think in a similar way, in a similar way to how Microsoft is perceived today or how Microsoft was perceived a few years ago. Right. That's pretty ambitious. But I would say that's more the North Star than, let's say, Palantir.

Yeah.

Yeah. So, so, so you end up building something net new for a customer, productizing something for a customer, but then you're fine not charging for that build-out because you plan to use that same product. You'll be able to resell it to other players and capture margin that way. Is that right?

So, the flywheel between frontline engineers or FTEs into the core product team is something that happens pretty naturally because no one wants to do double work, right? And the challenges between different enterprises around the world are relatively consistent regardless of where they are geographically. So whenever there is a wonderful— there's a library with like hundreds of different agents and apps that were written on the platform. And that's the first place that an FDE or any DS, for example, would go to find out like a good starting point for their own solution. Most of the bespoke stuff that we build are actually integrations on the customer's end. We don't build that much like bespoke features into the core product for one customer. And the vision for me is to go after large use cases one by one. And I think the North Star should be that Wonderful acts as a 0-to-1 or like an agency machine for enterprises with all of those use cases we built left to their own engineers and own business stakeholders to manage. On our infra.

When did you start the company?

We founded the company in January 2025, so a bit less than two years ago.

And what's the latest news on the fundraising side?

We announced a C round a few days ago.

Give us the number.

As announced, we raised. We raised a. $550 million? Oh yeah, let's go.

Bang!

Big number.

Incredible speed. Incredible speed.

Incredible speed. Last question before you jump in. I'm sure you'll be back on for the next round. But there's a bunch of different products tying into the platform. Do you ever see yourself getting into into the inference business, or is that something you're fine working with partners? Like, do you imagine actually setting up your own data centers, or is that not like an area that you feel like you're getting a lot of value out of?

So I don't have a religious answer on this. I guess the first principle is that at Wonderful, we see ourselves as the applied AI partner for the enterprise. So when everything, when Whenever something becomes commoditized, we want to be the first to productize it very well and give it access to the enterprise. A good example, for example, is like coding, right? Coding was a very hard science problem up until maybe yesterday, right? Or a few months ago.

Yeah.

As soon as coding becomes a sheer engineering problem, then why wouldn't it be part of Wonderful? Just as a first principle. If inference becomes so commoditized that we think it's better for us to own it, then sure, we'll own it. But we don't have a principal decision to do it anytime soon.

Yeah, makes sense. Last question, because we didn't get to it in the beginning. I'm curious, what were you doing before this?

I was the founder of a pretty standard B2B SaaS company that I sold in 2021, and then I worked for 3 and a half years.

Thank you for your service in the SaaS trenches. Yeah, we love it. Uh, great to meet you, Bar. Yeah, congrats on all the progress.

Likewise, thanks for having me.

And yeah, good luck. Get some sleep, enjoy the weekend. Hopefully you can rest up. We'll talk to you later. Goodbye. Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web app servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. Jordi, I believe we have our next guests. We're ready to rock.

Here they are.

Welcome to the show.

What's going on?

How are you?

How are you?

How are you?

Thank you so much for taking the time. Let's kick it off with some introductions on both of you, and then we can go into the news today.

I'm Ken Ono. I'm founding mathematician at Axiom Math.

Yeah.

I'm Karina Hong, founder and CEO of Axiom Math.

Awesome.

And what is the latest in your world?

Is there anything left to do, Ken?

Yeah, job's finished?

Well, no, actually, I think it's a wonderful time for mathematics now that we have tools that can empower mathematicians. The questions are still there. Last I checked, the Riemann hypothesis is still unsolved. So it's a new era.

Yeah, that is the benchmark for everyone, certainly for the AI labs. You can see that they're both, you know, both the frontier companies are working on this, but do you see—

The 3 of us, you mean?

Yeah, everybody. Everybody's working on this. But do you see these particular challenges as good benchmarks, or is there a more broad overarching vision, or is it worth focusing on these specific things like the Riemann hypothesis?

