Tech Turns to Mining, Meta VR Layoffs, Thinking Machines Shakeup | Matthew Prince, Chirantan Desai, Delian Asparouhov, Deepak Pathak, David Tearse, Blake Resnick

You're watching TVPN.

Today is Thursday, January 15, 2026. We are live from the TVPN Ultradome. The temple of technology, the fortress of finance, the capital of capital. Jordi has a new set of animations. The soundboard is growing in complexity. Every day there's a horse that walks all the way across the screen. And yes, guests can see that when they are calling in.

To be clear, to be clear, if we flashbang a guest, their entire screen is going white y Along with yours, along with ours. We're all in it together. Yes, yes. But it's somebody's birthday. It's Bobby's birthday.

Oh, happy birthday. Happy birthday. Anyone who's thinking about getting a gift for Bobby, get him ramped.

Get him ramped. I want every single person.

Time is money.

Save both.

Easy use corporate cards, bill pay, accounting, and a whole lot more all in one place. A little hot on the soundboard today.

But that's right, we're feeling good.

Yeah.

We have a.

We have a massive lineup. We have six CEOs joining the stream today. We have MongoDB, Cloudflare, Varda, Skilled, Carmen Industries, and Brink Drones. We're going to talk about the DJI ban on drones. We gotta add Matthew, we gotta add Blake Resnick to the linear lineup today. Linear, of course, meet the system for modern software development. Linear is a purpose built tool for planning and building products. So in the news today, Tesla is sharing that they made a ton of progress on their lithium refinery and then simultaneously AWS did a deal with Rio Tinto to buy a bunch of copper. Not that much copper. We'll get into how much, but they're sort of underwriting a new chemical process for refining copper. And it's all just sort of interesting that Marc Andreessen published the software is eating the world story back in 2011. And I think even the folks in tech who took that seriously were saying, oh, yeah, of course. Well, that just means that I didn't take it seriously.

I took it literally.

Yeah, but the people that took it seriously were thinking, okay, well, you're going past just the first wave of the Internet, the newspapers on websites, that type of stuff. And yes, we're going to get the Ubers. We're going to get more and more tech companies that are consumer facing or business facing. And you might see some transformation in legal services, financial services, logistics. Airlines will be using apps to let you book. But we're now in the era where the tech companies are starting to vertically integrate so deep that they own the mines and they own the refineries. And it's not just that they're actually building software for the mining technology.

Eating the world.

Yeah, yeah. And it's a new layer of a vertical integration that sort of just crept up on me because it wasn't an issue until the AI boom really necessitated massive industrial scale build outs. New industrial policy, which admittedly can't turn on a dime. But we're now two, three years into the ChatGPT boom, into the AI boom. The data center buildouts are getting bigger. We're seeing bottlenecks all over the place, whether it's chip shortages or energy shortages. And so the companies like Tesla AWS a lot of other folks are really looking deep into the supply chain, all the way into the ground, into literally eating the world. Because I see mining as sort of terraforming. I see it in some ways as literally eating the world. Like you said before, we move on. Let me tell you about Gusto, the unified platform for payroll benefits and HR built to evolve with modern small and medium sized businesses. So Amazon's buying the first new copper produced in America in over a decade. It's a very exciting project. And Tesla now has a Texas facility that can convert raw or directly into battery grade lithium hydroxide. It's a more efficient process. We can get into that. Tesla originally broke ground on this lithium refinery operation in May of 2023. It feels like they were sort of quiet about it, but now they're getting a lot louder. Obviously the narrative around China controlling 90% of rare earth extraction and rare earth elements has been in the news and this is sort of their response.

That shot with the cybertruck driving, it's good, it's good.

There's a voiceover we should, we should watch that video in, in a little bit more detail. Can we pull it up with some audio when we have a second? So this facility should deliver is the first spodumene to lithium hydroxide refinery in North America. We're deploying a new technology platform that is inherently much more environmentally friendly and cleaner. It's a simpler process. It's not that big facility. And this should be able to create enough lithium for 500,000 vehicles a year. EVs a year, maybe a million a year. I've heard Now Tesla's shipping 1.65 million. So this won't be 100%, but you copy paste this two, three times and you got enough coverage for all for the entire Tesla fleet. Take it through a kiln and cooler.

From there we take it through an alkaline leech and additional purification steps take it into crystallization and produce battery grade lithium hydroxide. Our process is more sustainable than traditional methods and eliminates hazardous byproducts and instead produces a co product named analzyme used in concrete mixes.

From breaking ground in 2023 to running rot through the kiln in 2024 to start a full integrated plant startup.

He looks so tired. He's just like, yeah, it's been, it's been great working for Elon. Hi. I love it.

Yeah, it's been relatively relaxed here.

Yeah.

The site.

We sleep in, the factory. In this case it's just dirt. But sleeping in the dirt, surprisingly comfortable. This refinery really enables us to have access to the critical minerals for energy storage for battery manufacturing. And it enables us to accelerate Tesla's mission by regionalizing supply chains for battery minerals and materials, by providing jobs by cutting emissions from the transportation network that's required for those supply chains. It really allows us to usher in energy independence for North America. Who would have thought that Texas would be the home of EV domination?

I know, narrative violation.

California really dropped the ball. So Tesla shipping 1.65 million, that was a little lower than the previous year. I think they were up at 1.7 million. But you get a couple of those facilities and it should be enough to cover everything that they need. There's also, there's been a whole bunch of news around rare earth minerals generally. Last year, a company called Iconic or Ionic Mineral Technologies made a large discovery of 16, 16 different types of minerals in Utah late last year. They this is in December and that was great.

As we do.

Yeah, yeah, yeah. It has been funny. We have been rumors about this, but they found a huge deposit of of all these different minerals. Obviously everyone's worried about China controlling 90% of the supply of rare earths, but we need to actually get these minerals out of the ground, refine them. And so that's where projects like that come into play. Amazon's partnership with Rio Tinto is a little bit different. They're not fully vertically integrated like Tesla is, but they are partnering with Rio Tinto. And Rio Tinto had been working on a new method of refining low grade copper deposits. So basically like there's a fair amount of copper in the ground, but there's a lot of dirt, a lot of other stuff that you have to process out that can be very expensive. But because of the data center expansion exploding, the price of copper has, I think nearly doubled in the past two years. And so copper is not the main ingredient to an AI data center, a token factory. But it is important because the biggest data centers use tens of thousands of metric tons of copper to move electricity around. So you need wires. There's copper in circuit boards, there's wire in the. There's copper in the transformers that move electricity around, and there's tons of miscellaneous electrical components all over your data center that need copper. So this particular AWS Rio Tinto deal is only 1400 metric tons over four years. I don't even think that's enough for a single AWS data center. But it's allowing Rio Tinto to underwrite this sort of risky new process methodology that if it works and they get the price down on this new process, then they can apply that to tons more copper reserves that's locked up in, you know, messy, messy ore that otherwise would be uneconomical to refine. So 70% of the global supply of copper is currently locked away in ore that's not economical to refine into actual copper right now. So if this new process technology works, you effectively triple the amount of copper. You triple the supply. And so price should fall. It's doubled.

You.

You triple the supply. Maybe price comes down. Good news for anyone who's in the.

Copper market trying to acquire Rio Tinto. Some quick backstory is one of the world's oldest, largest and oldest mining corporations with a history that spans over 150 years. Name was inspired by the Rio Tinto in southwestern Spain, but the founding again goes back to the mid 19th century. Spanish government was facing financial crisis and they started selling off their older mines. And so crazy story of just like a very old company continuing to be a key player in an entirely new tech trend.

Yeah, Rio Tinto has been in the news a ton also for lithium ion. They do some stuff there. They had one investment that I think they wrote down, then they spun another one up. They've been back and forth. Let's tell you about FIN AI, the number one AI agent for customer service. If you want AI to handle your customer, go to thin AI. So X freeze broke the Texas Lithium Ion Lithium refinery down a little bit more. It's the first in North America to convert the raw or into battery grade lithium hydroxide, Skipping intermediate steps. In the long term, probably more economical. It went from groundbreaking to first production in just 19 months. An unheard of timeline at this scale. The process is cleaner. No hazardous waste, A useful byproduct that can be turned into concrete. The single refinery can supply lithium for over 500,000 EVs per year. And directly challenges China's 60% grip on global lithium refining. Elon chimed in and said, it sounded like I saw another number that was maybe closer to a million, but either way, you do two or three of these and you're good for Tesla's supply chain at least. Maybe they sell it to other folks as well and what comes out of this will actually be sent to another company and then back to Tesla. But they are vertically integrating. Chamath Palihapitiya called it. He predicted this. He says, as we predicted in our annual predictions episode at the start of the year, unless someone shows up with superconductivity or carbon nanotubes, copper is the only game in town. And AI is a huge demand driver for a very under resourced material. So he's taken a victory lap with his thesis on copper. Pretty good call. Pretty cool that he has this, this clip here. So in other news, let's tell you about graphite.dev, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. And speaking of AI, there is turmoil in the trade wars, in the talent wars.

Barrett is back.

Barrett is back.

Barrett's back.

So back.

So this is from Kylie Robison over at Core Memory. She says. Breaking Thinking Machine.

Can we just give it up for Kylie for a second? Potential Scoop Athlete of the year.

Huge.

Putting up some crazy scoops.

Crazy scoops.

She had the anthropic X AI drama last week. She's on an absolute tear. She's on an absolute tear.

Yes, so she says. Breaking Thinking Machines has terminated its CTO Barrett Zoff due to unethical conduct, according to two sources familiar with the matter. CEO Mira Muradi announced the news in all hands with employees today. Sumit Chintala will be taking over as CEO and Mira posted as well confirming the news. She said, we have parted ways with Barrett. Sumit will be taking the new CTO of Thinking Machines role. He is a brilliant and seasoned leader who has made important contributions to the AI field for over a decade and he has been a major contributor to our team. We could not be more excited to have him take on this new responsibility.

And Alex Heath actually has some new news 30 minutes ago. He says more Thinking Machine employees are in the process of joining OpenAI after three of the startups co founders rejoined yesterday. Yeah, of course the Thinking Machine co founders had previously been at OpenAI but anyway, somewhat of a. Not great, not great. The I think potentially more controversial news was that Thinking Machines has branded squat racks at the office.

Yes.

Ryan Peterson posted a photo.

Ryan posted this.

A lot of people. A lot of people took this as somewhat of a red flag, right? People say, you know, the sort of common guidance is don't make merch until you're generating some real revenue. Right? And so if you take that further, it's like, maybe you shouldn't have.

What was Ryan doing to get this photo?

It seems like he just walked by.

He's going journalist mode. Here, you see this? He's like, he's not.

So if you zoom in, you can see these things look fantastic.

Well, well, Ryan, you're outside. You're outside the Thinking Machines branded gym.

Yeah, I'm somewhat. I'm somewhat envious. These things look fantastic. I'm kind of mad that we didn't get the branded plates, but this coming out while losing, you know, half of.

Your family, it has to be weird. I mean, how much did they raise? It was 2 billion. At 12 billion, something like that. I mean, it's gotta be so weird to start a company and on day one, have enough money for a custom gym. That's just like, unheard of. Like, you know, like, normally you start your company in the garage and maybe there's a weight in the corner you can lift or kettlebell.

One dumbbell. Not even two.

One kettlebell or something. But to just come in and say, yes, we're taking class, hiring a massive team, bringing on six or seven co founders.

Atlas was saying, the other red flag is, where are the 45? I think I'm gonna assume they're just not in sight. Like, you can't see them. Right. They would be mounted lower, right on the rack, but never change.

That's so funny.

But, yeah, so rude. In case you forgot, Andrew Tullock left thinking machines in Q4 of last year. Zuck had been trying to poach him and offered increasing amounts of money. And Andrew. The sense is that Andrew was like, I can't be bought.

But he needed the money. He had a shopping cart filled with chrome Hearts supreme. And he was like, wow, my next outfit's gonna run me 500 mil. I gotta get a new job. I can't afford it. I can't afford my lifestyle. Maybe it was a case of lifestyle inflation. They talk about that. He's just like, okay, I got a new SP3 Daytona. Now I got the Valkyrie and they.

Gave me F80 allocation.

I need some ca. I gotta get a new job. It's just not. It's just not covering the bills over here.

Tyler, do you remember the final amount that Meta paid, it was like, because I always appreciated.

There were a lot of rumors.

It was in the billions. In my head. It was like Andrew was like, yeah, I'm really honored that you would make such a big offer, Zach. A billion dollars. But unfortunately, I'm very mission aligned. I'm mission driven. There's really no price that I would take to leave the company that I started. And Zuck's like 2 billion.

I think all the rumors were actually from the summer when there was like all of the talent wars going on. So first there was, I think it was just one billion, and then just one. There were rumors of three and a half and that being turned down.

That's a lot.

But then later in the year, then he eventually wins.

Okay, so people kind of assume that the rumor is true because he eventually made the move.

Yeah.

But I mean, it's not necessarily higher than three and a half.

Right.

Because now we're seeing more people leave. Like, why are they leaving?

Yeah, yeah, yeah.

You know.

Yeah, it's.

I mean, it's very unclear. Right.

Yeah. The actual estimate was like, something like the Journal had reported one and a half billion over six years.

I mean, there is a risk to starting a company with six co founders. Like, that's just. That's just. That is unorthodox.

Unorthodox. So some amount of churn.

And they still got John Schulman.

Right.

Who's absolute legend. Co founder, OpenAI.

Yeah.

Like, everyone who is still the there or everyone who was there, goated. Like, they're all goaded. And it's like super hard to. One of my buddies, he's like, super smart. He was trying to get a job there. It's like insanely hard. You have to be actual. Go to work there.

Okay.

But now they're leaving.

It's a bit of a goat pen.

Yeah.

Goat farm.

Yeah.

I mean, Dylan Patel was really excited about this. He said, can I invest in Mira's company? God damn. Two more goats going there. Back in February 6th of 2025. So a little under a year ago, everyone's saying Shulman. But the other is as arguable. Just as arguable, just as if not more goat. And David Holt says, I'm calling it Marathi Heavy Industries.

Yeah. I mean, at one point, one of my buddies who works at a big lab, he was saying at that time, think Machines has like, the best team, like any of the big labs.

Yeah. So how would you describe what they do? Because we've heard a number of different. You Know, pitches from them, products from them, and some of them seem like, you know, this could easily make a couple billion dollars a year.

Yes. They've shipped a bunch of blog posts, but they also, I think their only like real product so far has been Tinker. Okay. Which was basically fine tuning, right?

Yeah.

An API for fine tuning and made it very easy. I think it was a super good product. People who've used it like really like it.

Yeah.

So I mean the general focus has been enterprise and.

Yeah.

So the ChatGPT everyone at OpenAI has been very obvious like this is the year, this is the year they're focused on enterprise. This year I think the idea was.

Like RL environments for specific businesses.

Yeah.

Yeah. So you get a really great fine tune. I know obviously OpenAI has a similar business line that probably competes directly, but you know, being a pure play can work out. We've seen what anthropic's done. It didn't seem like a crazy, crazy strategy.

Yeah. I mean John Shulman is like one of the OG main RL guys. The team is perfectly suited for this.

Yeah. So I don't know, maybe smaller team, more focused, more a couple big deals, they start finding their footing. How AGI pilled is thinking machines? Were they on our chart? I don't think they were. Right. SSI was on the chart of needs. AGI is AGI pilled. I don't think thinking machines made the choice.

Yeah.

Actually I don't know how AGI pelled is thinking machines based on what we know.

Yeah. I mean it's hard to say because it's like. Okay. Compared to like the average person, they're obviously way more AGI pilled. But the idea of like doing. You're doing RL for business like AGI, it's like G is general.

Yeah.

So if you're just focusing on specific business use cases, that's not very general. Right. You can imagine someone is just like building the massive model. They're just going to one shot everything. That's much more like AGI.

Is anyone doing abi? Artificial business intelligence.

I think that's what Internet would be a green.

Yeah. ACBI corporate intelligence. AVI artificial voice intelligence. 11 labs build intelligent real time conversational agents. Reimagine human technology interaction with 11 labs.

I think the question is there was already immense amount of pressure and really, really, really high expectations. I'm thinking machines around what they put out first. Right. I don't think these people leaving, maybe it makes it harder to deliver Right. On those expectations. But the Expectations are still sky high. It did feel like kind of rough messaging. Very, very rough messaging. Right. Unclear coming out, pushing this unethical conduct angle. People at OpenAI have obviously their bias, but. But they've denied it. Susan jiang, who's at DeepMind, said, Feels like the type of character assassination leak you put out when you find out one of your co founders might be splitting off. And Rune says 100% snaky PR move, so.

And Signal says there's more drama in AI than the real Housewives.

In the Wired article, the quote is, a source close to Thinking Machines alleged that Zoflo had shared confidential company information with competitors. So there's like that and then there's the like. Yeah.

And like to me that's just like, okay, it sounds like he had been like in conversations with OpenAI, he could.

Have just been complaining about the lack of 45 pound weights. And that is technically confidential.

Yeah.

He shares it with somebody else, they overhear it at a coffee shop and they're like, you shared confidential information.

Yeah, yeah. But that type of allegation is like you guys talked about AI that you shared confidential information.

Yeah.

So anyways, well, the scoops will continue. It is played out. I'm sure we'll get more.

What did Fleeting Bits say? Fleeting Bits said some quick thoughts on Barrett, Zoff and Metz and shown Holtz going back to OpenAI. Barrett was the Vice President post training at OpenAI before he left for thinking machines. He left OpenAI in September of 2024. Barrett would have, probably, would have probably initially had options for about 10 to 30% of thinking machines. That's insane. They had six co founders. How could that be so high? Post training was one of the more valuable skills at launch, I suppose. Okay. The seed round was 2 billion at 12 billion. So this would mean that he would have options now for around 8 to 25% of the current company. This seems all hypothetical here. There were rumors back in November that Thinking Machines was in talks to raise at 50 to 60 billion. So this puts is equity at 4 billion to 15 billion valuation. Thinking Machines could probably exit somewhere at between 30 and 60 today. Who has that money? I don't know, maybe Nvidia buys them or something. At least at some point over the next few years. I think the most likely acquirers would probably be Microsoft, Apple, Meta, Amazon and Nvidia. Okay, large cap company that wants to build a frontier model. Barrett would have been just over his one year cliff. So assuming he was not fired from cause or thinking, he doesn't want to litigate. He still has about 1 to 2 million.

Okay, that part's incorrect. How does that math work? I don't know. 25% vested.