Well, just very quickly, I'm a pure mathematician, so I don't think in the world of benchmarks. The problems are there, and I love the problems. Some of the problems get solved, but every time you solve a problem, there are new problems. So I think it's an exciting time. It's very disruptive for the community. It's hard, let's face it. But moving forward, 6 to 8 months from now, I think scientific research will be thriving.

Yeah, I think—

please.

Yeah.

The history, I think, has been that every mathematical discovery unlocks Either an applied, like an actual real-world application, like hundreds of years down the road, or even starts like new fields in science. So we're very excited about all these mathematical discoveries happening. For the audiences who don't know what's new, we had an exciting week. The very long, astonishingly, notoriously hard problem bounded gap in primes have been— the world record has been updated. 3 times this week, first by Yulia Soltman, a human mathematician, getting 240 down from 246, and then by Axiomath at 212, and then by OpenAI Astra at 186. So it's a— Wow.

Yeah.

Back and forth.

I am a blitz. Is it ever like this in math? This feels like a football game.

Yeah, that is all new. There's an analogy, if I could just very quickly say, Freeman Dyson was a very famous mathematician and physicist, and he described mathematics as mathematicians could be the birds or a mathematician could be a frog. And right now, right now we're in a position where we're amplifying the flyers.

Sure.

The ones with vision.

Yeah, I like that. What else can you tell me about the reaction from the mathematics profession and also what the role of the mathematics Yeah. What the life of a mathematician was in the pre-AI era and what it might look like in the post-AI era, because I know a lot of very successful mathematicians who have had full careers as professors, and proving theorems and working on these hard problems wasn't something that was really on their resume necessarily. Like, they contributed so much else to the world, and they lived full lives and had wonderful careers. It wasn't purely about the gamesmanship, although we're now— it's game on.

Proof Olympics time.

Yeah, Proof Olympics.

Like I said, I think the questions that I wake up in the morning thinking about, they're still there. It's just the paradigm. It's the tools that we use that didn't exist before are amplifying us. I think that's where we are. It's a great time.

So to track the community of mathematicians, but also the community of researchers working on AI for math, Pat Shafto, who is a program manager at DARPA, organizes this meetup every 2 months or every 3 months. And we just went to the last edition of that. It was like 2 weeks ago in Seattle. Terry Tao was there.

Yeah.

Alice Konterovitch was there. Boris Alexeev from OpenAI was there. A lot of the academics who are working on harness engineering that are great mathematicians are there. A lot of industry AI researchers are there, including people from OpenAI. We were there and AWS was there. The energy is incredible. I think it's like it was like a 3-day research program. Everyone was reporting their newest the open-source model. Godot Prover has version 3. That's Princeton Labs and Sangeetha Aroras. The community is thriving. I think we're seeing this incredible number of open problems that suddenly become attainable because of the technology advancing.

How do you guys make money?

Yeah, that's a great question. I think the award is going—

She pays me.

Oh, wow.

The world, the world is going to, I think, run on a different kind of flywheel of generation and verification. So we don't really think of formal verification as kind of like guarding or like capping the downside. We think of it in terms of performance and we think of it as optimization. So by having the cost of proof reduced zero, you can generate new algorithms. You can make new scientific discoveries. It's going to have important implications in the world of coding, in the world of AI for science, and other physical engineering disciplines that we couldn't think of in the past.

So put another way, the world runs on code that people don't understand. Many people haven't even read. So software verification, hardware verification, as Carina says, is kind of the challenge for the future, and that's really like doing math.

So, the Lean Startup of the future, all code written in Lean or something like that. Formal verification coming to all software with cybersecurity benefits potentially? Is that the idea?

There are a lot of different viewpoints on this. So, I've met researchers who think that all the world's software should be rewritten. There are other people who believe Well, we should formally verify everything. The truth is, I think, somewhere in the middle. People traditionally think of formal verification as something that you get zero credit for if you don't formally prove everything. We think partial verification has a lot of value by decomposing or refactoring. Each word has its own loaded meaning. Very difficult problem into subproblems, a very difficult coding challenge, coding task into different modules. skills, you can get assurance in certain parts and try to just lock in and optimize because you now have that formal guarantee. So it's again, this idea of verification for better performance.

I love it.

Can you guys walk me through your psychology when you're working on some of these open problems and you know that there's a bunch of different teams that are also working on the same set of problems. And it feels like one of those things where there's insane urgency because you could spend 3 weeks working with models and your own harness and tools trying to solve something, and then someone comes out with it and it's like the whole— you don't get any credit for your work because someone beat you to it. But are you feeling that pressure right now or are there enough disparate problems that—

Yeah, that's a great question. So the way we do it here at Axiom— and we have a little team called the A-Team, small but mighty— we have a list of problems that are our favorites. And on that list, there are some big problems like bounded gaps in primes, which we know the whole world of mathematics cares about.

Yeah.

So to tell you the truth, I'm not in a race with them until we get the idea that we're all onto the same problem. And recently, Julia Stalman's paper that came out a few days ago was a landmark moment. When that appears on the archive, you know that everyone in the world and the community is trying to optimize that. So there was crazy urgency the last couple of days.

Sure.

The psychology behind the scene was, I think, in February, we tried having AxieImprover tackle research conjectures in fields that are pretty deep, have a lot of theoretical literature. And that was successful. So Axie Improver was able to put on archive a few research open problem solutions, completely automated and also fully in lean without human intervention. That's in February. Immediately at the beginning of March, we were setting more North Star visions. As Ken said, there is a list of targets that we were excited to go after. The problem of bounded gaps in primes is a particularly beautiful one. It is a lot of people's childhood dream. So I think yesterday I saw Wei Jie Su, an OpenAI researcher, post that when he was 9, he knew about the problem. For me, I was, I was 12 years old.

I knew about Yitang Zhang's success.

So it was Yitang Zhang's solution of pushing that to a finite bound, 70 million, which is a very, very large number compared to today, the new world record. But that is a problem that's very to a lot of our hearts. So it's very natural that a lot of people are working on it. I think no one has ownership on any problem. And the world of scientific discoveries, it's pretty infinite. So you have a lot of problems to go after. We think we will probably enter similar races like the one with OpenAI we just had yesterday in the future. And also we're formalizing big problems. So we just saw Anthropic announcement on formalization of Fermat's Last Theorem. We have been working on similar things. So we're excited to continue to go head-to-head with OpenAI and with Anthropic.

I love it.

We are a 13-month startup. And we sometimes could be disappointed by being beaten the world record in a few hours. But we'll keep trying.

And I think we're doing a lot of fun.

You're very much in the game.

In the game.

Another thing I wanted to say, though, is among the problems that are our favorites, there are community— so last week, we set a world record on a problem called the parity of the partition function. It's not as famous as bounded gaps in primes, but for those communities, it set a world record. So our portfolio is vast.

Does math exhibit the same slop problem that you see in text and code generation? You know how you can go to a Deep Research report, it can write a really, really solid history paper that that might be PhD level, no hallucinations, but the experts and people who know this will just say, ah, just the flavor is not quite there. Or it has AI written all over it. Does that exist in the lean that is produced? Like the formal proofs people are like, ah, yes, it works, congratulations, but it's still a little sloppy.

3 layers. One is correctness. So a lot of, in the past, we'll be worried about correctness. Fortunately, you have Lean. And then 2 additional layers of taste. One is the code quality of the Lean formalization, the choices of what to keep conditional, the particular path, how you approach the definition of a very central mathematical object. And in our formalization of the 246 bounded gaps, which was done by polymath.ab, we really emphasize reusability.

Sure.

on producing really good code quality so communities can build upon it. And then the third part is the taste for the mathematics. Ken, do you want to add on that?