Yeah. So one quarter of the equity. So this person's assuming that he got 10 to 30% of the company and that the value of the company currently is 60. A four year vouch. He's over a one year. And so you're looking at a billion dollars of equity. Would be interesting to know how his OpenAI pay package. The rumor is that OpenAI set aside about 50 billion for the equity pool for the new PBC so they can afford a decent pay package for him. My guess is that it was personality dispute or disagreement over the commercial direction of the company with Mira executive team. And if there is something behind the unethical conduct claim, it could just be a game of telephone over him talking about to the other labs and mirror interpreting it as him disclosing trade secrets. Huh. Interesting. Well, let me tell you about Applovin. Profitable advertising made Easy with Axon AI. Get access to over 1 billion daily active users and grow your business today. It's called Thinking Machine now because there's only one left. No, there are three left. So it's still plural Rahul. But that's a funny post nevertheless.

Yeah, reading into this, there's got to be some real excitement on OpenAI has just been outflows, outflows, outflows. Getting some of these superstars back on board has got to be good for Mer.

So is this with the elves and Valinor? Do we know what Rune was posting about? Because we learned that Rune posts the elves have left for Valinor every two years apparently.

Okay, well he posted once before. Yeah. In 2024. About. In, I think. What was it? January 22nd.

Yes, yes.

So yeah, so now he posted this like on the 12th, so three days ago. I think most people have now taken it to mean to be about this. Right? People leaving thinking machines going back to OpenAI. But I think there's a couple ways you can look at it. Right? So if you look at it in the context of the last time you said this, it was the exact same quote that elves had love for Valinor. Like okay, what is Valinor in Lord of the Rings.

Right.

Valinor is where elves go. So they don't like diminish or like they don't like kind of fade away.

Yeah.

So it's okay. What does that mean? Like maybe it's. You're just basically securing like massive wealth, right?

Yes.

You're not, you're not going to be diminished by inflation or something slowly, you know, dying or.

The other way to read that is like you go to, you go to like one of these newer labs and you could be a superstar, but your, your abilities could fade away with time. Right. Because you just don't have the scale or the resources.

Yes.

Yeah, but so, so I think in the 2024 context, Valenor actually was the Neo Labs.

Okay.

Because you basically go start a company, be the co founder. You guys raise at 12 billion, you're just like, let's go like, yeah, I just secured insane amount, like massive bag.

Yeah.

But then now, now OpenAI is GroundEx. OpenAI is now Valenor. Right, got it. They're setting aside 50 billion just for pay packages. If you go back, I mean, you're doing pretty well. It's pretty safe. Right. You don't have this startup, you're not sure if it's going to work or not. So it's kind of a circular thing going on. There's other ways. You can even read the most recent one. There's like the billionaires leaving California, maybe Texas, Miami. That's Valenor. You could say the agents are the elves.

Maybe Barrett was so afraid of the Billionaire Tax act that he's like, I don't want to be a billionaire. I'd rather not be a billionaire than give 5%. Just give me a simple hundred million dollar pay package. I'm a humble guy. That's it. I don't want to be worried about that. AGI is coming.

Yeah, the new pay package.

Money won't matter.

Should be 999 million.

He's like, I don't want to deal with the headache of paying out taxes on unrealized cap gains. I'll be bankrupted.

Yeah, yeah. So yeah, Roon is talking about the framing around this before cognitive restructuring. Important people are leaving after cognitive restructuring. They're being fired for unethical conduct. The mom and dad are fighting more Thinking machine employees are in the process of joining OpenAI after three of the startup co founders, all X. OpenAI rejoined yesterday. Three. I thought it was only two that rejoined, but it does seem like they're.

It's two. They're counting Andrew.

Oh, okay. Rejoin. But he didn't rejoin OpenAI, so I don't know if that would count. Little odd.

You're just saying. Yeah.

Anyway, it does seem like the talent wars are unending in the AI race. We got to talk about saunas. We got to talk about the sauna. Wars because they're heating up and Vanity Fair has a piece on it. So before we read this, let me tell you about console. Console builds AI agents that automate 70% of it. HR and finance support, giving employees instant resolution for access requests.

Everyone's focused on the AI wars. Not enough attention going to the sauna Wars. The real heated rivalry is at the bathhouse.

Yes.

From bathhouse to other ship to Altar, a wave of bath houses in New York City are popping up just blocks apart, igniting a war of Zen. And this. This is especially relevant to the TVPN team because we've been in somewhat of a sauna war ourself, which we'll get into at some point in the story. The Flatiron District in Manhattan has been a center of Nuvo wellness for around 15 years, with its many boutique fitness classes, gyms, acupuncturists and stretching and recovery centers. Now a wave of bath houses that offer dry heat and cold plunges face off in a few blocks. There is bathhouse on West 22nd, with other ship two blocks away, and Altar set to open this winter. You could walk between all three in mere minutes. Williamsburg has its own cluster, right?

Parallels super cluster.

Yeah.

With the original location of Bathhouse and other ship that followed. Welcome to the Sauna wars, where dedicated bath houses compete with members clubs like West Villages, continuum, which costs 40,000 per year, that has a bathhouse set up in the financial district. Co working space. WSA's wet lounge as well. Co working space with a wet lounge. Not something bizarre, not something you hear about every day. The fitness chain TMPL has a whole subway ad campaign entirely around their bathhouse facilities, featuring women in swimsuits reclining in a suggestive way that one doesn't really find in saunas in real life.

Very odd.

They joined stalwarts like the Spa 88 and Air and Manhattan, CityWell and World Spa in Brooklyn, Spa Castle in Queens, and of course the Russian and Turkish baths in the East Village, which is the sort of place newcomers are brought to see a slice of the real New York City. It's usually teeming with people, some getting whacked by bushels of oak leaves in a treatment called Plaza and has two owners who rotate weeks of ownership. That's interesting. It's a funny concept. It's no frills and even a little gritty. For those who are used to conveniences such as booking ahead, there's none of that to be found. They do have a nod to modernity with an active TikTok account featuring guest endorsements recently highlighting Umar Thurman wrapped in a pashmina and calling it one of the greatest best New York institutions. The first time I went to 10th Street Bass was in 2007. I love the authenticity and how it made me feel I would find my way back there. When I felt depleted, underslept whatever it might be, I would come out feeling great, says James O'Reilly, who is one of the founders of the co working space Neuhaus. His latest project is the new Lore Bathing Club. This is where you go if you want to cultivate lore in your life, I guess. Founded alongside restaurateur Adam Elser, Lohr takes some inspiration from the Russian and Turkish baths as well as the communal sweat traditions of Europe and Asia and drops it into a 6200 square foot space finished in Travertine and White Oak in NoHo. The Bath World is quite a scene one where skincare brands such as Pharrell Williams Human Race send out a press release to announce they're temporarily supplying Lore locker rooms with their signature 7D gel sets. What is a 7D gel set Tyler?

I have no idea.

Other chip is hosting comedy nights. Alter will be selling bathing suits that are custom made in Brazil and has hired a lighting designer who has worked with Billie Eilish. So you're always feeling your best and comfortable and tan, Jafari says with a laugh while giving a tour of the space. There's already a wait list of more than 500 people for their membership which will cost $275 per month up front for $500 worth of credits to be used towards sauna and cold plunge, but also IVs and shots of vitamins, hyperbaric oxygen therapy, red light, NADA compression and PEMF. Kohler has an indoor sauna with a list price of $120,000, an outdoor sauna with a list price of 95,000 and even launched a collaboration with the wellness club Remedy Place for an at home 39 degree ice bath for 20.

Do you think Brian Johnson will get into IRL experiences? Physical locations? I know was it Dave Asprey did that for a while there were a series of bulletproof coffees.

Yeah there's one that one in Santa.

Monica and there's been a few like locations where you can go and they have a hyperbaric chamber you can get your your blood tested like a variety of sort of health and wellness like biohacker hubs some of them.

I think Brian would make a nightclub bathhouse probably because he would want to you know his new thing is love as the ultimate longevity hack and so could see him doing something like that. It is a sauna boom reminiscent of the glut of boutique fitness studios that flooded the market in the wake of the success of SoulCycle. It used to be that you would go to a gym and it was outrageous to go to an expensive group class, says Jafari. But now there are 99 boutique fitness fitness concepts in a 4 mile radius. Still, there are not as many as the peak days. Pre pandemic. Not every boutique fitness studio survive, let alone thrive like Solidcore and Tracy Anderson. Competition is always something we are aware of, says Emily Bent, Othership's co founder and director of marketing. The saturation in North America for bath houses is not even close to what it is in Europe. We're starting an industry and rising tides and all that. I have heard about othership and what an insane business Othership is from a number of people. So not surprised to hear they're expanding. While the message from founders blissed out on 180 degree heat is one of bringing together community and good vibes, the reality can be cutthroat. Consider the case of Bathhouse, probably the most well known of the businesses in the second wave of New York City sauna culture. Co founders opened their first location in Williamsburg in 2019 and a second in Flatiron in 2024, both featuring sauna, steam rooms, warm pools and cold plunges. Goodman, the founder, saw the company's more is more approach through the model of working on large event production. You think like an experienced designer, so going to Bathhouse is not a monolithic experience, but more of a choose your own adventure. When it first opened, Bathhouse was jokingly called the Bitcoin Bathhouse as they used the heat generated from min to warm the tubs. Which is I've been to both of the Bathhouse locations in New York.

They need to do LLM inference. Yes, this is a no brainer. Everyone who was in crypto mining needed to pivot to AI. They should just do that. They already have all the infrastructure.

This is the solution to the political problem. People are mad at data center power use. But if you have a nice bathhouse that you can go to for the.

Public, everyone's at the bathhouse constantly talking to Claude or something. I imagine that the water use in the bathhouse is actually probably pretty significant. I imagine that they're like draining the pools fairly often.

So I think you get the water use. I think they're probably draining them less Than you would hope.

Oh, that's not good.

So anyways, the comments originally were like.

There'S a bunch of things I want to talk about. First, crowdstrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. Second, we have a key question. Do you know who Uma Thurman is?

No.

Kill Bill. You've never seen Kill Bill? Tarantino. It's fantastic.

Oh, I. Yeah, brutal. Also Pulp Fiction. I've seen Pulp Fiction.

Yeah, Pulp Fiction.

Anyways, the comments were like, oh, they're laundering money through bitcoins. Tallmadge. But it's just a fancy pool heater. Those jabs did not prepare them for what happened earlier this year when someone posted on Rediscover. I noticed the hot tub and body temp tub were looking kind of dirty and gross. I thought it'd be fine, but then I ended up with a uti. Not great. A UTI doesn't just walk across the river. If it was a problem, it would be a pool at a single location at one given day. There's 24 hour computer monitoring. So I'm glad that the bitcoin heated pools are getting 24 hour monitoring and they're manually logged five times per day. And we keep the logs. We are adjusting ph and chlorine at all times. Every drop of water at every pool turns over every 30 minutes. All the vessels have their own independent system, so they don't mix. The water goes through a sand filter that takes out any particles up to two microns, like a receipt in their pocket or a tag that falls off. Then it goes through a UV filter. Big tank with three foot long light bulbs, a pure UV light. So they're going pretty hard.

Cool. Well, vibe co. Where DTC brands B2C B startups and AI companies advertise on streaming TV, pick channels, target audiences and measure sales. Just like on Meta. Anyways, speaking of Meta, we gotta talk about Meta because they just laid off 1500 people in their Metaverse division. It's 10% of their staff. It affects the company's AI glasses and other wearable products. The stock is up on the news. Obviously, 1500 people is a ton. That's a lot of people. But they still have 14,000 people working on reality labs. So there's a question about like, what exactly will they be doing? How much does this represent a shift? The messaging has certainly shifted to augmented reality instead of virtual reality, but it feels like they will be continuing a lot of stuff. The Wall Street Journal has a report Here. In December, the Journal reported that planning, that Meta was planning to make budget cuts to teams working on the Metaverse. Meta spokesperson said. We said last month that we were shifting some of our investment from Metaverse towards wearables. This is part of that effort. And yeah, you see the Reality Labs team work on a bunch of different stuff. And it started as sort of the Oculus acquisition, a VR headset. Then all of a sudden it was Meta Ray Bans, which is mostly just a camera, certainly not a VR headset. But it feels like it's sold many more units. I actually don't know the precise numbers.

I'm super curious how many of those 1500 people will end up working in VR in their next somewhere else. Yeah, it's just like, can the industry really absorb that many people? Right.

I don't know. I mean, there's projects at Google and Samsung and Apple Vision Pro, obviously, and then there's a couple startups. But yeah, where do you go? I mean, I imagine that a lot of those people are just broad technologists and can work on anything because they're probably working on servers or supply chain or purchasing or marketing. Like there's so much to do within Reality Labs. It's a massive organization. I mean they ran like huge TV campaigns for the Oculus and they will probably continue to with a quest. A year after the name change in Pivot, the Journal reported that the company's flagship Metaverse product, called Horizon Worlds, was failing to catch on with users. It had less than 200,000 monthly active users and most so called worlds were never visited by anyone at all. The company's presentations, attempting to stoke enthusiasm for Horizons Worlds, provided fodder for popular mocking Internet memes. People were making fun of it.

Yeah.

The Horizon Worlds product.

I still believe Meta is a fantastic name for Meta.

Yeah. Do you think Salesforce should rename? I wanted to ask Benioff that yesterday if he's ever just gonna go Force Metal Force Force. Metal Force would be a good name. But yeah, I mean he is moving past sales. He talks about Agent Force and Service Force and all the different groups he has. Maybe he's got to change the name at some point.

Be fine.

But Salesforce is kind of a lindy name. It's been around for so long.

Who has the ticker?

AI C3 or something? I don't know. Meta's been ramping up AI spending on AI, pushing its capital expenditure to 72 billion last year, with plans to accelerate that spending this year. It's also doled out offers Worth tens of millions, tens and hundreds of millions of dollars to AI researchers and engineers and recently acquired a Singapore based AI startup called Manus for more than 2 billion. The company's Ray Ban smart glasses equipped with AI have also started to take off. The company sold more than 2 million pairs and has delayed the rollout of its newest glasses in Europe as it struggled to keep up with Demand in the US. So 2 million pairs of smart glasses, the Ray Ban Metas versus 200,000 monthly active users in Horizon worlds. I wonder what the retention is on Ray Ban meta Ray Bans because if they're retaining less than 10%, then the actual MAU pool is lower.

Yeah, but it's the other thing that's so interesting.

I see people, I've started to see Instagram reels of people using the smart glasses really to do first person videos and often prank videos. So I think that there's actually a flywheel there that's stronger than what's happening, happening in VR.

Yeah. The thing that's interesting about not focusing on Europe is that my initial reaction of using the Ray Ban metas was if I was using WhatsApp as my primary messaging app, these would be like way more interesting. Right. Because you have the integration, you have the heads up display, you can get messages, you can read them, you can respond to them. It's super, super integrated. Meanwhile, they don't have an imessage integration and so they're not that valuable to me from a messaging standpoint, which is kind of one of the key use cases.

Yeah. The sort of Ben Thompson takeaway from this was that even though Mark Zuckerberg has effectively complete control over Meta and can rename the company and move very aggressively and acquire big companies, at the end of the day he does respond to shareholders and the shareholders have been watching the Reality Labs losses pile up billions every year. And at some point if you have a new sort of big money pit initiative that's AI and you're going to spend tens of billions there, you have to sort of make a sacrifice or at least show the shareholders that you're refocusing on the new thing. Even though obviously AI and VR play very well together. Part of the biggest problem with VR is that it's very difficult to develop. You have to develop this full game. AI can enable a lot of that, speed up development. And then there's also the sort of longer off take, maybe not that far in the future. But this idea that if you generate a world model like the Fei Fei Li World Labs project or Google has a project as well, genie. That might be something that actually gets you to put on the headset because it's infinite and endless and requires a whole bunch of upfront cost to train the model. But then once you put it on, you're just in that and you can prompt it yourself. You're kind of designing your own game. And that's a unique experience that seems uniquely suitable for VR. Ben Thompson's take was that the Meta Vibes was like a virtual reality project, effectively. Even though you didn't put a. Put a glasses on, it was created.

Well, and that's been my take on the name. It is like it's still a metaverse. It is a metaverse, yes. It just looks like a suite of apps.

Yes, yes, yes. And now you access that, you access that metaverse through a screen. And how close that screen is to your face might take years to get closer and it might be small when it's close. For the first, first iteration, I'm very interested to see how the Ray Ban displays are selling. 2 million Ray Ban smart glasses. I've seen them sell the Ray Ban displays at Best Buy and shield. Mona was talking about sort of his frustrations getting onboarded. I haven't. Have you run into anyone wearing them in the wild? I've run into people with the Ray Ban smart glasses all the time with the cameras on them, but I haven't run into any.

Seen the new ones, the chunkier one with the display?

Yeah, I haven't seen anyone just rocking them even though they are out.

I honestly, I feel like we don't see that many people never go outside. Well, I mean, on the weekends I do, but I'm like out and about. But it's kind of a small community. But I didn't get a chance to give a. I wanted to go back briefly to the sauna thing. I didn't get a chance to give a take. I am extremely bullish on Americans starting to sweat more because the evidence is so clear that if you use this on every day, you will probably extend your health span lifespan. It's heavily, heavily studied because of how popular it's been in the Nordics for decades and decades. So very bullish on sweating. One of the things that's interesting though is that. But New York has like, you know, New York is like the center of consumption. Right? People have tiny apartments, so they're just constantly consuming. Right. Food, you know, exercise, shopping, et cetera. But I don't think it's representative of the way that the Americans will Like sweat broadly because you can get a sauna for like a couple grand. And so all these bath houses are like a few hundred dollars. And so I think for some people they want the social experience, whereas for me, I don't like the, I'm not really like that into the like.

You don't like talking in the sauna?

I do like talking in the sauna. We've obviously had our sauna wars. Our sauna wars are. I guess we were talking too loudly in the sauna the other day because this guy just absolutely, absolutely snapped. I was, he was a bit of an angry elf. But, but, but anyways, like that most consumers are gonna be looking like, I can go and buy a sauna for.

$2,000 and I have space for it because I have a two car garage and a backyard and maybe a pool. And like for the average American that has a little bit.

Last time I lived in an apartment, this was like six or so years ago. I had a whole cold plunge that I snuck into my apartment and the sauna just. I had a two bedroom loft.

You had a sauna too?

No, no, I had two. I had a two bedroom. I had a two bedroom loft and one of them I just put a. It was just like cold plunge on a wooden floor.

That's insane.

And I had a sauna. Of course, I don't think you're allowed to bring like a mini swimming pool.

If it leaks, you're gonna destroy the whole.

Actually, I did. They did eventually tell me I had to get rid of it because there was a leak in the building and they went in to look and they were like, it didn't have anything to do with my hole plunge or whatever.

But okay, so if you were, if you were talking to Brian Johnson, you would say, skip, skip the physical institution that is the sauna. Start a line of saunas that can be delivered on a truck to homes.

Yeah, maybe. I mean, there's a company called Plunge that scaled insanely quickly.

The company's done really well.

Yeah, I think over nine figures of revenue off the cold plunge boom. But yeah, the main thing you have to understand is for other ship in these companies, you have to go I think into dense areas where people don't have a lot of space to have saunas. Because a lot of people are going to look and say, I either have one in my gym or I can, you know, people can go and just. Even if they don't have the two grand, you can use a firm and like you're effectively paying a couple hundred Bucks a month for like a private experience. Yeah, we'll see. We'll see. Gabe says the big sweat lobby is making moves. I like it, but yeah, I just, I've remained so bullish on this trend because it's just passive. Right. You don't have to like work up like the motivation to sweat.

Sure.

So you get a lot of the benefits of exercise, but you're not like, I gotta force myself to go for a run. It's just like, get in the box.

Get in the box.

The label box.

Yeah.

Let's tell you about label box. Label Box. Delivering you the highest quality data for Frontier AI. Get in the box.

Get in the box.

Label box. Label box should do a drop of custom saunas for everyone.

Daniel Gross says I am building out Meta's compute desk. I'm looking for folks with a background in the term desk desk.

It feels like New York, it feels like finance. It's not just a team, it's a desk.

Yeah. Deep learning, supply chain, commodities, semi key stacking, copper, energy, Excel, prediction markets, monitoring situations, et cetera.

This is amazing.

Candidates should have experience in at least one of these domains. Somebody's like, I've monitored a lot of situations. I'm qualified here. And novel ideas on how to apply LLMs to their work. If you'd like to help manage 100 billion of compute spend, please DM and include information about why you're a fit. If you're in doubt, if you're in doubt about being a good fit, please apply.

Yeah, at least one of these domains. I only know Excel. I don't know anything about any of the others. I'm just an Excel jockey. Good luck. But yeah, I mean, massive compute spend. Very interesting to see where the Meta Compute project goes. Obviously they're spending a ton. Mark Zuckerberg said today we're establishing a new top level initiative called Meta Compute. Meta is planning to build tens of gigawatts this decade, hundreds of gigawatts or more. Over time, how we engineer, invest and partner to build this infrastructure will become a strategic advantage. It's led by Daniel Gross, of course, and Santosh will continue to lead the technical architecture, software stack, silicon program developer, productivity and building and operating the global data center, fleet and network. Daniel will lead a new group responsible for long term capacity strategy, supplier partnerships, industry analysis, planning and business modeling. They'll work closely with Dina Powell McCormick who just joined Meta as President and Vice Chairman to work on partnering with governments and sovereigns to build, deploy to invest in and finance Meta's infrastructure I'm looking forward to working closely with Daniel Santosh, Dina and their teams to scale Meta Compute. It's a good name. You need a name for a team. What did Ben Thompson have to say about this? He said a narrative seems to be quickly spinning up that this signaled that Meta is finally going to get into cloud computing, competing with Amazon, Microsoft and Google. And Well, Ben Thompson, he doesn't really think that's going to happen. What exactly would be the point? Amazon invented the space and invested heavily to have a structural cost advantage in commodity compute. Microsoft had a natural customer base with enterprise. Google has differentiated infrastructure and an outlet to fully leverage their AI investments without fear of cannibalization. What exactly would Meta bring to the table that other hyperscalers do not have? Such that they would have the confidence that they could be anything other than second choice competing on nothing but price. So asking the hard questions about where Meta Compute goes, they don't need to sell it if they truly scale they've been what I mean wasn't there compute budget in the pre AI era? Tens of billions of dollars in capex and that was just for reels just for ads. There's plenty to do if they don't choose to sell it anyway. MongoDB Choose a database built for flexibility and scale with best in class embedding models and re rankers mongodb has what you need to build. What's next?

We have the CEO of Mongo coming on in just five minutes. In other news, Beijing has told Chinese firms to stop using CrowdStrike if you didn't have a reason to start using CrowdStrike this is.

Is this the TVPN effect? They saw us run a couple ads over in Beijing and they were like yeah, they're too powerful.

Too powerful.

They're too powerful.

Apparently this is already very tiny business for CrowdStrike, so not super impactful. But in other news, OpenAI's number one hater in the entire world Nick is.

Back on the timeline.

He was the Hater OpenAI Hater of the year last year. Yeah, he's probably he's really gunning for it again this year.

His camera is all just pictures of multiplexed his camera rolls just all Sam Altman photos for sure.

But the News OpenAI Broadcom AI chip.

Delayed and to says this is not a delay. It's clearly a 5D chess move where Sam is amassing enough time to release state of the art GPT6 and put in Google Anthropic out of business.

Everyone's having fun trying to Find trying to find where this was actually reported.

Yeah. Sam Altman also is putting together. There's an announcement from Ashley Vance about OpenAI and Sam Altman backing a new bold take on fusing humans and machines. It's called Merge Labs. We've heard about this before. We talked to Rob Toews on the show about BCI and brain computer interfaces potentially being a big trend for this year. Obviously neuralink has been cooking for a number of years. Sam Altman also has his bet with Merge Labs and they now have $252 million in the bank and an all star crew and superpowers.

And remember, Sam Altman wrote in 2017 a piece called the Merge. I'll read in short, a few paragraphs from it. He said, more important than that, unless we destroy ourselves first, superhuman AI is going to happen. Genetic enhancement is going to happen. Brain machine interfaces are going to happen. It is a failure of human imagination and human arrogance to assume that we will never build things smarter than ourselves. The Merge can take a lot of forms. We could plug electrodes into our brains, or we could all just become really close friends with a chatbot. But I think Emerge is probably our best case scenario of two different species, both want the same thing and only one can have it. In this case, to be the dominant species on the planet and beyond. They're going to have a conflict. Imagine if dolphins. Imagine if Benioff has been working with the dolphins to secretly help them merge and they're like in a. In a crazy dolphin human race to see who can merge first. Because sent dolphins created neuralinks can just.

Like generate B2B SaaS product ideas constantly.

So funny. X has this, this new thing repeatedly. AI always is very good at being funny, unintentionally. But a bunch of people were posting about Benioff's story yesterday, talking about. And so X summarized it as a headline. Marc Benioff recounts dolphin vision behind Salesforce's start.

Dolphin Vision.

Dolphin Vision.

Sam has some crazy old posts. I was looking at this one from February 15, 2023. This is in 2023. It's $30,000 to get a simple iPhone app created and $300 for a plumbing job. I wonder what those relative prices will look like in 2028. The likely coming divergence between changes to cognitive work and changes to physical work could be quite dramatic. And I mean, I saw Nick was showing us an iPhone app that cost about $100 of credits or something like that. Maybe a little bit less than a vibe coding app. You know, you're going to use credits to Inference, a model that can actually build an iPhone app. But we're definitely closer to 300 bucks than 30,000 at this point. It's remarkable. Sentry Sentry shows developers what's broken and helps them fix it fast. That's why 150,000 organizations use it to keep their apps working well. We have our first guest of the show, CJ from MongoDB in the restream waiting room. We won't keep him waiting any longer. Thank you so much for taking the time.

What a setup to jump on the show.

How are you doing?

I am doing fantastic, man. It's like there are 700 people here. San Francisco is back. I feel like San Francisco is back.

Amazing.

And we are reintroducing MongoDB in San Francisco today. So Phil, the energy is just amazing.

Fantastic. Congratulations. Take us through some of the announcements. What does it mean to reintroduce MongoDB? What are the misconceptions where people might think of it as one thing and now you need to reintroduce it?

I would say MongoDB. First of all, database market has been around for a long time, 50 plus years, maybe 60 years. Oracle is going to celebrate its 50th anniversary. Year and a half. That was my first job out of college is I joined Oracle. MongoDB was a truly disruptive database that got created in 2007 and in New York City. I mean this is one of the few tech companies that got created in New York City.

Put it on the map.

I know, I know. And then the leaf, you know, people ask me what is the leap of MongoDB all about? And the founder said, hey, this is a natural way a software engineer or a builder should interact with a database versus a rigid way of interacting with database. So it's pretty cool. I just became CEO a couple of months ago, came from Cloudflare and I'm having great time meeting innovators, customers around the world. On the misconception. Here's what I would say that even though the database was created in 2007, the whole idea behind it was speed, agility and dealing with a lot of unstructured Data like emails, PDFs, images, videos, log files, this and that. So when the founder created this database, it was not with AI in mind. And today when you look at any AI application, your favorite consumer application you may be using on a daily basis, you see there is images in there, you may have create a video in there, lots of interaction. This database was created for the air Wave. And we want builders building on MongoDB as they create agentic applications, or any applications for that matter, that has structured data so that they can move fast and hopefully create the big iconic businesses, legendary businesses in Silicon Valley.

On top of MongoDB, where are you seeing the most excitement and growth in your business? I can imagine that there's a ton of greenfield projects that are happening in large clients, large enterprises where they want to move even faster. So they're grabbing MongoDB in addition to other modern tools on a new greenfield project. Then there's also a new startup boom of AI enabled software companies. I'm sure that's growing as well. What have you been seeing? Where is uptake the fastest in the last few years?

I would say John, what I learned, and I can feel it right here on the floor, is that you had pretty much customers. We have 60,000 customers, 62,500.

That's fantastic.

And our customer growth has accelerated in last 12 months, which is pretty cool because a lot of developers are now building on MongoDB. Yeah, nice. I like it.

Every time you say something great, I'm going to hit that. So fair word.

All right.

I love it. So one is our customer base re accelerated as developers are building on MongoDB. But it's a tale of two cities, right? One is large enterprises. John, as you called out, they are growing still with MongoDB, they have some mission critical across every industry. Financials, health care, government. I can tell you many across the globe just building on MongoDB and you have AI native companies or digital native companies, whether they are in San Francisco, Seattle, Tel Aviv or even New York, they are building on MongoDB. So the customer growth is coming from there. And customer growth accelerated in last 12 months for us, which is really cool. Many, many companies, when they start decelling they do not re accelerate and we re accelerated. So that gets me really fired up and really positive. And the innovation that we announced today that if you want to truly build an agentic system at scale that doesn't hallucinate, you can build it on your laptop. You always require a great database. You need an awesome agent rag or you know, semantic search that just works, that is accurate and this is truly a mission critical app that an insurance company may float it to their customers or a government may float it to their citizens or whatever the case might be. We want to make sure that this is a great data platform. So we announced voice for embeddings. We created a small embedding model called nano which is 18 the size of large. It shares the embedding space and works with pretty much on cloud, on prem multi cloud and Jensen talked about at ces the whole thing around multi cloud, multimodal and multi models.

And we do the same.

So very cool. Excited. We gave that message this morning to the builders here. We have a full show all day today, which was also live and people are fired up.

How are you thinking about M and A this year? Obviously, the company's doing really well. You're a public company. You have a public currency for acquisition. There's a lot of different pieces in the landscape that could be synergistic. But what's your thought process for partnering with other startups, either in a full acquisition or another partnership? How are you thinking about it this year?

So, John, I'm a big believer in organic growth. I like we have a large market that we participate in. We have only penetrated 2.5% of that market. 100 billion. We are at 2.5 billion. So we have a lot of room with our core data platform. Now. We are always looking for specialized skills, say maybe in the AI area. So in 2025, we bought Voyage, which was created out of Stanford. Amazing people. The founder teaches the legendary course on machine learning. So we bought that company and it got amazing, amazing engineers and researchers. And that professor is now Chief AI Scientist at MongoDB. So we will do those type of acquisitions where they are small, either for a technology or people, or sometimes both, so that we can truly leverage and accelerate the pace of innovation.

Has anything in the various compute bottlenecks been a headache or kept you up at night? I mean, we're reading about Amazon Web Services is paying for copper refining because they need to go so deep in the supply chain that they need to go to the mining level. Memory prices are through the roof. There's delays on all sorts of custom silicon chips. I know you might not be affected, but how are you thinking about the different bottlenecks in the cloud computing supply chain?

I would say the bottlenecks. If you remember a couple of years ago, everybody was worried about will they get enough GPUs from Jensen. Everybody's like, oh my God, bro, are you going to get enough GPUs from Jensen? And that was like a big worry. And now when we speak to customers, that's why they are like, okay, will GCP or Amazon, like you said, AWS and others, will they have enough capacity right now? It is not something that people are panicking. And Amazon or Google, I would say they are trying to stay ahead on that curve so that it doesn't become that. Hey, CJ, we cannot run MongoDB for your favorite customer X because we are out of capacity, man. We just can't do that. So we are trying to definitely work with them, understand how they are dealing with both compute as in CPUs as well as GPUs. And those innovation cycles are also very fast. But right now we are relying on them to solve. But we are not worried about it because they haven't given us the signal that we won't be able to do this.

Well, congratulations on the event.

Jordy, do you have anything else and the new role? I know we're out of time, but I'm sure we'll have to.

Yeah, I'd love to dig more into your background and tell the fuller story, but we know you have a busy day today. Congratulations and we will talk to you soon.

Give our best to everybody at the event.

Thank you, Jordy and John. Appreciate it.

Have a great rest of your day.

Goodbye.

Lambda Lambda is the super intelligence cloud building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. We have Matthew Prince from Cloudflare joining in just a few minutes. In the meantime, we will go back to the timeline.

There is some massive news I did on Merge. I wanted to say a couple things. So Mikhail, who's one of the co founders of Merge, says tldr. We're developing a new paradigm for BCI using molecules instead of electrodes. Oh, interesting. If you're excited about this and want to contribute in protein engineering, synthetic bio delivery, immunology, ultrasound devices, neuroscience or data ML AI, we'd love to hear from you. So that sounds extremely fun. Merge is pursuing ways to make BCI much higher bandwidth, millions of neurons and less invasive. No device implants. Instead of electrodes, we're focusing on modalities like ultrasound that can cover large areas of the brain and on proteins that can connect these modalities with neural activities. And so again, quite a different approach than Neuralink.

Well, let me tell you about TurboPuffer, serverless vectoring, full text search built from first principles on object storage. 10x cheaper and extremely scalable. Our next guest, Matthew Prince, is in the Restream waiting room. Let's bring him in to the TV pen Ultradome. Matthew, good to see you again. Welcome back. Happy New Year. How are you doing? Good, how are you? That's good to hear. It seemed like you weren't doing that well a week ago when you Posted about a court in Italy fining you. Can you take us through the process? Was this a surprise? How did everything lead up to your post which went mega viral?

Yeah, first of all, not a court, quasi governmental. And let me give you a little bit of background. So in Italy, there was concern by a bunch of the football clubs, so the soccer clubs in Italy, about online piracy of the matches that were there. And in response to that, the Italian Parliament basically designated a quasi governmental organization called agicom that they would have the ability to essentially allow any media company executive to put any website they didn't like on a list. And then initially, just the ISPs in Italy had 30 minutes to block access to that website. But they then amended the law to be able to include other companies, including Cloudflare, but also folks like Amazon, Google, Microsoft, anyone that is doing anything that is cloud related, they have gotten swept up in this. And so the requirement was that when a website was put on this list, that we would have 30 minutes to remove it globally. In other words, this small group of media executives got to choose what you could watch in the United States. And we already saw significant exposure of this beyond just illegal streaming, where we saw political content being removed. They accidentally blocked all of Google Drive at one point. They accidentally blocked the all, all of Cloudflare's network at one point, which broke the Italian Internet. So they turned that off pretty quickly. And so we had been in litigation in Italy to basically say, this doesn't make any sense. And what particularly doesn't make sense is we hate the illegal streamers. They cost us money. We don't make any money off of them. They waste our bandwidth, they take up resources that we could have for paying customers. And so we have a huge team that is trying to shut them down. We actually won a court decision right before Christmas that said that we could get all of the internal documents on how this commission was created, what its rules were, everything else. And then, you know, probably just coincidentally, of course, immediately after Christmas, the first meeting they had, they imposed a $17 million fine on us and said that it will continue to escalate if we don't pay it and register for this scheme and commit to let them censor, censor the Internet, which obviously we're not going to do. That, by the way, is 2x our revenue. In Italy, they have interpreted the law as saying that they can charge up to 2% of global revenue for any company that they find. So again, the insanity here is pretty high. And yet that's not even what the law says, the law says. The law just simply says it can be from tens of thousands. And in fact, there's parts of the law that they relied on that are so old that they were actually quoted in lira. So it was back before even the EU existed. To give you some sense of how crazy this is, even the EU has said that this law, they have significant concerns over this law. And so we're in a strange situation where, you know, we're getting a lot of pressure to literally just turn Italy off and not let them connect to cloudflare's network, which would, you know, unfortunately, everyone saw a few months ago when we had an outage, how big of an issue that is. That doesn't seem right.

You know, putting your tinfoil hat on. What is there some. Do you think there's some local group that wants your Italy business that says like, hey, that's like, is it purely censorship own goal?

I think that there is. So I think, first of all, it's a bunch of older football owners who don't understand how the Internet works, who just are saying piracy is a problem. And instead of doing what most of the leagues around the rest of the world do with us, which is work with us in order to identify what those illegal streams are, and then we not only take them down, but we actually put up a page that's an advertisement for the correct stream in its place. Instead, they were just like, we're going to just make it so that you have to take these things offline. Damn the consequences. I think there are other people inside of Italy that see this as an ability for them to be able to control some political content that they may not like. Obviously, Italy is a contentious political state, and being able to shut down things you don't like off the Internet is a very powerful political tool. So I think that's the tinfoil hat hat side. But I think mostly this is just thuggish behavior by a bunch of people who don't understand how the Internet works.

Ben, take off the tinfoil hat, put on the Steel man helmet and try and Steel man for me, where the work should be done of fighting piracy on the Internet, because I have some empathy for a league owner who doesn't want their content streamed illegally.

And it seems to be such a hard problem. My view is UFC kind of just threw in the towel and was like, the pay per view model is over. We have.

We're going to Netflix.

Paramount.

Paramount, yeah. And so, yeah, it seems like it's a nice thing for you to do to take down these illegal streams. Maybe there's some legal reason that that falls on you.

It's a purely, it's a financial reason. It's like when, when the streams come up, they, they clog our pipes and that shuts down the ability and we have to pay for that bandwidth.

Yeah.

So we have all the financial incentives.

To shut it down.

It is a hard problem.

Right.