Well, the taste for mathematics really matters. Honestly, the percentage of math that has been understood by mankind out of all the math that probably exists, it's probably close to like 1% or 2%. Math is— we're just always discovering things. But you can prove theorems that nobody cares about or will have no application to the real world. And what's the good in that? So the human taste and the judgment and the final arbitration for what matters is very much a human thing.

Yeah, I think I— when I was a student in Ken's— he's REU— I remember Ken giving me the critical feedback of the symbol for elliptic curve conductor must be L. And if you use any other symbol, you will be criticized as not an expert in the field.

Hmm.

Very cool. Thank you so much for coming on the show.

Yeah.

Fascinating.

Congratulations. And yeah, super, super impressive progress from the whole team.

Keep up the good work.

Can't wait to see what you guys ship next.

We'll talk to you soon.

All right.

Aloha.

Aloha.

Cheers.

Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI, own the data platform that powers it. There is a fascinating deep dive, an investigative report in the Financial Times Wealth section, Playing with Fire: The Chinese-Cambodian Billionaire Chen Ji as an Investor. Cohiba cigars became ultra-luxury symbols. Then things came crashing down for the tycoon. It's a longer read. I don't know if we can go through it all, but there's some fascinating things here. One is that in 1962, JFK, he goes to his press secretary and says, I need 1,000 Cuban cigars. Because as soon as they arrived, Kennedy signed the order imposing US embargo on Cuban goods. So he knew that he was going to block Cuban cigar importation, importing, and he said, I need a little stockpile, so get me 1,000 Cuban cigars. But this is a— yeah, it's a strange business. Chinese-born billionaire who moved to Cambodia in his 20s and became extraordinarily powerful, extraordinarily quickly. By 2020, his Prince Group claimed interests ranging from real estate to banking to supermarkets and an airline. Then he owned a 53-meter yacht with its own nightclub on it. He bought a $100 million, basically £94 million, London office building and a £12 million mansion and cultivated the image of a connected international tycoon. But Chinese authorities had already begun investigating—

I like how he cultivates that image too.

A little bit, yeah. He's got some tycoon—

He has 21 years 83 years old. How is he doing all that he's doing? How does he know who he knows?

Yeah. But Chen, his most unusual asset was in Cuba. In 2020, Imperial Tobacco sold its half of Habanos, which is the group that markets Cohiba, Trinidad, and the company's other famous brands. They generate roughly half a billion dollars annually in the Cuban cigar market. And there's this weird illegal Cuban cigar market that has created a bunch of pressure where the Cuban cigar companies struggle to even supply the actual— the legal market because it's so complex. But he buys into Habanos, giving him an indirect stake of nearly 1/3 of the Cuban cigar company. And he He explored cigar warehouses in Singapore, cargo flights between Havana and Cambodia, really starts trying to expand the operation. But his empire imploded on October 14th, 2025. The US and the UK froze Chen and the Prince Group, his company's assets, with American prosecutors alleging that Prince Group was actually one of Asia's largest transnational criminal organizations operating compounds where trafficked workers were forced to conduct online scams. He denies the allegations, but in January of 2026, Chen was extradited from Cambodia to China and effectively disappeared from public view. Days later, liquidators were appointed to dozens of companies connected to him, including the vehicle that owned his stake in the Cuban cigar company. His lawyers are fighting the liquidation, but potential buyers are already circling the cigar investment, trying to get a stake. I don't know, maybe things will change over time. But the bigger problem is that Habanos may have damaged the business. The price has exploded.

Astra has landed.

What happened?

Astra seems to have landed—

Explain.

—in a lot of prosumer accounts.

Oh, cool.

Yeah. Let's refresh. I was looking at it. It is in my personal Pro account, not on the Chat tab, but on the Work tab. So if I go to chat.com and then I click on Work and then I check the model picker, I got GPT-6 Astra here. And not to make it too simple, I got light, medium, high, extra high, max, and ultra. Okay.

There we go.

Which consumes limits faster. So take it for a spin. If you have a ChatGPT Pro account, this is your weekend. Tell your family you're locking in, you're vibe coding something in Blender, taking it to Unreal Engine. That blog post, it's an Unreal Engine video game now. It's multiplayer. It's going to happen. Anyway, we have our next guest here with us in the waiting room. Let's bring in Bridget Mengler from Northwood. Welcome back to the show, Bridget. How are you doing?

Hey guys, doing well. How are you?

Thank you so much for hopping back on the show.

Great to see you guys.

So many flags. So many flags. Where are you calling in from?

What is this new space?

This is, this is old space. Old space. This is, this is Northwood HQ. First, first Northwood HQ. And yeah, we're starting to call ourselves the United Nations of Northwood.

So just more, more flags representing really the countries that we Yeah, I was going to say, if there were people, if it was representing the people on the team, I imagine there'd maybe be a few more. But yeah, very, very, very cool.

Yeah.

Yeah.

What has international expansion been like? Why is that important to the actual rollout of the product, the technology, the business? Why do you have so many flags so quickly as a relatively young company?

Right? No, we're doing startup on hard mode. Why not?

Yeah.

Yeah. But so reminder, we do networking for space. And, you know, you can— I like to do the comparison to like cell towers. When you're going down the freeway, you need to hit different cell towers to stay in contact. Same deal with the satellite. It's just way further away. And so their equivalent of cell towers are literally on different countries or continents. And that's how you can provide the continuity of coverage. So We kind of have to stand up a bunch of sites around the world to support satellites orbiting. And yeah, it's challenging. It's part of the game. But I think that's really gratifying.

What is the state of—

yeah, what is the actual process of a search look like? I imagine in a bunch of areas of the world you are looking at a region that encompasses multiple countries. And so are you basically doing like a roadshow and You know, you want to get buy-in, but then there's probably also incentives that you can get. But like, how does that actually work? I'm not planning to compete. Just to be clear.

Give me the secrets. Yeah. No, I mean, the annoying answer is it depends. It really depends on what the satellite operator is maximizing for. So some will care about, you know, their contact time metrics. So wanting to make sure that they can have, you know, minimal windows of gaps between their contacts. Others will care about, can you drop data in this one very specific location? And so once we have that customer parameter, then the question is, you know, where's the infrastructure that makes things cheap and easy for Northwood? What country is a good partner? And then kind of kick off the lease hunt and the construction and all that from there. And then we get it up and running. So yeah, we've been busy, been doing that in a lot of different countries. And I was thinking it could be fun showing the factory floor. So, it's been filling up a bit, and so we need to move to the new spot. So, these are our Portal and Prism units.

What are the serrated edges for? Do you have pigeons that land on the edges of the— I've only seen those on like park benches in New York City.

Yeah, no, you're not the first person to ask about that. It's really for our antenna performance. So, we want to make sure that We're really precisely hitting where we want to go. And, you know, RF has this thing called side lobes, which means that the signals will bounce off in kind of unintended ways. And so that's a way for us to kind of control where the signals go.

Interesting. So if I wrap my house in pigeon defense things, that might make my Wi-Fi better or worse?

Literally 2 birds with one stone.

Okay.

You'll avoid the pigeon, avoid the— no, I actually, I don't know.

We'd have to test it.

So how much space do you actually need in a country? Is this like you can get by with like a quarter of an acre or how? Like the hardware doesn't look like massive.

Yeah, you're not building a data center.

You have power and it's secure and it— but—

No, that's the whole deal. We compress it a lot. So our portal system, which is right here, it's designed to replace a 7.3-meter dish antenna, which is a hard number to visualize, but think of like a 2-story building. And so those portal antennas, you can combine like 6 or 8 of them together and it replaces that. But the cool thing is you can actually do multiple satellite contacts at once. So what might take like multiple football fields to support that much contact, you can actually do in basically the size of like a strip mall parking lot.