Because again, the streamers are clever and they're always bouncing around and doing different things. But we shut down hundreds of thousands of illegal streams every single day and we work with most of leagues around the world, so. So there doesn't need to be a law in order to create the incentives for us. What I do worry about though is anytime that you create some sort of rule in the name of privacy, in the name of protecting children, in the name of whatever, that lets a local government and whether that's a state government or a nation state be able to set how global infrastructure works. That just seems per se bad.

Right.

You shouldn't be able to say Italy decides what you can see and watch in New York or California or Texas or Canada or China, frankly. And China shouldn't be able to impact like let China be crazy inside of China. And they can have their own rules and regulate their networks however they want. But the fact that this has extra territorial application that, that literally some, again Italian, basically soccer club owners can dictate what content you watch in California or New York, that's insane. And that's something that we have to push back against.

Yeah. So what's the response been like in Washington, in D.C. i imagine that at some point, even just a passing statement from the admin, hey, Italy, back down on this, we'll, you know, buy a little bit more wine or something, some trade deal, some sort of thing, or.

Not impose 100% tariff on all your goods.

That too, yeah. What are the levers that people are considering pulling what has been pulled? How are the discussions going?

Yeah, so I spent the beginning part of this week in D.C. meeting with everyone from senators to the United States State Department, the number of different officials within the administration directly in the White House, ustr, which is the United States trade representatives. And I didn't come across a single one that wasn't disgusted by this policy. And so like, I don't think the Italians quite understand the hornet's nest that they've stirred up. This is, this is now sort of the cause celeb for Europe. Europe using US tech companies as their piggy bank where I mean the Crazy stat is that Europe makes more off finding US Tech companies than they do off taxing their own technology companies. Companies. That's insane. Right. And so something has to change here. And this is the most clear example that again, you know, whistles to all the dogs in Washington right now saying, like, we can't have, again, Europeans, let alone, you know, this sort of Italian cabal being able to dictate what is and is not online.

Yeah. What's the reaction been from, I mean, your competitors, there are other content delivery networks, other hosts. Do they agree with you? And they're just less outspoken because they're not in the founder mode that you're in, perhaps. Or are there some other companies that are just less affected or less targeted because they aren't as much of a peer play? How is the rest of the ecosystem reacting to all this?

Yeah, I think the hyperscalers are really concerned about this. I think anybody who has a really large business where they have, you know, millions and millions, we have tens of millions of customers. You know, I think what's unique about us is we have a free version of our service because we believe that every small business, every developer should be able to have robust cybersecurity tools, robust content delivery tools. And so that. That gives us a much broader portfolio than some of our competitors that only have, you know, they won't even pick up the phone for less than a million dollars like that. That's a. So it's a. It's a little bit of a different thing. But if you're in aws, if you're a Google Cloud, if you're a Microsoft Azure, if you're an oracle, like, they are deeply concerned that this is going to impact them because they all have broad sets of customers, and it's very difficult for them to just say, we're going to let Italy pick who can and cannot be a customer of ours. That's, again, just insane.

Okay, last question on the fine for me, and then I want to move on to AI since we have you here. What's the result of the fine? Did you just pay it or are you debating it?

No, no, no. So we're, we're challenging Italian law because it's. Because it's. Italy doesn't have any. Any mechanism to stay this. I would imagine that there are going to be some, Some pretty pointed phone calls from senior U.S. officials, senior Italian officials saying, like, cut this out.

Yeah.

In addition, you know, we are considering various options so we may shut down our free service in Italy, which, again, is unfortunate. It'll hurt a bunch of developers there. But we, we can't do charity for places that. Again, the other thing that's coming up is the Milan Cortina Olympics, where we provide tens of millions of dollars of free cybersecurity services. I was actually a personal guest of the former president of the IOC to the Paris Games because we provided the information that took the terrorist attack that almost shut down the Games and allowed them to arrest 90% of the terrorists beforehand. We helped stop massive denial of service attacks, again for free, that were launched from Russia, which is trying to completely discredit the Olympics because of the doping scandals and bans that they've, that they've faced. And so, you know, again, I really believe in the Olympics. I really want the Olympics to succeed. But again, it seems wrong for us to be providing tens of millions of dollars of free cybersecurity for a showcase event in a country that's just acting completely insane.

Yeah, that makes a ton of sense. Talk to me about the 2026 App Innovation Report. Talk to me about this idea of a technical glass ceiling. I've been going back and forth with a lot of the AI researchers that come on the show. The models are amazing. And then I'll ask them, like, well, this particular app doesn't seem to be getting better on my phone. What's going on over there? Why isn't this app significantly improved? Why are there still bugs in enterprise software when everyone has access to frontier models? What are you seeing in AI adoption? What are you feeling internally? Just take me through your current thinking.

Well, we are incredibly bullish in AI. We use it internally. We use it across all of our teams. We actually had our company wide kickoff today and our CTO challenged every one of our teams to say, how can they use AI more? And it's really quite incredible what, what it's done. What I will say though is we're, we're definitely in some places, for some companies hitting some sort of a limit, it seems, in what they're able to do. And the exception to that, I think, has been Google Gemini has continued to outperform the rest of the market. And I think that the answer to why that's the case is actually not about the chips. Obviously, they've designed their own chips. I don't think it's about the people. They have incredibly smart people. But so does Anthropic, so does Open. I think it's actually about the data. And we are one of the few companies that actually has a view into this and so we haven't shared this very publicly, but I guess, you know, no time like the present. And so how much more of the web would you guess that Google sees and is able to train their AI systems on versus OpenAI?

Yeah, I'd probably say 50% more.

Yeah. The answer is actually 3.2 times more. Wow. Which is insane, right? That means that for every one page that OpenAI says, Google is seeing, you know, 3.2 pages, so there's just so much more data that's going into it. And again, OpenAI is the next closest. After that, Microsoft sees Google sees 4.8 times more of the web than Microsoft. They see similar to what they see with Anthropic as well. And it just drops off from there. So what I worry about is that because Google has this unique access to the web that nobody else has, that the game might just go to them. Because I think that at the end of the day, whoever has the most data wins. In the era of AI, I think what a lot of these companies are seeing is they're hitting kind of a data limitation. And one of things we're really thinking about is how can we make sure that we level the playing field. Either bring Google back to where everyone else is, don't let them leverage search in order to get a unique advantage in AI, or how do we lift everyone else up to the same place where Google is, make sure that everybody has access to that same data. If we don't create a level playing field in terms of the data, I do feel like there's a real concern that Google is going to run away with this and no one else will catch them.

Interesting. Yeah, we've noticed that. I mean, we test all the different LLM apps, and even just from a branding perspective, when you're in the Gemini app and it says doing a Google search, all the other apps will say we're searching the Internet. But doing a Google search just gives. The brand is like, oh, it's doing what I would do if I were on the computer. And it's interesting to hear that there's actually more data being accessed. It's not just the brand.

It's also one of the only things that Elon and Sam can really agree on or shared motivation is making sure the AI really competitive with Google.

I mean, people, people forget that OpenAI was started because Elon and Sam got together because they were so concerned about what was happening with Google. And I think that that concern is rightly placed. And Google has such a data advantage where because of search Everyone has left them behind their paywall. Everyone has let them see parts of the Internet that no one else sees. Again, we can look at the restrictions on various spots. What are the robot txt files and everything else for a huge portion of the Internet? And the answer is that Google just has privileged access. And until we level that playing field, either by bringing Google back to the same level as everyone else, or bringing everyone else up to the same level as Google, I worry that this is their race to win. And that's bad. The world certainly shouldn't have one AI company. I don't think should have five AI companies. I think we should be playing for a world where we have 500,000 AI companies and we make it as easy as possible for any of them to thrive and survive.

Okay, talk to me about the defaults on robots txt, where people are allowing scrapers, where Googlebot can come in, where Gemini can come in. What's been your thinking on defaults? How has it evolved? And then what's the reaction from your customer base been to those defaults?

Yeah, so I think this is actually a really nuanced issue and the answer is going to be different for a number of different companies. And in some cases we don't know what the right default is. So here's, here's where I think it's the things that are very straightforward. Excuse me, if you have a knowledge base. So cloudflare has a knowledge base on how to use Cloudflare workers in order to build systems. Excuse me, tickle in my throat.

No problem.

If you have something like that, then that's. Then of course you want that to be in AI systems and that should be default scrape, no matter what. On the other hand, if you're a publisher which is ad or subscription supported, maybe you don't want that to be in because AI bots don't click on ads. And if they're taking all the content that you're generating and then just regurgitating it in a place where you don't get any benefit from that. That's a real problem for the business model of media companies. And again, I think what we're seeing is the media companies have really locked down who has access to their content. And you're seeing much better deals from the likes of Conde Nast, Meredith Reddit, who are getting more for their content from the AI companies. Because at the end of the day, whoever has the most data is going to be the winner. That's there. What I think is somewhat unfair. And again, back to the Same issue is Google because of search is still getting almost everything for free, whereas everyone else is having to pay for it. That seems wrong, that seems like something that we need to fix in some way and again that either we need to either bring everyone else up or bring Google back down to everyone else's level. But that's, that's one of the things we're worried about. I think the most interesting thing that people aren't talking about today, it's what's going to happen to retail, what's going to happen to small businesses that are out there. AI is this massively consolidating force. And if you look at take three of the biggest retailers in the world, Walmart, Target and Amazon, what's interesting is again these are hugely well resourced companies, smart people working at them, no lack of financial resources across them to do whatever is right. And they've come to three completely different conclusions on what you should do about agentic commerce. In the case of Walmart, throw the doors open, open it up to the catalog, let every agent in. And in the short term I think that's to going that's clearly the right answer. But I worry over time that the agents are going to just keep trying to figure out how can we cut Walmart out over time. And I don't know what brands mean in a world where you know your agent is shopping on your behalf. Brands are a way for humans to have shortcuts that understand quality and value that goes away in a world of agent commerce. Agents will figure that out for you and report, report back and brands might not matter as much. And including a massive brand like Walmart, Amazon is taking the complete opposite strategy. They're literally suing Perplexity, saying your agents aren't welcome here again too well.

But that's because they have a massive advertising business and they know that the agents aren't.

Well so does Walmart.

Yeah, but it's not how material is it for Walmart versus Amazon?

It's a pretty big like there's any of the sponsored listings on Walmart, they're all doing the same stuff.

Stuff.

I think, I think I would, I would argue it's something different. I think Amazon thinks that they have to win the agent war themselves and so Alexa has to get much better at it, whereas Walmart doesn't have as much of a stake in it. So this is a giant bet I think from Amazon that their agent will be able to shop better than anyone else's. That isn't my experience with Alexa yet, but Again, a huge amount of resources that are going into that. Target has what I think is sort of the oh, let's split the baby kind of approach, which is agents are allowed, but they have to be the ones that are listed on the Target website. That obviously is not going to be a long term sustainable approach, but I think that that shows how challenging this is going to be. And what I really worry about is going forward, if you think about what small businesses that you shop with, your local butcher, your local flower store, those things, a lot of it is because of the interpersonal relationships that you have with those shop owners. If those go away, if your agent is doing this all on your behalf, if we all become some version of the Jetsons where how does George do his shopping? He asked Rosie the helpful robot to go do it for him and eggs show up and he doesn't really care where they came from. I worry that that's going to put enormous pressure on small businesses and I don't know that small businesses have the tools today to be able to figure out how to exist and thrive in this world of agentic commerce. And that's something that I'm enormously worried about.

Jordy.

I think, yeah, I think it's warranted. I still think at the end of the day with physical products, it's like, you know, I care a lot about, like, I'm drinking this here. I'm drinking this because our friend Andrew Huberman, this is the brand that he's aligned with. Like, if I just sent an agent to go buy me a yerba mate, maybe it comes back with something else. But then I'm still like, yeah, just get me the, get me the, get me the podcast in a can, please. Anyway, so a lot to figure out. The exciting thing I feel like is that every. So many different companies are reporting that LLM traffic is converting at a high rate, which tells me that it is a great tool for product research. But there is a lot of questions of like, yeah, how do you. GEO is kind of like a dark art right now. It's very hard to. It's far easier to kind of understand how you're showing up in results than it is to maybe influence it yet. But a lot of things to figure out. I did have one. I wanted an update on the ski season from you. I saw numbers traffic down like 20% year over year probably.

So one thing before. I agree we don't quite know how this is going to play out, but I do think it's important that we articulate what we're trying to play for. So I want to play for a world where there's not five AI companies, there's 500,000, where there's not a small set of media, massive conglomerates, but anyone can be a content creator. I mean, you guys are a perfect example of what I want to make sure continues to exist. As we get into a world that is more AI driven. And then for businesses, we need to make sure that you can be a small business and still thrive. There is a way to have a level playing field and that there can be tens of millions of businesses around the world that can actually do that. And that's what we're thinking about. That's what we're partnering with folks like Visa and MasterCard, American Express, PayPal, you know, Shopify, Adobe, Salesforce to try and figure out how do we give those tools. And I think that's one of the most important questions. The business model of the Internet over the next five years is going to change radically. Rarely does that something, you know, that big, have that big a change. And I think if you're not thinking about what that change is going to be, then, then, then I don't think you're worrying about the most important question today in terms of the ski season. No reason the skiing should be distracting you from thinking about what the future business model, the Internet should be. It is grim. I'm sitting in Park City, Utah. This is the worst ski season. I mean, I'm 51 years old. My first year of skiing, I was the 76, 77 season. This is as bad as that season.

No, it's good timing for you. Like a lot of people, you pray for snow. You were kind of like a bad ski season is kind of convenient for you right now. You need to be pretty locked in.

I'm trying to get Vail to sell me Park City and then return it to local ownership and make it the star that it is.

And I think so. So Vail is trading at less than. Less than two times revenue. We've covered kind of the history of the business. Like, at what point do you just, like, just buy the whole. Buy the whole thing?

I don't think it makes sense. I don't want to run Vail. That doesn't sound like much fun, but I think somebody should. And, you know, if it's, you know, if you could, if you could acquire Vail for, you know, $5 billion with, with some leverage, and you need about a billion dollars of equity, I know a guy who will, who will write you if you're an activist investor.

Well, that's what I was saying. One way you get your prize, you know, your trophy, Park City is you just buy Vail and then you spin it out. Spin out Park City yourself and then.

That seems maybe we'll get to that point. I would much rather somebody else who's like good at those things. That goes again like a, you know, Elliott Management. If they want to go buy Vail and then sell me Park City, that would be it seems like an almost. It's getting close to being a no brainer.

Amazing. Well, we're hoping to follow that story. Thank you so much for coming on the show. Good luck with the, good luck with the Italians and thank you for all the support coming on the show.

How's your Italian?

My Italian is not, not particularly good.

But he has a great Italian lawyer. I bet, I bet he's got a guy who's over there. Lot of good lawyers. It's full employment. We'll have a great rest of your day. Enjoy the rest of your and we'll see you soon. Goodbye. Thank you.

Shopify.

Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces. And now with AI agents, Delian Asparuhov, the co founder of Varda, the partner at Founders Fun. Hit him with the flashbang. Got him. We got to hit him with a smoke grenade too. Ask him to do an intro as the smoke clears. Throwing smoke, throwing smoke. And look who he is. It's Deli. It has the smoke clears. We have the partner from Founders Fund on the stream. It really takes a second to, to clear. Okay, okay, we're clear. We're clear. Delian, good to have you here. Hope you're doing well. Thanks for hopping on the show. How are you doing? How's your week going? You're ripping checks. You're in business. Money's flowing.

Is it?

Yeah.

Is the bear market over? You know, last year you kind of, you felt like things were getting a little crazy. Maybe you didn't find any opportunities you love. Then you started the year with a bang.

I feel like, you know, I, I think I'm just like addicted to investing over the holidays. I realized over the past like nine years that I've been in Venture. I think seven of those years I've signed a term sheet within 48 hours of Christmas Day. And I don't know if that like speaks to the fact that like I just like I'm addicted to work or something about Like Santa Claus makes me want to like send out some presents or something like that. But yeah, it definitely feels like a busy season. I was catching up with another investor and he's like, same thing here. Like one of the best deals we.

Did was basically over Christmas.

Is the deal flow from like serendipity at the holiday parties or is it just networking from the Internet? Like where are you meeting founders these days?

I admit that like, I think a part of it is if you were trying to preempt around, no better time than to do it over the holidays because like nobody's chitter chatting at the office, there's no like hallway conversation, etc. You can have these just like very quiet conversations. Founders like it because they don't have to deal with a bunch of noise. You like it because you feel like you can do this one on one negotiation and basically just set the deadline of like, I've now done this couple years in a row where it's like, yes, signed by New Year's, otherwise like deals off basically. And it feels harder to enforce that type of deadline like during the like, you know, school year basically. And so I think that's part of why on both sides I like it. It's like as a founder, you can kind of test the waters and if it gets shot down, whatever, nobody heard about it. Everybody was gone for the holidays. You come back into the office, your employees don't know, your board doesn't know, no VCs know. And if it does go, then like, you know, great.

I think, I think it's convenient. As a founder, the holidays can be so annoying because like it becomes like kind of poor form to just be like, you know, slamming people, your team with emails around Christmas Eve, Christmas with the email and yeah, so I think it's a win, win. Just obviously anytime you're talking terms, it becomes wildly distracting because you're like, this is maybe the most important thing for my business right now. But if you can use that stretch of like seven days where you're not really supposed to be like, you know.

You'Re not operating, but you're allowed to negotiate terms. By the way, I did want to point out, I did make sure that I wore my TVPN quarter zip today.

So you have a, you have a horse on that. Is that a little.

Yeah, we got another way.

Other way. Into the sunshine.

It's coming on. It's coming on on the screen.

Wow, you guys have a whole CGI.

Oh yeah, yeah. 2026 is a good year for us. Talk to us about the bimodal strategy early stage growth stage. Is this actually pushing you out on the barbell? Was series A your sweet spot and now it's seed and series C, is this going to be a permanent thing? How are you thinking through that?

Yeah, I think if you look at some of the best like deep tech, hardware, industrial sort of companies over the past like 15 years, they have a very interesting like PPs, you know, sort of growth curve where effectively you have like typically a seed round that happens. And I'm just going to give like you know, somewhat fake numbers here but you know, indicative of like you know, a buck per share, basically the like early incubation days, you then have like series A that'll be like four or five times higher, basically like five bucks a share. And a lot of times what you'll see is years of super de minimis, basically PPS growth until you sort of get to the other side of like the R and D production hell, etc. And then all of a sudden you just have huge step function changes like over and over and over again. And this is true on everything from like the early days of like Nvidia, Tesla, like Apple, all the way down to like some of the more recent, more deep tech, industrial sort of startups. And so what to do as like an investor basically? Well, you know, you can I think do the seed rounds where I do think they're attractive, risk reward, you get some amount of IR just like through series A, etc. And then what you effectively want to do is do that last round where it's flat and then allow it to basically, you know, basically grow, you know, from there. Yeah, and I think what I've like started to realize like that is the style of company that I like to do, it has a very different PPS curve relative to like SaaS, AI, et cetera, which are honestly just like much more like, you know, clear, smooth, sort of like, you know, typically obviously exponential curves. If it's like, you know, sort of great company. Right. Like if you look at ramp for example, there was just clear step ups all throughout. You obviously had to like, you know, they did a great job of like, yeah, you know, marketing themselves down in like the 23, 2022 basically, you know, sort of crash. And I think that was proven to them and then basically smoothly upwards basically since then. And so the way that's played out is like, you know, now over the past couple of years I've much more consistently done basically like super early stage PDF and a founder Basically like, you know, early, early, you know, industrial deep tech companies. And then I sort of sit there and bide my time and wait until what I think is like that basically last round right before the inflection point and then basically sort of do do that round. And I've now done that over the past year. I did basically three, like later stage basically rounds, like, you know, series B, series C and then like, you know, basically series C that I just closed terms on over the holidays. And all of them were companies that I either been tracking basically since seed or literally been an investor since Seed. And we talked about one of those, you know, part of what we were talking about earlier today was, you know, talking about one of those very particular companies. But yeah, I think I'm starting to realize that I like that way more than I like doing the like, you know, sort of messy middle in some ways. And so, yeah, want to double down on that and you know, whatever. Tuesday night was cool. Where I literally felt that happen, where I signed two term sheets that were indicative of both ends of the barbell.

Was the amount of work in DD that you did on the Series C company significantly higher. Like how did you actually split the time up? Because this is framed as, you know, 5050 of Dalian's time. But I imagine that there's just more data to pull in, more research to do, more customers to talk about, but.

At the same time, you know the company, so. So much better.

Sure, sure, yeah. So what was it like, process wise?

Yeah, I mean, I'm not somebody, I don't think I would ever be able to do like a raw, serious C where it's like I've met the company for the first time and have the time to go do the diligence, et cetera, from scratch. Because it's just like, dude, I've got a fucking company that I gotta run. Like, I can't do that. The thing that I can do though is like get involved early or if it's something that I'm like deeply passionate about, even if I'm not involved, like track the company super regularly such that I'm kind of doing all that pre diligence basically, like ahead of time. So, like, let's talk about one of the, you know, some companies that I think exemplifies this strategy, Hadrian, where we just announced a $1.6 billion post round led by T. Rowe Price basically last week. Maybe, you know, we're going to have.

The show, we'll bring it, we'll ring it twice for The Hadrian.

Exactly, exactly. I think it's Dallas, you know, fifth unicorn or something like that from like, you know, they see under 25 post all the way through. Yeah, yeah, feel good about it. Chris said I was texting him earlier. He's like, you're allowed to, to toot your own horn. I'm gonna teach you VC brags, you.

Know, hit me up.

Yeah, yeah, yeah. Give us the update on Hadrian.

Yeah, that's one where it's like, look, I got involved in basically like company formation. Basically like you know, sort of seed round where it was Chris and like maybe one or two sort of people. And so what did diligence on that one look like? Well, it was effectively like, you know, four and a half years of working with Chris very closely, going to every single board meeting, watching the company like adapt over time. And in like, you know, Q1 last year in 25, it was clear there was just this inflection point that I felt like I could appreciate but not everyone else was really appreciating yet. The first was they basically, you know, for those who don't know, Hadrian basically builds these like automated CNC basically, you know, facilities. They basically cut metal on behalf of, you know, aerospace and defense companies. So, you know, think a lot of the biggest logos in the founders fund portfolio won't list, you know, sort of all of them since I'm not sure which ones are allowed to be public. But you know, ones that, you know, build rockets, make drones, you can guess which ones those are. And they cut metal on behalf of those companies. For a long time they were just in the R and D production grind, trying to figure out basically how to make the economics of the business work. Switching from what were originally quick turn small prototype contracts into production multi year, basically large contracts. In Q1 last year, there were two things that happened that just made it very clear to me that it hit my Spidey instincts of oh, this is the point that I'm supposed to come in and do the series seed. My whole job was to have done the seed and then wait for this exact moment. The two things that happened were one, they just got through production hell. Basically, like the economics started making sense. They were producing sort of way more volume. And then the second thing was, we talk about this a lot, but the defense industry is obviously going through a huge shakeup and there's a whole set of neoprimes that are coming out that are going to get anointed. And Hadrian was clearly in this transition point from what was previously just purely like a back end sort of supplier of parts and a vendor into somebody that was becoming more of like a strategic neoprime that was going to be code bidding alongside others, but as a neoprime that is focused specifically on production. Right. So if we think about groups like you know, Andrew Saronic Shield I, you know, name your favorite, basically, you know, mock industries, name all your favorite, you know, neoprines. They all roughly follow the same model, which is like they are product developers fundamentally, right? Like they go and like design net new products. They may do a decent amount of production and assembly themselves, but they still are reliant on like out of sort of house sort of vendors for a vast majority of their like you know, subcomponents. Adrian fundamentally is a very different type of neopride, right? They are neopride that is explicitly focused on production. They do not produce basically like end products themselves. They only amplify others and products. But that is a significant bottleneck that the defense department or war department is looking to go unlock and is leaning on a Hadrian for a wide variety of programs, both directly as Hadrian as a contractor as well as like co bidders alongside other neoprot. And so I think when I think about like, you know, the way that hadrian was interpreted Q1, Q2 last year, people were squinting at, they're like, man, these economics have only turned in the last quarter. You know, isn't this just a back end vendor that doesn't have a ton of upside? From my like insider perspective, having watched it for five years, I'm like, dude, this is the type of business where once the economics turn a corner, they don't unturn a corner, you know, I mean it only gets easier from here. We finally got to fucking positive. We're going to 20% margin, 40% margin, 60% margin. And it's like, you know, accumulating advantage. Like materials get easier, the factories get easier and the states want to send the stuff. It's just like, you know, you're rocking and rolling. But like when people look at like a SaaS business, they don't appreciate that. Like yeah, in like hardware, especially when you're producing something that is like kind of a commodity, there's sort of like semi infinite demand. There's like not semi infinite demand for data bricks. You know what I mean? Like, you know, the scale of the market is not nearly as large.

And Ollie would put.

To differ.

I'm sorry, we're obviously investors of data.

Hey, also T row price is T rowe Price is smart enough to be in both databricks and Hadrian.

You know, to be clear, big data bricks, fan. But at the end of the day, the amount we spend on metal is more than, I mean I don't probably really know what databricks does, but like whatever they do, the amount that we spend on metal is more than whatever.

We spend on databricks.

Well, I mean reflect on the, on the AI boom, the slide hop versus steel debate. It felt like AI went through this sort of reset on the narrative towards the back half of last year.

Do you have room in your office for a victory lap or are you kind of.

But it feels like we're back to, back to normal. I mean there's multiple $10 billion deals announced already this year. Nvidia is acquiring stuff for 20 billion, something like that. It feels like the pace is, is just continuing unabated. There's no real reset there. How did you process like the bubble in venture, that discourse? Where, where do you feel like we are now? Where do you think about the new like fundraising landscape? The K shape dynamic between the large funds, smaller funds? Like what's your, what's your current outlook on the venture capital market broadly?

Yeah, I mean it's where like you know, I won't name names but there's like various venture firms that have had like funds that they have like half funded, done really great deals out of and some of these like, you know, sort of top tier names have significant markups even on those investments and still aren't able to like fully close out the fund. So it's like as an lp, you're able to like, you know, basically get in immediately, know that you're like, you know, LP position is going to be marked up and a part of it is that you're comparing it that versus like there's so many of these mega, you know, sort of labs, mega rounds where there are so many SPVs that are available that as an LP you're actually trading that against. Well, I could just go get the direction position, no fees or much more diminutive fees through like an SPV from like a lesser manager. And so you are seeing like, you know, if you look at like, you know, I think there's this like, you know, graph recently. They're just showing like you know, total venture dollars raised by venture firms significantly. Like it's still linearly basically down since 21, 22. And that was either shown, I think it was like on Carter or something like that. They were showing basically that data. But I would bet if you looked at like total LP dollars deployed into like tech venture, direct sort of investments, fund investment, etc. It is almost certainly significantly up since 2021. It's just like Mubadal at some point can access some of these mega rounds just as easily as they can access like directly into basically like some of the funds that they used to have to invest into. And so you're getting this odd dynamic where you know, we talked about this even last week a bit but like a lot of people are doing these like feedless careless spbs that are basically like branding exercises for them. So yeah, I think it's like it's a harder time than ever to be an emerging manager. Like you know, there's somebody that like, I forget who did this analysis but it's like it's taking this fewer emerging managers taking longer to close and then it's harder than ever to even as you go and you're like a multistage mega fund to still go raise the like equivalent of like the Andreessen $15 billion, you know, sort of total, you know, growth fund. It is not trivial for people to do that because the things you're going to go invest into in these late stage assets are now, you know, readily accessible directly via the, you know, LPs through all these alternate sort of vehicles.

And it's not like a huge secret what companies are the real winners. You don't need to pay fees.

It's not like there's like 100 logos, it's like there's like seven logos that matter basically are significant number of dollars. And so as an LP you're like, why do I care about paying these fees versus like I know I'm getting into the exact same logo anyways, but through this other vehicle.

So you have sort of like the private map 7 that everyone's excited about. LPs are getting direct access to. Is there any effect of the fact that the public max 7 are doing so well? I was, I was kind of laughing to myself like if I want exposure broadly to self driving cars and I buy Google and Humanoids and I buy Tesla and like there's a number and AI. So you buy Google again. It's like you can get a pretty broad portfolio of like sci fi bets just by owning the Mag 7 and maybe those get disrupted, maybe you got to pick up other big, big companies, but it's not like they're completely asleep at the wheel. Is there, is there a like a little bit of a magnetism from the mag 7 that might be pulling liquidity towards just public tech companies these days a bit.

But I would say also for most of these LPs, they have like, you know, and especially the endowments, they have like pretty clear ratios that they want to maintain across basically like public, private, etc. And so in some ways through the public market sort of accelerating, they have, if you remember in like 2022, during crash, there was like the reverse thing. Basically we're like, you know, basically all of a sudden the private books basically weren't, you know, getting marked down nearly enough. They're trying to represent like 20% of an endowment's book and they're like, we literally have to freeze all private investing. Now you're in some way seeing like, you know, the reverse where, you know, people in some ways counter swung so hard. We're now some of these groups that want to maintain like a, you know, 7% ratio on private are all of a sudden down at like three and a half and are trying to figure out basically where to deploy. That's where they're coming with these huge text to the table. But they're seeing that there's a multitude of options and they want a relatively limited set of names. And why not just go basically more and more direct into those names? As a founder, at some point you're like, well, I like these venture guys on Sane Hill Road. But at some point when I'm a public company, these are going to be the guys that hold my stock when I'm public. And so I don't mind prioritizing them basically directly and having them on board in some ways. To tie this all the way back to Hadrian, it's like, yeah, this past round Hadrian had a multitude of different offers from more traditional Silicon Valley venture firms, firms. But it's like, why wouldn't you take a T. Rowe Price where Tiero Price is going to be the person that likely just, you know, underwrites part of your ipo, holds it when your public, you know, scales with you while you're sort of public. And so you're seeing obviously, you know, you know, those types of firms come in much earlier into venture and then founders by default, you know, sort of prioritizing those. You'd obviously had the like wave in like 2021 where a bunch of the VC firms tried to become RIAs and enter into like the T. Rowe Price, you know, sort of lane that almost obviously turned out to be like a disaster where it's just like these guys are not sort of public equities experts and do not understand what it means to manage that book appropriately. So I feel like most people have pulled back from that. So do you think it's a shifting landscape where I just don't think that the current state of liquidity preferences that are given to. If you invest into a company of 500 billion in the public markets, you do not get a liquidity preference as an investor. I think it's less and less likely that some of these super late stage private companies keep giving that type of liquidity preference to investors. And then I think also the fee structure on these venture firms is almost certainly going to compress for some of these super late stage types of rounds and funds.

Do you think we'll see any SPACs this year? Well, specifically for companies in our world, obviously there's SPACs constantly treasury companies and.

Stuff going out, But Venture backed SPACs?

Yeah, I definitely have had now a variety of companies that I'm involved with have had SPAC sponsors pretty aggressively approach and knock on the door. I think there are certain companies where it makes sense. If you are a type of company where it's like look, I just need 500 billion or 500 million on the balance sheet.

Billion would be great.

I'm sure some of them love to have 500 billion on the balance sheet. But yeah, you know, and you're some super deep tech long cycle thing. Yes, makes sense. I think there's just enough scar tissues from the 21 time period where it's like there's a very small handful of companies that like survive through that spat. Crazy, right? Like there's that one robotics company that works with Walmart. It's like one of the few that has actually like, you know, sort of traded significantly up, but like the vast majority.

MP materials. MP materials.

Materials. A SPAC though.

Chamath. Spac.

I thought it was a chamath.

That.

Was like, I think almost like a part. Like it used to be public, went private, restructured and then like fortress.

Fortress value.

Fortress value.

I'm just, I'm just excited, I'm just excited to see what American Exceptionalism Acquisition.

Corp. Symbotic was the one that I was thinking of by the way. I'm pretty sure it's one of the few stacks from like 2021. Let's look at.

Here's one. Here's a SPAC candidate. What about that, that, that Moon hotel that was getting checked.

Moon Hotel.

I mean at $400,000 a night it's pretty simple. Just do the math.

Simple economics.

It's Just a math problem.

I like that they're working on that. Give them a couple decades. Let them cook, Jordy. Let the AYC founder cook. Thank you, Gabe.

In the chat. Not a math spec. It was a pipe.

Pipe. Okay.

Yeah, pipe into that one.

Yeah.

Symbotic is probably the best example of like basically stacked at you know, a looks like 8 billion dollar valuation now a 42 billion dollar company, basically like incredible outcome. But that's like of the basket of like this backstory in 2021 that's like, you know, one of like, you know, 15 that basically ended up like that. The basket going down. And so I think for most companies that that makes sense.

Yeah, it feels like you get to do a big fundraise, you get a bunch of cash on the balance sheet and then. And almost certainly you're going to have a stock chart that just looks like this, that follows you around and gets screenshotted in your face for the next like, you know, five years. And then you look at some of these spacs and it's like, oh yeah, they're up 40% this year. And it's like. But they were 10 times bigger when they went out. And so you have this weird dynamic. Even if it puts cash on the balance sheet, it seems like it's a lot of weird dynamic. Very different culturally. To be running a public company dealing.

With the stock price is symbotic. Just a super underrated company. Seems like a Ford sized business.

Wow.

I mean, yeah, it's large. I think it's pretty underappreciated. It's basically like, I would describe it as like the best warehouse robotics thing since Kiva, but it's like, you know, publicly traded independent company. I understand these like pretty darn significant like you know, rollout into all of Walmart, definitely not discussed enough. I don't remember like the venture investors, you know, that were, you know, you know, around the table, but it's like based in Massachusetts. That's a part of why it doesn't get like, you know, a ton of, a ton of interesting heat. But yeah, I definitely think it's like pretty under the radar.

Yeah. Well, thank you so much for taking the time to come chat with us. Have a great rest of your day. We will see you soon. Tactical nuke.

Oh no, this is me doing barbells for my barbell strategy.

So I'm gonna just do my barbecue.

Keep wrapping it up.

Keep repping the barbells, boys.

Goodbye.

He had to nuke him. The truth Restream. That's a truth nuke one livestream 30 plus destinations. If you want a multi stream go to restream.com going from the truth Nuke to the Restream.

Speaking of robotics, next up we have Deepak from Skilled AI. Got to give a shout out to Luke Metro. Metro joined a company. It's pretty under the radar. Left Anduril to join Skilled. Yeah, There wasn't a lot of chatter about Skilled publicly at the time. And of course, Luke is a goated picker and it's been marked up significantly many, many, many times since he joined, so.

But until the CEO arrives here in the Restream waiting room, let's talk about Reggie James. He's joined General Catalyst as their creative director. He says it's an honor to be trusted to hold this position. He had the opportunity to spend a lot of time with incredible folks across the industry during his break, but the ambition that General Catalyst has to tackle huge problems is simply unnatural. There's so much work to do to show the story of the immense world being built here, but I couldn't imagine doing anything else right now. I'm incredibly excited for this.

Super bullish on this. Reggie obviously is Oramax. He's motion maxed and. And I think if he can bring some of that to gc, it feels.

Like, I mean, GC is a massive fund, but a little bit quieter. They're not going direct as much. There's not a lot of noise and that just feels like more of a blank canvas than going into. If you put Reggie at a different firm that maybe has a whole bunch of other firm doing a ton of stuff.

Yeah.

Loud. You could. Yeah. This feels like a really natural fit and I'm excited to see what he does with it.

Yeah.

Let me tell you about Figma. Figma make isn't your average vibe coding tool. It lives in Figma so outputs look good, feel real, and stay connected to how teams build create code back prototypes and apps fast.

Billionaire rams owner Stan Kroenke becomes America's biggest private land owner.

For Stan, let's Stan, are you Stan?

I guess he's listening. He's listening to the show.

I'm a Stan. Stan and Stan.

Stan buying nearly 1 million acres of new Mexico ranch land.

Stan's land is growing.

That's fantastic news. Farmland investing becomes increasingly popular, especially now that maybe somebody comes in and wants to build the data center on your farm and pay 10 times what you paid for it. At 2.7 million acres, Koronke's holdings are larger than Yellowstone National Park. Or the equivalent, roughly 2 million football fields. He bought 1 million acres of new Mexico ranch land in December from the family behind industrial conglomerate Teledyne. According to the trade publication Singleton Ranch's transaction is the largest land purchase in the US in more than a decade. Land is back, back on the menu. Let's go.

Speaking of let's. Greenland is in the news, of course. Dylan says this is the equivalent of affordability in Greenlandic politics. And he shares a screenshot here. It's a quote from a Greenlander who says, I hunt whales and seals in the United States. They think whales and seals are cute and shouldn't be hunted. That's what I'm afraid of. Don't bring your fear of hunting. Don't bring whales.

I like this. He's looking for a carve out. He's looking for a carve out.

Let me keep hunting my whales. Let me keep hunting my seals. Hey, you know what, Abelson?

If you join up with America, can we work to get Canuck Abelson on the show to talk about the business?

So Canuck wants to stay with Denmark just because culturally it's cooler to hunt whales.

We can shift the culture we kidnapp.