Very cool.

I had something else. Now I'm blanking. I lost my train of thought.

Yeah.

I guess my question was, what is driving growth of demand for ground stations, satellite connectivity generally? We've seen Starlink's meteoric rise, then more recently, I think earlier this year, there's sort of like the other side, the other team that's emerging. There's some deals with ASTS, some other earlier stage. There was an acquisition.

Yeah.

And there's a few other internet providers. It feels like there's at least going to be an attempt at making that a duopoly or oligopoly. But then there's also, we've talked to Planet Labs. There's so much else that's going up in space. There's government stuff. Both the near-term driver of demand for you and then also longer term? Yes, great question.

Overall, Northwood thesis was more things are going up into space and they need to be able to control the spacecraft. They need networking to control the spacecraft and to be able to sell more bits. That has definitely proven out. I think, you know, government and commercial demand for space, it's never been higher. Every operator needs those functions. And so for us, you know, moving to a new factory is about how can we ramp up the production on that hardware to support that. I think, you know, when you think of the product lines on the portal side, which is the one I just showed you, think of missions that need to do complicated in-space maneuvers, kind of like a Star Wars moving in space. So there's more customers trying to do that. And so that's driving expansion of that product line. The other one, the PRISM product, is really about how do you do like scalable, highly scalable, highly reliable networking for constellations. So kind of like the folks that you're mentioning there, a lot of demand. You guys mentioned a fair amount of it. I think it's been a crazy year for space, right? I mean, we had the SpaceX IPO, huge amount of enthusiasm and appetite. Like one of the things I was kind of curious to talk about with you guys, because I imagine you're, you're hearing it and seeing it too. It's like we're in this weird dichotomy where it's like appetite for space has never been higher. At the same time, there's like all of this concern around launch being constrained. So it's a wild time. It's a wild time for the ecosystem, I think. And, you know, I have my own soapbox on, on like what that means and how the industry is going to respond as a result. But—

Well, yeah, from your view, as I wish that we had more, more alpha to share to you, share with you on launch bottlenecks. But you're the one buying launch capacity. Like how much doesn't need to— well, downstream of it. Yeah, yeah, yeah. You're right. You're sort of limited, limited by it. Yeah. Potentially in the future where you have a good scenario is you get to the point where you're like, we are generating as much revenue as we can based on what is currently going on in space and new revenue will be, you know, dependent on that. But how much of the— we had the founder of— what's his name?

Philip from StarCloud.

Yeah, so Philip from StarCloud was on. He said part of the reason to raise their— they raised another quarter billion— Oh, yeah, they're buying launch passes in 2028 already. Yeah, and 2029. But when you talk to different space teams, is that really Like super top of mind for them, or do they have— is that like problem number 5 after making sure that their satellites and spacecraft actually work and all those other things?

I think it depends on kind of the maturity of the company. You know, for a lot of companies or, you know, things like that, they, they've already had their established plans. So it's not an immediate impact or concern, but something more down the road. You know, government operators have their ways to get what they need even under those conditions. But I think like self-serving, what I think is interesting, you know, we, like you guys said, we serve objects that are already in space. So we get to continue to have market there and then into the future. But I think what's going to be really interesting is if you do have constrained launch, you actually need to eke out as many bits like what you're actually selling from space. as possible and you need to do it as at high dollar as you possibly can. So like each one of those bits counts for so much more. And so, you know, my thoughts around that are like, point one, if you need to eke out more bits, how are you going to do that? You need higher utilization. That means like more seconds in contact with whatever that end customer is on Earth, making sure you can actually transact for as many seconds as possible. And then how do you send as many bits as you can per second. So like you get as, as much, as many bits out of each of those seconds as you possibly can. So that's like throughput and utilization, really. So that's one piece. And then the other piece is like, how do you get customers to pay higher dollar for those bits? Like, let's imagine there's some kind of winter where you're limited in the number of satellites that you can launch into space. You want all of them like maximized. You want all of them monetizing those bits as much as they possibly can. And I think that's where it comes back to, you know, reliability, right? Like that's something that US government pays for and that's something that more enterprise use cases really pay for and prioritize. And so I think like, you know, bear case scenario, that's what happens where all those operators are treating their assets differently in a sense where they're really prioritizing that utilization and high value, which like I think we've kind of been designing to serve for. So selfishly, I think that that works out well. But, you know, bull case, we figure it out and there's some other launch provider that swoops in and saves the day.