I'm down to hunt whales. I'll go hunt some whales and seals with you right now. Let's hang out. What are you using? A spear? Spear gun, rifle, nuke.

What if he's selling the blubber out the back door to power data centers?

We got flashbangs. We'll throw a flashbang in the ocean. All the whales will be doing this. Oh, no.

And then we'll something about with the harpoon. I don't like. I don't like seals.

Yeah, Joe wasn't always reading. He's reading Moby Dick. We know about hunting whales in this country. We would be happy to partner with Greenland on spot.

I think I'm against whale hunting. They're just so majestic.

Majestic? That's a synonym for cute.

But seals.

You're the problem.

Oh, you don't like seals? I have deep beef with seals growing up surfing in the cold waters of Northern California. By the way, last time I talked about surfing, I got mega canceled on YouTube because I was talking about how surfing makes like brings your body temperature lower and then you spend all day trying to heat up.

People were like, you were against surfing before work.

This guy is in a suit talking about surfing. What does he know about surfing?

Well, you're about to get canceled by the seal lovers in the chat.

No. So seals will like mess. Would mess with me. I'D be out alone surfing in Northern California. Water is very sharky and these things are the dogs of the sea. So they'll run up and they would just slam into me and all of a sudden I think I'm getting attacked by a shark. And then I'd just be in the water by myself. Beefing with a seal.

This is an incredible skill issue for you. Oh, seals annoying me. Oh, I need seals to be hunted. It's fine. Hang out with the seals. Let the seals mess with you. They're the dogs of the sea. We love them. We also love Plaid Plaid Powers, the apps you use to spend, save, borrow and invest securely. Connecting bank accounts to be remove money, fight fraud and improve lending. Now with AI Buco Capital. He tried to warn you. He said, I tried to warn you. I'd say we're getting close to the bottom, but I have no idea. Nobody wants to own these things. He's referring to the problem with application SaaS isn't vibe coding. It's that vibe coding is only one of the many bear cases. There's no valuation support for many of these names. Free cash flow is a mirage due to excessive stock based comp. He is calling the top on SaaS boys. The SaaS boys are not happy with Buco. He says, speaking of stock based comp, the dilution at some of these companies is absolutely disgusting. It's insulting VCs.

Remember thousands and thousands and thousands of Snapchat. Snapchat gives away the whole company every 10 years.

Yeah. What happens to competition into margins, into future profits when the incremental feature is free to make? Many of these companies are zero. So you can see his prediction is you see a lot of different VC funded SaaS companies encroaching on each other's territory. Oh, I'm in this category. I'm just going to build this.

Yeah, the calculus changes when it's like, hey, this feature is going to take us like nine months to really build and be competitive. And it's like, oh, it's actually going to take us a month.

Two seconds. Seconds. It's done. It's not clear where the value will accrue. Investors are thinking of Netflix gathering all the value on top of cable, or OTA stealing all of the value on top of legacy systems of record. Last and certainly not least, even if you don't think application SaaS is a zero, everyone else seems to. So who's going to step in to buy these things? Probably private equity. One of them has to accelerate for real for the money to start flowing in. Yes, interesting. He could go on and on.

Anyway, Naruto in the chat says, I've done the swim from Alcatraz in sf. People running an event sit in boats and prevent seals from biting the swimmers seals.

Send this post to Greenland, let them know we know how to fight the seals. We know how to beat them back when our Americans are swimming.

Yeah, I mean the biggest mistake they're not making is flashbanging the seals. Flashbanging the seals they didn't study.

Yeah. Scientists have made a spray on powder that can seal serious wounds in seconds. Speaking of seals sealing wounds, the South Korean developed powder reacts with elements in the blood and turns into a gel to form a barrier to stop bleeding. This is not that new. There's coagulants that exist. We've talked to a few folks about this.

That's the company founders. Jake.

Has a similar coagulant. And then also there's quick clot from the late 90s I believe, but exciting to see more development in the healing of wounds.

Thoma Bravo has a piece in the Financial Times and they say broader integrated software platforms for business are positioned to win out against those selling specific services. The explosive rise of AI startups follows a pattern we've seen before. Small companies racing to apply new technologies to specific business problems. Let's give it up for applying new technology to specific business problems doesn't get enough love. Promising huge efficiency gains in their narrow slice of the market. These startups typically use standard AI models and distinguish themselves by writing software on top in order to target specific tasks. Their value proposition is that specialized solutions can compete without worrying much about how businesses function as complete systems. But this is a vulnerable strategy. Like.com companies that burn cash without sustainable business model models. Many of today's AI startups burn money paying for access to AI models While lacking advantages that competitors can't copy. Their core bet is that AI can solve specific business problems so efficiently that companies will willingly go beyond their integrated systems to get superior performance in one area. In short, best in class beats good enough, but all in one place. But connecting to full business process processes to find a sustainable growth path over time is very challenging task for companies built to do just one thing. While some might be seeing sharp initial revenue acceleration, they also see substantive costs. One study by Cruise Consulting suggests they spend double what traditional software as a service company spend on compute and infrastructure, while also paying for specialized talents that command high pay packages. These structural challenges show why we believe established software platforms such as Salesforce, SAP and Microsoft and Portfolio Companies like Anaplan and Coupa that we invest in at Thoma Bravo will be more durable despite the risks of AI disruption to their businesses. While AI startups struggle to defend narrow territory, broader platforms build advantages that often grow stronger over time. Rather than just making one specific area better, they work on strengthening the broader business of a customer. This model produces intelligence for both the company and the provider. Building capabilities and SaaS companies have years of making improvements. For example, software platforms can model cause and effects across business functions. A narrow AI tool might optimize an inventory, but if it doesn't consider cash flow, supplier relationships or supply chain risks, that optimization on a single variable could actually stand to hurt the business. Established software companies are thus better positioned to address business concerns on AI integration, competition implementation challenges, cybersecurity, legal ethical questions, costs and reputation risks. This is because digital transformation has always meant balancing efficiency gains with risk management. Trying to see if there's anything else here that's worth reading. I think the main thing is being in a position like Thoma Bravo where you're meant to be deploying tens of billions of dollars amid when you have the shifting sands is like pretty hair. Like there's a ton of opportunity. Very exciting time, but it's also harrowing.

Yeah. Well, let me tell you about Vanta Automate Compliance and Security. Vanta is the leading AI trust management platform. And we have our next guest. We are in our Lambda Lightning round. We have Deepak from Skilled.

How are you doing?

Good to have you on the stream. Thanks for taking.

Thanks. Great setup you guys.

Fantastic setup.

Thanks so much for testing it for the first time. And as you can see there may be issues in between. We do not know.

Yes, please introduce yourself and the company since it's your first time on the show.

Yeah. So thanks for having me. I'm Deepak Pathak. I'm a CEO and co founder of Skilled AI. So at Skilled, what we are doing is we are building a general purpose robot brain. Now, any robot, any task, one brain. When I say that it's like even a humanoid or quadruped robotic arm, hands, everything governed by a single brain. And that is what we do.

Let's get into the news from this week and then we have a bunch of questions.

Sounds good.

What happened this week?

Well, this has been a busy week for us. I think back to back we had multiple announcements. We began with the announcement regarding learning from human videos. We showed a lot of results. But the latest announcement, which is just from yesterday, is about the funding announcement.

How much did you raise close to.

$1.4 billion, which values the company now at $14 billion.

Incredible. Incredible. Yeah. Let's talk about where you're most excited about the application of robotics in the very, very near term. There's so many different obviously angles, industries kind of form factors that are exciting. But where are you guys focus on today? Working with customers?

Yeah, I think this is. Whenever in public folks talk about robotics on the consumer side, the excitement becomes okay, we have robot coming to our homes next month and my what will do X, Y and Z. And people don't realize that robotics is not a new field. This is one of the oldest area in technology. It precedes computer and in fact it precedes AI. For 70 years you see all these movies and even real world demos of robots, but still we do not see robots around us. And we began to feel that robotics has not arrived yet. But if you go to any, go to any big factory, go to car factory or Amazon, you will see like tons of robots, thousands of robots doing things, manufacturing items, etc. Now what's the gap? The gap is you cannot get robots beyond these demos or beyond these factory settings to consumer scenarios because the brain has been with and that's what we bring to the table. Now in this setup, it is a step by step approach. Even in factories, even in enterprise, the robots are there, but they are in a red tape area, like behind an area where you cannot go. And if a human goes there, they have to turn off all the robots. And that's where you start first. You start to get these robots alongside people. So what we are doing right now, even we are already deploying in variety of areas like point to point delivery systems, warehouses, manufacturing. But the change is we are doing it alongside where people can come and interact and do their own work and the robot will be working aside with randomized scenarios, people is important because people create randomness and chaos and that's where AI needs to come in.

Yeah. How are you feeling about the supply chain? Are there any bottlenecks that you're worried about coming down the pipe? We seen so many unexpected bottlenecks pop up in just the data center build out. Talking about copper today, that was not on my bingo card. What should we be worried about if we are going to be scaling robotics over the next couple of years?

Yeah, I think this is exactly why we are scaling the way we are. If you think about going like for every different application there's a different hardware requirement. If you are doing something for data center, you need a different Kind of robot for doing things and specific other scenarios. And this is where getting all the supply chain all together at once, it takes a long time. But as I said earlier, robots are not a new thing. Industries have already been deploying them for a variety of scenarios. The ROI has been less because you have to create a separate area, separate warehouse, customized scenarios for them so we can piggyback on those supply. So what the model we follow is we partner with hardware companies or even end users who have hardware and we provide our software as a service on top and that enables very fast go to market strategy for us. And these robots, which may be earlier in a secure area where humans cannot enter, now, can be used alongside people and start to work.

How much hardware do you guys actually build internally specifically for testing? I can imagine there's certain kind of products where you're working just on the software side, working with a specific custom, but then on the internal side for like, you know, having faster feedback loops. I imagine you're building quite a bit as well.

You can today. What has changed the last five to seven years is you can purchase a lot of commercially available hardware. And in fact, every single robot hardware probably you've seen on the Internet has made its way to our office. And we already have large number of them as long as the company sells and gives us access to fully control the robot for software. So we take the hardware, we put the model to the end to end model goes directly from pixels in to actions out, and it takes control of the whole hardware and makes it a skilled robot.

What form factor do you think, what robotic form factor do you think will be the first to make consumers feel like robotics has really arrived? Right. You talked earlier about how it's like commonplace in commercial setting, but it's happening like in a warehouse. No one's really seeing it. And so they're not kind of like feeling the acceleration yet. But on the consumer side, when do you think people will have that kind of like, moment?

This is a very, very, very good question. And I think it goes to the root of what people think of robotics when they say, okay, I order one in my home. Okay, let's say the most common thing in Silicon Valley, like, I want robot in my home in this year. But when has this happened that you suddenly have things at home and you do not see them in your daily lives? What about groceries? Like, do you see robots in grocery store? What about Target? You go to, you order something, you waited for two hours for it to get picked up, ready, and then it's a pain to get even pick it up. What about hospitals? What about any of these consumer facilities which deal with consumers but are on the enterprise? And that's where you believe you will begin to see robots. First it will become a commonplace thing and then you start to get them in your robot. Like if I give you a robot to you like today, you will probably use it and you will like it for a few months. Then it will sit in your garage like a massage chair. I don't know if you have. And then never get used again. So it has to first come become common in your daily life that you're not surprised. Okay, I have this in my home and it's ready to go. And that's where we are. Enterprise is a stepping stone to the eventual consumer deployment. And yeah, I can go deeper into this. What is the other reason for this? But this is one of the practical reason. The other reason is data. And that you can go for going to.

Yeah, what kind of bottlenecks are there on the data side that, that you guys are seeing?

Bottleneck as in when there is something in high quantity and it's coming to a squeezed hole. But there is no data for robotics. And this is the, this is the main issue. It is not one of those AI areas like as I said robotics was when the initial AI projects were done at MIT and other places, they were in the context of robotics, it has been 70 years. Chips have. The gram has come down from a room size to like cannot even see that. So tiny transistors, robots are still not around them. And the reason for that is unlike language or vision where you can build GPT like models, LLM like models, there's no data for robotics. And that's where the technical part becomes very critical. So what we do today is we bootstrap. You have to somehow bootstrap the process process to deploy the robot on a site. You cannot say I'll get a data by deployment. But to deploy, I need a problem. I need the brain to function. For the brain to function, I need data. So how do you break this problem to bootstrap that what we do is we start with two such alternate sources. We learn by watching people. So learning by watching people act like there are tons of lifestyle videos on YouTube and Flickr sources, we watch them and have the robots learn from that. And the second source is simulation and put them together to transfer these results to the real world. And once they get deployed, then we can establish this data privacy where the more deployments increase, the more data you get back. And that increases further the performance of your model.

Very cool.

Makes a lot of sense.

Well, thank you so much for.

Yeah, great to finally meet.

Great to finally meet. We're huge fans of Luke Metro. He's a great pickup.

Yeah.

We'd like to nominate him for Employee of the Month. Employee of the Year.

We don't know anything about the other employees.

No insight.

But he's the employee of the month in our month.

We think he's a great guy.

He's a good post.

Employee of the Month every month.

Fantastic.

That's it. That's it. Well, so great to finally meet you and congratulations on all the progress.

Have a great rest of your day. We'll talk to you soon.

Cheers.

Goodbye. Public.com, investing. For those who take it seriously, we got stocks, options, bonds, crypto, treasuries and more with great customer service. Our next guest is David Terse from Harman Industries. While he's coming in, I'm going to tell you about cognition. They're the makers of Devon, the AI software engineer. Crush your backlog with your personal AI engineering team. Good to see you. Welcome to the show.

Thank you.

First time on the show.

First time on the show.

Thank you so much for coming in person. Please kick us off with introduction on yourself in the company.

Yeah. David Terse, co founder and CEO of Carmen Industries. We actually launched today our heat processing unit. So an hpu. And this is a cooling system designed for gigawatt scale AI factories.

Okay. Cooling the entire data center, individual chip, both. What do you think?

All of it. So typically a lot of these are cooled with a similar loop because you've got high, high memory bandwidth, you've got GPUs, you've got air handling, all that type of stuff. All of this is about reducing the complexity of the infrastructure.

Okay.

So our overall solution is actually a 10 megawatt modularized solution. And that's very different than the 1 to 2 megawatt ones from the existing solutions out there. And we reduce that mechanical yard by about 80%. So your speed to deployment increases rapidly with this.

So heat is generated as I put electricity through the silicon chip. An old Facebook data center, they would just open the window and blow it out with fans, I assume, you know, air cooled systems. Now we're seeing a lot of water cooling. Walk me through, like try and explain what is the actual flow of where the heat goes, how it's moved around.

Yeah, absolutely. So pretty much every single electron that goes into a data center gets turned into heat.

Yeah.

Comes right into these chips they generate a huge amount of heat. So now when we're talking gigawatt scale, this is a gigawatt worth of heat.

Yeah. We talked about there's a bathhouse in New York called Bath House and they originally would heat the baths with via bitcoin mining. There's so much heat coming off of it.

Yeah. And racks used to be kind of in this like 5 to 20 kilowatt per rack. Now we've jumped up to 120 kilowatt per rack. The Vera Rubin coming will be 600 kilowatts per rack and they're trending towards a megawatt in Iraq. So when you think of how much power that is, megawatts, 1,000 homes consumption and putting that in a rack and then all gets turned into heat. So you have to move that the most efficient way possible. And today 20 to 30% of your energy is going towards cooling. That is a OPEX constraint, that is a thermal constraint. And it's just, you're not generating revenue with that. So we're trying to shift that and we reduce that at a minimum of by 25% energy consumption on the cooling side. Now you can have more headroom that you're using towards compute.

Yeah. And what's the, what's the material, what's the substrate for extracting the heat? Air, water, something else.

Yeah. So we interface with the existing systems that are there. So we, we integrate with the glycol loops that go to the chip level, to the CDU side of it.

Okay.

But inside our actual system we are using supercritical CO2. So we're running a high speed compressor that was developed in house by our team. We have a huge amount of history from, from SpaceX Rocket Lab. About a third of our team comes from there. And we've really designed this from the ground up. Our first one we came up and running in eight months with a four person team, ran it at 30,000 RPM. The next system will be coming together in Q2 here. And yeah, we're using supercritical CO2 due to the high density of the density.

So it can absorb the heat. It absorbs the energy, it heats up and then you move it somewhere else.

Yeah, we remove it, we vent it to ambient. But, but why we're calling this a heat processing unit is because today we view heat as a liability and as a constraint. But heat is energy and it's an asset. And so we're building the platform that allows you to remove that heat from your system as efficiently as possible. And it opens the door for future use cases of heat reuse.

Sure. Yeah, I want to talk to you about that because I've seen some of the hyperscalers have put out statements saying that there is waste heat from our data centers, but we're going to be able to pipe it into the local community, basically give everyone free heat in their homes. And that sounds really good as long as I guess it's clean air. But it seems like a good use. What does it take to actually go from a data center in the middle of nowhere to a data center that's closer to a bunch of homes to get the energy, get the waste heat there, and maybe just warm up someone's house during the winter.

Yeah, I think Europe will be more ahead than us just due to the proximity of those data centers to urban locations.

Just smaller, smaller countries. Right.

So where we see also a great opportunity is to do waste heat back to electricity.

Oh, interesting.

And actually supply that back to the data center. So in your colder regions at night during the winter, you'll have a big enough delta T where you can actually do that with our system.

Okay.

So that becomes now additional headroom that you can use towards compute.

Okay.

As well as. Then I kind of half seriously half joke that right now data centers are not in my backyard. That's the vibe that everyone has. But what if we built communities around these and you have free utilities if you live there?

That would be huge. I think people would really like that.

We were talking about this yesterday. Just like if somebody's, you know, you hear a data center coming into your backyard anywhere in your town, there's no, there's no. For a lot of people, there's not going to be direct upside just because even if it's not there, I can still use all the different AI tools. Right. I'm not like limited just because there's no data center even in my state. I like that talk about discovering the opportunity, the journey to getting here. Maybe your background. When we met, you were in this was probably a couple years at this point. You were in the idea maze looking at a lot of different opportunities and seems like you found a super exciting one.

Yeah. So kind of going through that is.

Yeah.

I was an entrepreneur in residence at Riot Ventures looking for what I was going to build yet Shout out will. And it really takes people that are true industry experts, world class engineers to develop something new. And I ended up partnering up with my co founder, Dr. C.J. kalra, who had spent pretty much his entire career in the thermal energy space. Most recently head of technology at Antora. Prior to that worked at a couple different startups, but started his career at GE Research working on supercritical CO2. So that's exactly what we're doing here with our system. So, so when we started working together it was looking at this broader thermal infrastructure layer. 50% of all end energy use is used for thermal management. We're either heating something up or cooling something down and we're doing it with outdated technology. So we're leveraging the tens if not hundreds of billions of dollars that has been spent across turbo machinery in the rocket industry. High speed power electronics, silicon carbide, permanent magnet motors in the EV revolution, packaging this all together and bringing it to the most pressing need, which is the thermal management of data centers. Right now this is a huge.