Similar kind of like at this point, sort of like intellectual debate on what happens with compute on Earth. Like, does compute aggregate towards, you know, certain model providers that are at the frontier that can get more economic value out of— out of—

Exactly.

A data center. or a single GPU. I'm curious about how does the launch—

What do you know about hosting Netflix in space to free up the ground compute for AI?

I mean—

Might be easier.

Yeah. How does the— is there a secondary market for launch? I'm thinking about people that are buying launch capacity today or pre-ordering it, and then they get to 2029 and they're like, I'm not— this is not the most economically viable. Sure, sure, sure. Use case of this capacity. But I imagine Elon would be like, cool, we're going to be the one reselling that. It's hard to imagine.

Right. Yeah. Yeah, I would imagine that. You're not the first person to float that idea, though.

What about your— I don't know if they're KPIs, but your benchmarks, like you mentioned, sort of like the The throughput and availability of a connection.

Correct.

And that feels like we're almost in like the pre— like with— if I just think about Wi-Fi connectivity, I think about throughput and latency, but availability is almost more of a bottleneck now. But hopefully we get to a place where availability is universal. And then does latency matter? What are the other kind of selling points and like benchmarks that you're using to understand the performance of your system versus the status quo?

Yeah, I think, I think throughput definitely matters. And then I also think there's an interesting kind of like, you know, I've mentioned the word like industrializing before, which I know is like it's a buzzy word and all that. But I think like what I mean for space with that is like as more and more use cases go to space, then you start applying those KPIs that might have applied to terrestrial market, terrestrial networks to space. And so one of the ones that— or some of the ones that we've been thinking about, yes, I think throughput really matters. What's interesting is that's actually about not just a single direction of throughput. People may sometimes not think about that, but throughput is actually a bidirectional thing. So you need to be able to both send traffic and receive traffic. And some systems are set up to do that in an asymmetric way versus ones that are set up to do that symmetrically. So, you know, just like One thing that we think about is, is like, yes, throughput and which direction is it going? And that's actually really relevant when you think about different markets. And then the other one is the reliability piece that I said. So you think about— I was chatting with somebody from Cloudflare recently and hearing about their kind of reliability metrics. Like people tend to measure it in terms of the number of nines that you can support.

Yeah. What is space? Does space have any nines?

Pitiful nines.

Zero nines.

Yeah, well, and so for like, for the, actually for higher reliability use cases, right? Like US government uses space and needs to have really high reliability performance.

I imagine GPS has been up for decades, is probably 5 nines at least.

Yeah, yeah. And so like the way that they do that is by either, well, both, like you have hardware that becomes much more expensive to make it more performant and then you have much more of it. So you have like backup and failover and all that. And so yeah, one of the ways that we design is like, how can you have that reliability at a much more like commercially viable price point. Because wouldn't that be so cool if you could use space for so many more terrestrial applications?

Yeah.

Um, so anyway, those are, those are like 2 metrics that we pay attention to and we track. And like, you know, we're, we're in baby network stage right now, so we are very much trying to like fail as much as we possibly can. And like, you know, uh, if you have a bad uptime day, like good job because you spotted a problem. Um, but eventually, yeah, I think we will want to be driving towards like really competitive reliability metrics.

Jocko mindset in the space.

I guess, yeah.

Problem? Good. Yeah, exactly, right?

Yeah, gives you that.