Have you been bouncing all around the map checking like on the ground at various data center projects?

A little bit now? These things sometimes are more heavily guarded than Fort Knox. These guys have machine guns outside of these things.

Yeah, no, I know. But once you're, once you're working with somebody and actually doing a deal, I'm sure that the doors get open.

Yeah, absolutely. And we're going to be launching these actual deployments starting later this year. We're standing up our first factory gigaworks one which will be 100 units per year of 10 megawatts as a gigawatt size capacity right out the gate. Because the demand is so extreme compared to where the supply chain is right now.

Yeah. So if we see one of these data centers, maybe it's 150 megawatts, maybe it's a gigawatt. How many heat processing units would they buy from you? Is it a modular system where they might be using some legacy system as well and then bring you on incrementally or do they need to go all in and work with you exclusively?

No, we're very much plug and play with the existing solutions that are out there. So you could, if you have an existing data center with an existing cooling system, we could, could phase in our system over time. But yes, a lot of these are clean sheet designs that are going up right now. And right now this is one of the biggest bottlenecks for them for deployments.

Okay. So yeah, there's obviously just a bottleneck of like if you just can't get any sort of cooling, you know, from the market like you, you'll just be delayed. Talk about the actual economics of someone who goes with you for a new build. Is this Higher capex, lower OPEX over time. Or how should a CFO who's working on a new data center think about working with you and the trade offs involved?

Yeah, absolutely. So our actual capex of the system is the same as existing solutions out there. And the snowballing effect of reduction of size, reduction of infrastructure, reduction of piping make this a cheaper solution right out the gate. Faster deployments because time is money on these deployments as well as then when you actually look at the efficiency of our system. So in your worst conditions, this is you need to cool your chips at 20C in Texas during the middle of the summer. We have a 25% energy reduction in the cooling system over the state of the art. When you start looking Virginia, some of these, you know, more moderate climates, we're looking 60 to 100% improvement in performance of this system, which that unlocks huge amounts of headroom. Yeah, so it's not as much about the OPEX savings because we do do that, but what we really unlock is the ability to allocate more of your electricity towards compute which that's the money maker in this thing.

Yeah, that makes sense. Talk to me about manufacturing a heat processing unit. Are there co packers for this thing? Can you go to a turnkey manufacturer? Once you have a design or a CAD file, you call in Hadrian or someone. Or are you going to vertically integrate and kind of make these in a warehouse somewhere?

Yeah, we're going to vertically integrate a huge amount of this. Now there's certain key components that we have a supply chain for sure around the motor, around a couple other pieces.

In there, tubing and piping and stuff like.

So we're taking very much the SpaceX approach. People think SpaceX vertically integrated from day one. It didn't. They bought a lot of things off the shelf and then they looked at at where they could either make an improvement or bring the cost down. And we're doing that exact same thing. A big piece of this also is diversity in our supply chain. We need to make sure that we're not single string in any one of these single components because the big guys, the hyperscalers are saying you need to produce a lot of these and you cannot slow down my deployment. The reason that they're coming to us is because we're speeding up their deployment. So we're making sure that we have de risked every piece of our supply chain. Our chief operating officer, she came from SpaceX, she was leading all of production at Millennium Space Systems before she joined us. So we have A rock star team. And that's always what Carmen has been about.

Yeah. What do you think? I mean, it sounded like you wanted to have these in production pretty quickly. What do you think about the lead time? The battle of the press releases, the idea of going out and getting a couple big headline deals done and then using that to either raise money against those or even just have them pay you up front and then that kind of finances the growth. How are you thinking about, like, the investor relations aspect of the business?

Yeah, I mean, I come from an investor background. We did actually announce the Series A this morning as well.

We got a gong for you. Hit the gong for us. The details.

Give us the details. Who did it?

It was Riot Ventures. They were our lead. We closed it back in September.

Fantastic.

Kind of under the radar, of course. Yeah, it was Riot Ventures, Space vc, Wonder Ventures, and then Sunflower Capital. And we also brought on Pat Gelsinger.

Oh, yeah.

As an investor as well.

Yeah. CEO of Intel. Hit that gong. Congratulations.

Yeah.

Gong rings out.

I like that. I like the technique.

Good technique. Talk to me about the name. It feels like you'd be a space company.

You're not.

Is it. Is the.

Not yet.

Is there something more to the Karman line or is that the reference?

Yeah, it's Theodore von Karman, famous physicist. He did kind of coin the term of the difference between our atmosphere and space, but just his work in supersonic and hypersonic flow was a lot of the inspiration for this.

That's amazing. That's amazing. What's next? You're hiring. How big is the team? Where do you want to be in a year? How are you thinking about growing the business?

Yeah, we're 23 people right now. We have the core team in place to build these first units, get them up and running. The scaling piece of this will be all technicians. To be able to actually manufacture a huge amount of these systems technicians.

Where are you pulling from? Where are the. Where do people do this work typically? What skills? Is it a trade school? Do they have college degrees? Are they mechanical engineers? Who are these people?

Yeah, so the kind of first wave of them is going to be people with a, I would say, very diverse skill set within manufacturing. And that's because when you're building these first units, you don't actually know exactly where the sticking points are going to be. And you're doing this kind of design for manufacturing while you're also standing up the manufacturing piece of it.

So you have to be able to troubleshoot as you're building something out Exactly.

So you want. Yeah. You want very high skilled generalist technicians. And then we will move into more specialized as we go, increase the velocity of that, the actual production side of it.

Are you in the Gundo?

We are actually in Long Beach.

Long Beach, Yes. Okay.

Bigger facilities down.

Bigger facilities, yeah.

If you need a lot of space, it feels like a lot of companies have graduate. They go to.

How big are the actual systems?

Yep. They're the size of a shipping container.

Okay.

So we got to build 100 of those to do a gigawatt in a year.

Okay.

So it's very doable.

Yeah. And would you actually stack them up, 100 of them at a big data center site?

Yeah, absolutely. I mean, you go look at these data centers today and you do an aerial view of one of them and you have more square footage getting taken up by the exterior mechanical yard, all for cooling, than you do the data center.

Yeah.

So with a one or two megawatt one, you're building a gigawatt data center. You've got five hundred to a thousand.

Wait, is that the current status? Okay. Because I would assume that there'd be some sort of mechanical economy of scale where if instead of, you know, you have a thousand tubes and a thousand, you know, like compressors and a thousand. And if you can just reduce that by just doing one really big one. But there must be something.

But then there's on site construction with that. It's a custom build all the.

Oh, okay.

So these were really designed originally for hospitals, universities doing cold water systems. You'd put two or three of them on the roof.

Okay.

They work great. These are concentrate controls, carrier train type.

Systems at the mug at the 1 megawatt level.

Correct? Yeah.

So you're scaling that up, but still staying in a deployable form factor that can fit on the back of an athlete.

Wheeler.

Yep. And that's what CO2 is great for.

Got it.

High density. We run a high speed turbo compressor inside there that was designed in house.

Yeah.

That's how we get the 10 megawatts in there. Got it. And then the actual fan system is separated from ours. And that's actually, we've decoupled that from our design.

And you're. You're building a thousand of these this year.

100.

100.

Okay.

Okay. And what, what's the rough. What's the rough price range for one of these things? Yeah, you can give me, you can give me a wide. I know it's. Call me, but like a wide, wide range.

Yeah. You're kind of in that, we'll call it 750k. A megawatt type range.

Very nice, Very nice.

So it scales up very fast.

Yeah. What are the like environmental considerations of compressed CO2? People worry about emissions, but what's the net impact of using CO2?

So CO2 is actually the gold standard for refrigerants.

Okay.

Because Today they're using R1234A something from Honeywell.

Yeah.

They while they have tried to push down the global warming potential of those, they now form forever chemicals over time.

Okay.

So CO2 is a global warming potential of 1. Because it is your baseline, it does not form forever chemicals over time as well.

Let's go.

Yeah, exactly. So it's.

We're going to get. We want to get CO2 cannons in here.

Yeah, There we go.

Yeah. No, it's actually a great working for, for many reasons on that. It's non flammable and then the energy density of it is just dramatically higher than the existing solutions out there as well as it's abundant, it's cheap.

Yeah. Interesting. And so walk me through like a data center construction in I don't know, 2028 or something. Is it possible that they're off water entirely? There's no water cooling or would that still be a piece of the puzzle?

Yep. So our system, while you use water as in like the glycol loops.

Yeah.

Ours is a closed loop system. Okay. So it does not consume water at all.

Interesting.

Which is a big plus for these days. We're getting a huge amount of pushback.

Yeah.

Of course. Gigawatt data center can pull over a half a million gallons an hour to keep it cool. It's an astonishing amount of water.

Interesting.

So we've seen a huge amount of reception from the hyperscalers around closed loop zero. Water doesn't form forever chemicals, natural refrigerant. All of these things are speaking exactly how they want to build data centers going forward as well as that compact package is speed to market. At the end of the day it's how fast can I build this?

How are you prioritizing your pipeline? It's kind of a. It's like there's not. I imagine there's some potential like mom.

And pop, there's hyperscalers and there's Neo clouds.

Great for the hardcore.

Go to semi analysis. Look at clustermax. Call every single person.

Yeah, yeah. So we've definitely been talking to a lot of the big name people out there as well as the Neo clouds, the ones that are kind of step below that I would say for probably every one of these big gigawatt campuses that you hear one of the hyperscalers announcing, there is three to five that are kind of shadow data centers that may look like an Amazon warehouse. And they're co locators, they're selling that compute to others. Each one has, I would say, its own unique reasons why they get excited about what we're working on. But yeah, that's actually a very ongoing consideration right now is how do we allocate our 2027 production numbers? Right now?

Yeah. Where are you trying to. How many units do you want to produce in 2027?

So this fall, we will stand up that factory with an annual run rate of a gigawatt. So in 2027, we'll do about a gigawatt worth of actual production.

Cool.

That's amazing. Well, congratulations. Thank you so much for coming on down to the TVP and Ultra Dome.

Absolutely Great to have you.

It's amazing. This feels like the perfect opportunity for you and a very necessary problem to be working on.

Absolutely.

Awesome.

Thanks for having me.

Thank you so much. Let me tell you about Railway Railway simplifies software deployment, web apps, servers and databases run in one place with scaling and monitoring and security built in. And I'm also going to tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Just do it. Stop making excuses.

No more excuses.

No more excuses. And we are ready for our second guest in the TVPN Ultra Dome today. We have Blake from Brink.

Whoa.

In the studio with some physical hardware, we have.

Make room. Make room. You want to put it right here in the front.

Maybe you can put it on top of here. Maybe. I don't know. Wherever you want.

Come on.

In the front. That works. We have a drone. Two drones in the studio. Thank you so much for bringing this. Thank you so much for coming on down to the TB pin Ultradome. Second time on the show, but reintroduce yourself. How long have you been working on the.

By the way? This thing is alive.

It's alive.

We got a live one.

It's a heartbeat.

Yeah, no doubt.

So, yeah, briefly reintroduce the company and just explain kind of the scale and size of the business these days.

Yeah, no, my. My pleasure. So story. Story of me. The company started my engineering career off over at McLaren Automotive. Was 14 when I was working there.

So amazing.

I think I was there.

You're. Why you're. Why you're. You're. You kind of can take a Little.

Credit for the W1.

The. Well, that and the Reese. You know, the Lando's.

Oh, yeah, I'm sure.

You know, fortunately neither of those things. I can take a tiny amount of.

Credit for the seven.

Okay. Oh, no way.

How's that?

That's amazing.

I worked on some parts that are on the underside of that.

Really?

Amazing.

That's amazing.

So that was an insane first engineering job. Obviously been all downhill in some ways. Yes, dude. I mean, it's hard to match walking by mp4 4 every day on my way to lunch.

Wait, what facility were you actually working out of?

Woking.

So out of the McLaren Technology Center.

No way. Which.

Yeah, like this.

How did you actually get the classic, classic summer job? Classic summer job? I was at 14, I was reffing like 10 soccer games every Saturday. Working hard, but not as smart.

Yeah, yeah. Wait, so how did you get the job?

Well, I built a fusion reactor.

Before?

Yeah, before the internship. So that really helped. Basically I would stay up until like 3am Los Angeles Vegas time every night. And I just kept calling their CEO. I never got through to. But after maybe the third or fourth attempt, I did get through to his personal assistant. And yeah, she took some pity on me and ended up facilitating the internship.

That's amazing. Okay, well, catch us up on Brink. Scale of the business where you are, product suite, all that?

Yeah, 100%. So Brink manufactures drones for public safety, end users. We started off with this guy, which is an indoor drone for SWAT teams. So the mission of this is to take off from an armored vehicle to fly up to a target structure, look around the backyard. And then this was the first drone in the world to have a glass breaching capability that would literally shatter a window. Make entry indoors.

Yes.

This right here is a small lidar. Generates about a half a million points per second. So as this aircraft is clearing rooms, it's actually drawing a floor plan of the structure and then streaming it back live to first responders.

Remind me how it actually breaks the glass.

Yeah. So basically it's a wheel that hangs off the front of the bird that has two tungsten carbide end effectors that are embedded into it. And the wheel spins up to roughly 30,000 RPM.

Oh, sort of like a saw almost.

Yeah. But it causes glass to just explode. In particular, worst nightmare we have to share.

When we were coordinating for this appearance, they said if you guys bring some glass to the ultra dome, we can smash it on set. And we were like, I don't know.

I actually broke some glass yesterday. I was playing a little basketball. Yeah, yeah, we break glass here.

Yeah.

Next time we'll have to have you bust in through the, through the door, take down the steel wall.

I would love to make that happen.

Yeah.

And yeah, this thing can also enable two way communications. Right. So when it finds a suspect, it can enable, you know, a crisis communicator to actually de escalate whatever's going on.

A hostage negotiation.

Yep. So Vegas Metro was our first customer today. About. About 20% of the SWAT teams in the country are actively utilizing this drone.

Wow.

What are the rest of them waiting on?

Well, you know, a lot of SWAT teams are super small and they're like a shared effort between a bunch of 10 person police departments.

Got it.

This is, you know, a rather premium drone. So they're not like in the direct addressable market for this particular bird. Yeah, if you're looking at the addressable market, we probably sold to, you know, 40 or 50% of those.

Sure, sure, sure. And they might, if it's a small team, you might need an extra operator. Someone who pull away from the door kicking that the rest of the team's doing. Not everyone has the staff.

Yeah, that's exactly got it. But yeah, I mean, from the beginning of Brink, I also had this vision of drones arriving at the scene of 911 calls before first responders could.

Right.

But I was familiar with how technically complex it would be to realize something like this from my time at dji, who was my last employer. And yeah, I thought building a city wide network of recharging pods that could integrate with computer aided dispatch and you know, enable drones to fly very reliably over people, et cetera. It was just like a little out of scope for a teenager without any money. So started with this. But you know, a lot has changed for Brink over the course of the last five years. Today we're just about the second largest quadcopter manufacturer in America and we are the world's largest public safety drone company. And that's enabled us to. Yeah. Realize that product.

Yeah.

This is new. When did this come out?

About a year ago, little over a year ago.

Oh, wow.

The drone was active. Yes. Yeah, for sure.

It's okay.

I'm touching.

Oh yeah, no, it's great. I'm glad you are. So, yeah, this is really the world's first purpose made. Nine one response drone. So it takes off from a citywide network of drone recharging pods mounted on top of police and fire station roofs. And it flies to emergencies before, you know, first Responders can show up, and then when it arrives, it streams back live video feed of whatever is going on. So first responders know exactly what they're about to walk into. It has an onboard thermal imager, so it can see heat, it can see through smoke, see hotspots and structure fires. You know, all of that good stuff. Communicate that potentially.

And I imagine police officers would want one of these on site, even if they are getting there first in some situation. Right. Maybe they're witnessing something and just having that third person view from the sky so that they're left focused on trying. Obviously they have body cams and things like that, but having that extra angle I think is important 100%.

Yeah. How does deployment work? I mean, imagine a 911 call comes in, an address is getting collected, and then at some point does someone type the address, hit go. And then the drone can just go from an address to and route itself, or is there a pilot that's firing it up? Like how autonomous is the actual deployment to get from the base station to an address?

Yep. So we grab addresses from computer aided dispatch or other pieces of software.

Computer Aided dispatch, what does that mean?

Basically, that's the software that dispatchers use to deploy ground units of various kinds.

Got it. Okay.

When someone calls 911, you know the GPS coordinate of that phone hits cat.

Okay. Interesting.

Yes.

So we'll grab that GPS coordinate, then we find the nearest drone to the emergency with sufficient battery state of charge to actually get there. The doors on our recharging pods automatically open. We launch an aircraft, we plan a path from its launch site to the emergency, avoiding manned air traffic, no fly zones, et cetera. And then when the drone arrives, it can engage in some number of pre programmed, like on arrival behavior, like orbiting the GPS coordinate, pointing its camera in.

A certain direction, saying something totally sensing itself.

Or it can then be taken over by a dispatcher or tele operator. The drone could be repositioned, camera can be moved around. Maybe that teleoperator does use the drone to communicate to someone. Or they might decide to deliver an emergency medical payload like Narcan, a defibrillator, personal flotation device, et cetera.

How do these takeoff points, charging locations actually work? When a police law enforcement agency decides or a city decides they want to set this up, are they going around and finding local businesses and getting them to opt in? Are these just like sitting on the roof of like a Starbucks or something like that? How does that all work?

You know, that does happen sometimes, but generally fire departments are Just already in the right places, in jurisdictions. Because.

Maybe it's helpful to talk about the range of this first.

Yeah. Each one of these birds can cover roughly 20 square miles of jurisdiction.

Wow.

Which is quite a bit.

So, like most cities, you actually maybe need a handful. You have great coverage, correct? Yeah. Yeah.

Talk about the supply chain to actually build a drone company. Obviously you're a great engineer, but there's so many pieces on here that America hasn't really been dominant in drone manufacturing. We talked about GoPro and sort of the rough go that they had trying to get into the market, being underpriced in many ways. Like, what did it take you to build the first one? How did you build the supply chain? How much is built in house today? How are you thinking about building these guys?

Well, fun fact, Brink has been sanctioned by China twice. And I'm also personally sanctioned by Twink.

No way. You're one of the guys on the list.

I know I am.

It really is. I mean, there's a lot of lists you can get on as an entrepreneur, but none with more kind of prestige than being sanctioned. I think I was.

I was proud of that one.

Get ready to vacation in America.

100%. So we had to move away from a Chinese supply chain.

And I'm sure you were buying some stuff early on. Okay, there's a battery. Buy some stuff off the shelf, hack it together. But eventually, the geopolitical reality and your 3D printing.

Yeah, so Lemur does include some carbon Fiber Reinforced Nylon 3D printed parts. Responder, though, is fully injection molded.

Okay, got it. And so walk me through the supply chain. What are you making? What is America good at making? Do we at least have a good drone motor company? I've been hearing that drone motors in particular are difficult to make in America. Or at least a lot of the companies sort of offshore or left or went out of business.

Yes. So Listen, I mean, DJI controls 90% of the global drone market. It's really a monopoly. I think this isn't so well understood. Like, the drone industry is a monopoly owned by dji. Autel is their nearest competitor. They're another Chinese drone manufacturer. They have about 5% of the market. All foreign drones were just federally banned. I don't know if you guys have been tracking this, but right before Christmas.

Wait, but. But only in a. For government buyers or all consumers.

All new foreign drones have just been banned from entering the country.

Okay, so if I go to Best Buy right now, there might be a DJI Mavic 3 on the shelf. But there's not going to be any new ones once they go out of stock.

That's probably the next stock. Exactly. Basically like the next Mavic won't be allowed into the country. The next Matrice series drones will be allowed into the country. They can hypothetically continue importing their current in market drones, but in reality, in reality, even those imports are being slowed down very much at the borders.

Okay.

So there's, there's this huge need for, you know, a DJI of the west or like a leading drone manufacturer for the free world.

Yes.

And that's the organization that we want to be. Yeah, I mentioned, I mentioned all of this though, because.

Wait, so what does that mean then? Does that mean you start selling these to the public consumer?

I want, I want, I mean, I.

Want, I think, I think over time, yes, we probably do start engaging in those types of businesses.

Yeah.

You know, consumers a little far from our current space.

Yeah. But if it helps you get to scale, bring down costs. Yeah. There's a potentially good feedback loop.

I think though, like private security, industrial Automation, Homeland Security, etc. Would be like more, more natural, immediate new verticals for us.

But.

Yeah, yeah, I mean, over time we want to be.

That was underreported. I didn't, I didn't catch that. They were, I remember, I remember some news around foreign drones with, within purchasing from the government. I didn't, didn't clock.

Yeah, they were. Walk us through some of the history though. There were some local bans, some state level bans, some individual agencies that might say, hey, we're not going to approve these. You heard about, you know, you can't use TikTok in Langley. That seemed obvious, but now it's grown a lot. What was the actual history of the rollout? Who's been driving this in Washington?

Yeah, so the progression here was, number one, the federal government banned itself from purchasing Chinese drones. Then some states started passing similar legislation. Florida, for example, example, banned their public safety agencies from acquiring new Chinese systems. About a dozen states have followed suit. And the new thing is what I just articulated, you know, the government saying, you know what? No more new foreign drones are allowed into this country, period.

Okay.

Regardless of application.

Yeah, walk me through the geopolitical risk of, you know, foreign drones. At one point I had a DJI Air 2 or something, this little mini 49. What was it? 490 gram or something like that. And it was sort of like collecting dust in a, in a, in a drawer somewhere. And I was thinking like, okay, if somebody hacked this and really tried to fly it out of its little case and out of the drawer and then out of my closet and then out in my house. Like, what is it going to do? Is it that dangerous?

Walk me through the. I've always thought that this is like you rewind and you know someone in China, so some entrepreneur in China is like, I'm going to make a hit consumer product in America that will be able to give you a real time map of everything going on. And there could be so many drones that people won't even notice when a drone is just hanging around some strategic asset or anything.

So is that the risk? Surveillance?

I would say there are three primary risk vectors. Risk factor number one is that DJI drones just stream back video, thermal imaging information, GPS coordinates, altitudes, et cetera to China, which.

Going straight to the DJI Cloud.

Yeah, 100%. Yeah, 100%. So that's, you know, that's the first big like tranche of risk. The next tranche of risk is look at what's happening in Ukraine. Like drones, specifically DJI drones and other, you know, aircraft that are similar are being deployed to incredible effect in the context of warfare. And they're being consumed at insane rates. If America wanted to engage in adopting those types of drone tactics, we would need to produce many hundreds of thousands of drones a month. And we just do not have that industrial capacity in this country. If you ban foreign drones though, guess what happens? That industrial capacity gets built up because the domestic market starts to be served by scaled American drone companies.

You bootstrap wartime demand with consumer demand. Today there's enthusiasts that want to be able to get footage of their ski trip or hiking or any number of use cases.

100%.

And then the third tranche of risk is what you articulated. I would say though, when you start contemplating systems like, like that.

Sure.

You know, that risk starts to feel much more serious.

Yeah.

Because it might be able to take off at any time.

It's not from a recharging pod.

From a recharging pod.

And so yeah, I was surprised. What was the operation in Ukraine where they parked next to the air base, took off like that. Like I still feel like that risk. People are not taking it seriously enough about that kind of event happening like on American soil.

What was it? Spiderweb.

Operation Spiderweb. Is that it sounds right.

I forget what it was called, but yeah, that was certainly dramatic. So talk to me about you're having success in the police market. Are you thinking of going to the federal government, the Department of War soon? Is there a crossover there or is that more of a distinction market that you're putting off for the time being?

Yeah, I mean, the way I think about things is there are 20,000 police departments in America, 30,000 fire departments, 80,000 police and fire stations. In the future, I believe, you know, the top half of those buildings are going to have a drone recharging pod on the roof. That's 40,000 deployment sites. We charge on the order of like 60 to $160,000 per year per launch site. So even being very pessimistic, the domestic market's around $3 billion recurring. International, excluding Russia, China, Iran, and all the countries that will never be able to sell to is a similar size. Maybe like a $6 billion pie. Even if we're only successful at capturing a third of that. Like, we're, we're a large public company. Axon's prior, closest comp. Right now they're trading at roughly a 20x revenue multiple. So, like, that'd be an amazing outcome for us. But long term, we do want to be the DJI of the West. Right. And we want to operate in every customer vertical or we can add value and yeah, that includes some of the areas that you listed.

Yeah. On the engineering side, if you could wave a magic wand, what technology would you want to advance most directly? Is it the battery, the motor, any other, the actual sensor stack? Where are we most bottlenecked from just a technology development perspective?

You know, I would say, surprisingly, connectivity can be a challenge. Yeah, we're about to release something very innovative in that domain with our next drone. But yeah, that can be a major issue. Like cellular isn't always perfect, especially at altitude.

I have this very, very first world problem that on the drive home from the studio, there's like a 10 minute stretch where there's just. I cannot get a. I can't have a phone call. So I start a call when I leaving, and then I'm like, I'm gonna have to call you back in like 10 minutes. And then I call people back and I'm like, how in the middle of Los Angeles on a major like street. Yeah. Do. Can. Can I not take a simple phone call?

Keep in mind, our drones fly everywhere in jurisdictions, right? So every city has little pockets like that.

Also the Verizon outage yesterday, you have to be you. I mean, I have, I didn't have any issues with it, but I mean, there was like 30% of Verizon users I think yesterday had no coverage. Some massive number, mostly east coast numbers. I Guess, but yeah.

What about Starlink satellite connectivity? Is that doable on a drone this size? I have the Starlink mini. I could imagine the extra mini or the nano fitting on here. Is that in the cards?

Yeah. You should watch our next launch.

Okay.

Okay. When's that rough day? Couple months.

Q1.

Okay.

Yes.

Yeah. Yeah. Interesting. What else goes into training police forces and SWAT teams on using these effectively? Have there been any creative uses, creative stories that you've seen where people have deployed this in maybe novel ways where you didn't even think, oh, but they could use it in that particular way? Are there any interesting stories that you guys have shared?

Oh, there are so many. I mean, especially when it comes to lemur, right? Yeah, yeah, there are really a lot. There was one that I watched personally where, like, a person was kind of pretending to be not responsive, so the operator of the drone actually kind of started nudging them and was able to get a response out of it.

Interesting.

So, yeah, I mean, all weird. Weird things along those lines happen just constantly. We can fly this as well if you guys are, please.

Yeah, that sounds great. Let's fly.

Drones are alive.

All right.

Yeah.

So takeoff site there.

Yeah, we see here.

There's a Diet Coke. That's gonna get knocked. John, I'd move your Diet Coke.

Okay. I think we're flying this one.

Okay.

This is amazing.

Yeah. But, yeah, I think the. The indoor drone is maybe the right.

Yeah, let's.

We got the camera view back here.

Oh, we got the camera view up here too. Okay.

Oh, wow.

It's fed in and everything. Nice. Yeah. Let's fly this around. This sounds great. This thing would definitely get me to comply if it was flying around. This is crazy. There we go.

So we'll show off a couple of functions.

Okay.

Yeah.

Show us some stuff.

You know, right now we're in a GPS denied environment.

Okay.

You know, we won't be able to see any satellites through a metal roof like this.

Sure, sure, sure.

The drone is utilizing its onboard cameras, running visual inertial odometry algorithms.

Yo, yo.

Yeah.

In order to localize. And it's also using some onboard lidars to do the same. Yeah. The aircraft is streaming back 4K video and thermal imaging information.

Okay.

So, yeah, Dimitri, you want to maximize thermal.

Really?

Oh, the thermal. Oh, wow. These are all of our thermal signatures. This is very cool.

There we go.

Wow.

And those sensors are mounted on a payload that can rotate a full 180 degrees so it can look straight up.

Okay.

Also straight down, which is pretty operationally Useful. We also have some lights that are built into our camera pod. Dimitri, you want to hit those?

Whoa. So that's a good light for economy flashback.

Yeah.

And you can barely see the.

Oh, you can strobe it too.

And they both have strobe function.

Okay. Yeah, yeah, yeah, yeah.

So that's.

That's all.

Wow.

That's fantastic.

You can see that kind of articulation.

Oh, yeah.

It goes up and down.

There you go.

Great.

Dimitri, you want to land it and then I can give it a quick call.

Yeah, that's amazing.

Give.

Give it a quick call. Yeah, it really. It really is like a creature.

Yeah, yeah.

The way.

The way that has. It has some. Some gusto to it.

It's.

So the airframe has a cell phone number.

Oh, wait, really? Oh, so you can just call from an actual phone.

Okay.

Don't dox that phone number. You're ringing off the hook.

One, one, one, two.

Wow. Yeah. Okay.

This is how, you know, Austin, negotiations are happening around the country.

Just call the drone. That's so simple. I mean. Yeah, it was probably just.

It was the right solution.

Yeah, simple. Just work off the default rails. That's amazing.

How on earth did hostage negotiations. If you want to turn it off?

And then there's one more thing I'll show off quickly, which is Turtle mode. It's a very durable drone.

Okay. Yeah.

If it ever does crash and lands on its back, it can self.

Right.

So you'll see that it can self. Right. Oh, there you go.

Wow.

Turtle mode.

Yeah, that makes sense. That's very cool.

Okay.

That is very cool.

Yeah. Thank you.

How.

Question, Jordan.

How did. What were the actual logistics of a hostage negotiation historically?

Wasn't it a lot of megaphones screaming?

Yeah, you're yelling, but then they're trying to yell back, but they're in a house and they're trying to not be visible? I can imagine it's hard to hear. Makes the negotiation a lot harder.

I mean, that absolutely happened. A lot of instances of SWAT teams just having to make entry into structures where they didn't know what was waiting for them on the other side of the door also occurred. And that's not good for anyone. I mean, that's incredibly dangerous for police officers. It's dangerous for the suspects. It's dangerous for civilians. It might just be, like, generally in the area. So, I mean, our thought with Lemur is, why do that at all when you can send in a little drone to clear all the rooms, find suspects, enable negotiation. Hopefully, that just resolves the situation. But even if it doesn't, at least now first responders know what's waiting for them and they can put together a plan that takes all that data into account.

Yeah. What are the economics of the first responder units that you work with? Are they, are they wary about capex? Do they want more of a subscription product with services and they get regular updates? Do they just want to pay one time and then you get out of their hair? What's the correct business model for this?

Yeah, I mean in general nowadays we're selling full blown drone program. So that might include some number of indoor drones, some number of standalone outdoor drones, and also jurisdiction wide nine one response drone coverage. In year one, we deliver all of our current state of the art hardware. So an agency would get like responder 1, then at year 2.5 of a contract, we automatically upgrade them to responder 2. Same thing happens at year 5. If a drone ever gets damaged or destroyed over the course of those five years, we'll repair or replace it. We include data storage, we include unlimited software licenses for pilots and viewers. We include regulatory assistance with the faa. Everything they need to run their full drone program.

Yeah. How have you been processing the advancements in AI over the last couple years? It's been useful. We have better video models, we can understand images at the same time. The, the baseline is like a megaphone. So this is a huge upgrade. No doubt. And you might not be able to do on device inference. I don't know if there's an H200 in there or something like that.

Right.

So yeah, where is AI being helpful? Where's it being useful? Where has it been? Okay, wait and see.

So our drones won't even take off without utilizing some of those techniques. You know, when you have a strong GPS signal, you can mostly rely on that for localization, especially if your aircraft has a GPS RTK system. But GPS isn't always perfect. Like when we're flying around in this structure or if we're flying near a big building or under a tree. You do need redundant localization systems. What we've implemented to solve that problem is a vision system that relies on those sorts of techniques in order to pick out tracking points and to calculate relative velocity. So it's really pretty fundamentally baked into our product at that level and at higher levels that touch end user apps.

Yeah, one of the most obvious use cases, just break a window, go into that building, find a person, and then I want to talk to them on the phone and that's it. And maybe I can't get the video feed out because the connectivity is difficult. But if we can at least maintain a phone line, we can have a conversation and progress can be made and hopefully people can be saved.

Yep, for sure.

Fantastic. Any predictions on how drones will be applied in wildfires and just firefighting broadly? Is that something you guys touch yet? Would you ever touch it? How are you thinking about that?

Yeah, I'm absolutely thinking about that space pretty seriously. Drones can add value in a number of different ways. First of all, they can help out with wildfire detection. Early on, it's possible to put a drone recharge charging pod in an area that has a high wildfire risk and then just have it fly around.

There'll be, like, solar panels powered. Yeah, you think? Totally.

100 and. And likely, you know, Starlink insured back off for connectivity.

Okay.

Yeah. Some of the footage that came out of the LA wildfires was so bad, like, you would. I would see some of these cameras, and it looked like a. It was a camera that was a potato.

No, we only got drone footage that was days later.

And you'd be like, yeah. And it would be like, okay, there's a tiny speck of smoke here, and it's like, how is that, like, the best that we have when it causes tens of billions of dollars of damage?

For sure.

Drones can also be used for logistics, you know, moving around chainsaws and that kind of stuff in areas that are just difficult to traverse with conventional vehicles. And I think in the future, drone swarms will be used to actively fight forest fires.

Just drop a ton of material all over the place.

Yeah.

Or even, I think, when. Even. Even for, like, ember detection and, like, putting out, like. Because obviously, when some of these wildfires start, they're. They're the actual fires happening here. Embers are floating kind of randomly with the wind. And then there's a moment where, like, if you just had, like, effectively a fire extinguisher that could, for a second, just, like, hit a certain spot, you could stop, like, a bunch of structural walls.

But back to this guy. What's the decision? To have the white posts come up on the back rotors and then go down on the front rotors.

Yes. So those cans on the back of the drone are actually our GPS RTK antennas.

Oh, interesting. The motor's actually just here.

Yes.

Okay. These are antennas 100%.

And they need a field of view of the sky in order to receive. Yeah, yeah, 100%. So, yeah, those two things are antennas. We've embedded antennas on the front Two motor pods as well for a mesh networking radio. And then you see those two little wings coming off like the. Yeah, yeah, those, those are actually our LT antennas.

Very cool.

And it's, it's generally good practice to separate antennas on drones as much as humanly possible to avoid crosstalk. And that's why we engineered, engineered this configuration.

Well, congratulations on all the success and amazing update.

This has been the most wild in person interviews ever.

Let's glass break next time.

Yes, yes, we should. We will definitely get some. Bring the sheets, smash it all.

That is by far the coolest demo.

Okay. Okay. Fantastic. Well, thank you so much for coming on the show.

Thank you.

Congratulations on the progress and we will talk to you soon. Well, you thank. While he's walking off, let me tell you about Phantom cash. Fund your wallet without exchanges or middlemen and spend with the Phantom card. And I'm also going to tell you about Gemini 3 Pro. It's Google's most intelligent model yet. State of the art reasoning, next level vibe coding. And they got deep multimodal understanding over there. Jordi, is there anything else in the timeline that you think we should get to? The one thing that I have is that we have some breaking news that Doug o' Laughlin over at Semianalysis has severe LLM psychosis. He's been one shotted. He uses Claude everywhere and anywhere and everywhere, including a client Happy hour at 6:50 in New York City. And Doug says, absolutely. You've met me at a very Claude code time in my life. I listened to him on Chinatalk or on, sorry, on a podcast with Dylan Patel and Jordan Schneider from Chinatalk Transistor Radio. And he was obsessed with using Claude code to organize his notes, write markdown files, orchestrate his entire life. He says everything is a skill issue. AGI has been achieved and he's very excited to be using agentic coating agents everywhere and always.

Last thing we need to cover is the PBR99 pack.

What?

That has been released.

They launched this during dry January. What a wild move. Zig when they zag.

The new 99 pack wraps 99 PBRs in a gigantic all new original illustration featuring characters from the Godzilla universe. I mean, this collab with Godzilla put above your fireplace, right? I'm sure, I'm sure all the wives out there will love when the husband comes home with a 99 pack. And it's the song.

99 bottles of beer on the wall. 99 bottles of beer. You take one down, pass it around. 98 bottles of beer on the wall.

They did it.

And they, yeah, they did it. They did the meme. But yeah, congrats to the team at PBR over there. They're doing stuff. They're innovating. Well, that's our show today, folks. The bomb has been planted. Leave us on Apple Podcasts and Spotify. Subscribe to our newsletter.

We love you@tbpn.com we got a fun day tomorrow. We have our friends from Sequoia popping on. We got the founder of Athletic Brewing.

Oh, yeah, we have some Athletic Brewing over there. Are in the refrigerator.

And of course, we'll have the Mansion section.

Of course. Excited for that mansion section tomorrow. Thank you so much for having a wonderful, great rest of your day.

We'll see you very soon.

Goodbye. Nice work, brothers. I'll see you on the next one.