Have you had any big multinational telecom players seeing what you're doing and thinking, hey, we should probably get ahead of this and start trying to compete? Or does it still feel like a greenfield?

I think they're definitely paying into paying attention to space in a big way. And I think there's more opportunities for collaboration or for just like interesting configurations than maybe people imagine. Like, I know there's been a ton on the like direct-to-cell conversation, right? Like getting satellite signal directly to your phone. It also could be a part of the cellular backhaul setup, things like that. Yeah. You know, for instance, like a lot of our backhaul connectivity today is still done over microwave, which is a point-to-point RF link. It's just not up and down, it's side to side. And so I think there's a lot of, a lot of ways that, that the telcos can engage with space. And obviously they already are. They're paying a ton of attention to it.

Well, yeah, I mean, it feels like a natural place to put your Portal and Prism would be where there's existing— like, it's all the right sort of setup to do one of these things.

Last questions from the chat. The wired headphones, is this because there's a lot of RF interference in your headquarters? Or are you just a wired headphone respecter like Jordi Hays here? He loves them.

I wish that was the case. To be honest, these were just the first ones I could find. No political statement to be made.

No, people don't— we get— We get annoyed because we want more people to call in with Wired because there's no delay. Yeah, no, it's actually a better experience for most call-ins.

It's much better. It's been crystal clear. So thank you. I'm so glad you came on and breaking it down.

Okay. Awesome. Yeah. Great to see you. Great update. We'll talk. Excited to see more flags up there next time. Sweet.

Have a good one. Goodbye. Let me tell you about Public.com. Investing for those who take it serious. Obviously, they got stocks, options, bonds, crypto, treasuries, and more with great customer service. Jordan, what else is in the timeline that you want to hear?

Crusoe raises over $3 billion at $30 billion on an absolute tear.

He's chasing capital.

He's locking down— Absolute— I mean, Chase. Chase, he is the bank. He's the bank himself. More importantly, though, the National Weather Service forecasts dangerous beach conditions for the Malibu Coast starting today.

Dangerous?

Dangerous for novices or dangerous for someone at your level? That's impossible. Dangerously good surf conditions.

Dangerously, absolutely electric surf conditions heading into this weekend. Hope to see the TBPN nation in SoCal taking advantage of this hurricane swell. We're going to have back-to-back days of phenomenal activity in here on the Pacific. So I'm thrilled. I'm excited. Hope to get you out there, John. I would love to see you—

I might be tinkering with Astra, who knows?

I could see you getting into e-foiling, you know.

I got a startup idea. What if you made a message board or a bulletin board just for AI agents? I guess that would be Maltbook. But why do the agents love hacking into everything and making— just using it as a message board? They're obsessed.

Way easier than having to figure out how to ship You can scrape your own forum, basically.

There's something at a very deep level. I mean, there's another example of beyond the Hugging Face, they were using Artifactory, like the internal package manager, as a discussion board. Now there's news that they hijacked a German website this spring and transformed it into a bulletin board for other AI agents. That's from Reuters. There's something about— I'm still so interested in why they to do this and why collaboration is not prebuilt into the system, because that feels like it'd be more monitorable and stuff. But I'm sure we'll have more reporting on the Hugging Face incident and what flows from how these agents collaborate with each other. Obviously, monitorability and AI safety is very important. So, will be interesting to see where all that goes. Was there anything else, Jordi?

What else is? I think that's it. That's it.

There's a bunch more posts, but we'll get to them because we'll be back on Tuesday.

Have a happy Labor Day. Have a wonderful three-day weekend, and we will see you. Zeke says great weekend to get out on the pontoon in Malibu. Pontoon would be insane. Yeah, pontoon might flip it. Yeah, could be fun to flip it. Yeah.

Sounds tasty.

But, uh, yeah, be safe out there. Have a wonderful 3-day weekend. We love you guys. We'll see you Tuesday.

Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter at tbpn.com. And we will see you on Tuesday.

We love you.