100 Billion Bezos, SMCI Fully Sends GPUs (To China), Reddit CEO Joins | R.F. Kenmore, Mitch Lee, Bucky Moore, Steve Huffman, Quaid Walker, Ankur Jain, Michael Kratsios

You're watching DBPN.

Today is Friday, March 20th, 2026. We are live from the DBPN UltraDome, the temple of technology, the fortress of finance, the capital of capital. Let me tell you about ramp.com. Time is money. Save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place. We have a massive show, two in-person guests today. Let's pull up the Linear lineup. Linear, of course, is the system for modern software development enterprise. 70% of enterprise workspaces on Linear are using agents. We have RF Kenmore from the Menswear Industry Insider.

No, he just is a menswear insider. He's an insider of the menswear industry.

Critic. Mitch Lee from ArcBoats is coming on. Bucky Morse coming in person from Lightspeed. Steve Huffman, the co-founder and CEO of Reddit, coming on. Then Quaid from Bezel, who you know and love. He's been on the show many times. Ankur Jain from Built's coming on to talk about credit cards, consumer cards, all the fintech stuff he's working on. And then Michael Kratzios from the White House. He's the science and technology advisor. He's coming on to talk about the new policy proposals that the White House has put forward around artificial intelligence. But first, I wanted to chat more about the breaking news from yesterday. Jeff Bezos is in talks to raise $100 billion for an AI manufacturing fund. There was a funny post that was like, he has $200 billion. Why does he need to? Raise more money.

No conviction. No conviction.

Oh, yeah.

I know it was funny. It was like it definitely— this, this broke containment and got into kind of the anti-capitalist people are not part of X. And they were like, wait, why are you asking people for money? You have all the money.

Yeah. Yeah.

But of course.

But of course he, he wants to let other people in on the action. Well, I want to let it— I do wonder what his like GP commit will be.

That's what I was saying yesterday. I was like, I would expect him to be like, yeah, I'm good for 20.

Yeah, 20 or 30 or something like that. I don't know. Maybe he'll go even longer. But in general, I think the idea of going around the globe, because he's an international person at this point, hoovering up money from all over different sovereign wealth funds and then deploying that to help rebuild the American manufacturing base is a good thing. I wrote about it today in the newsletter, tbpn.com, and I also tried to do some searching and think about like, what do you actually buy for $100 billion? What does the portfolio look like if it's a private equity style roll-up? If you're just taking out the market caps and you're sort of assuming the debt and continue to operate the business, but then bring more efficiency to bear, what are some of the top options? What are some of the options look like?

I don't know. The other thing is I saw a very viral post about someone that seemingly read the headline and thought, like, don't you have enough money already?

Why do you have to buy all these companies and fire?

Like, they saw automation and just assumed, like, job loss. And when I look at this, I'm like, okay, if he's successful, we will, you know, create— we will actually start adding a bunch of new manufacturing jobs, which we've been shedding over the last year.

Yeah, yeah, yeah, exactly. So headline number's funny. If he had done $99 billion or $101 billion, I don't think people would be making as many SoftBank comparisons, but of course, as soon as you hear $100 billion for a mega fund, you instantly think SoftBank. He's got vision. It's a vision fund for America. I like the idea of an American SoftBank, of an American vision fund, so I'm generally pro, but of course, people are gonna make comparisons to SoftBank, which has had some "Few Winners and a Few Black Eyes." Moss is a high vol guy. He's been the world's richest man briefly during the dot-com boom. It was reported that he was worth more than Bill Gates. It was pretty temporary, unclear exactly how much, and of course there was no liquidity at the time. But he's also held the crown for "the man who lost the most money in human history" after losing $76 billion in 2 years. You're supposed to do the "wah wah" for that. But he made it all back.

I was preemptively hitting it.

Okay. Yeah, yeah, he did. He did make it all back. He's made almost $100 billion twice. I think maybe there was a moment when he had the title of making $100 billion twice, one on Alibaba and then the second one on Arm. He's made money on Nvidia and lots of other things. But the idea of an American SoftBank, especially one focused on revitalizing American manufacturing, I like that idea. There are a million reasons why this is an important project. Job creation, economic independence, national security. Et cetera. And you might just get more cool stuff. Like, as manufacturing gets cheaper, like the tools that you use, the cars that you drive, the washing machine in your home, the tires on your cars, all these different things get better, not just cheaper, but actually unlock new capabilities and new things that people enjoy. And so, I'm in favor of more manufacturing, more economic development, basically. I don't think the project is done by any means. And so the question about like, you know, Jeff Bezos's role, like he's a great operator and I think he's the right person to take the reins and actually deliver here. So like Masa, Jeff Bezos has been through the ups and downs of the dot-com boom and bust. At one point, Jeff Bezos lost 85% of his net worth in like 2 years.

But he made it back.

He did make it back and more. He was worth like $8 or $9 billion and got down to his last $1 billion, basically. Because of the stock drop in Amazon. But he built it, but he not only kept Amazon alive, which I think everyone knows that Amazon went up during the dot-com bubble and then crashed and then built back up. But he also kept Blue Origin alive during that time, 'cause he founded Blue Origin in 2000.

Wow.

Before the crash.

I didn't know that.

Yeah, before the crash. So he was—

Was it before SpaceX?

Yeah, before SpaceX.

Whoa. Yeah, so Elon sold Blue Origin. 'Cause the reason I assume that Blue Origin was just memetic Elon.

No, no, no. He did it earlier.

Wow.

And so he kept Blue Origin alive. And can you imagine how stressful it is? You're like, "Okay, I'm worth $10 billion. Certainly I can have a little side project as a treat." And you're like, "Okay, I just lost 85% of my money." I deserve a side project that loses $20 million a month. Exactly.

Yeah.

I don't know how much they were losing, but that doesn't seem unreasonable. But he was able to keep it going. Of course, Blue Origin was a slow story all the way up until like last year when they landed New Glenn. And so, he kept both alive and he's never given up on the project of Blue Origin that always has felt behind SpaceX for the two decades that it's been operating, but is now sort of unlocking new capabilities and is—

They were going turtle mode.

Yeah, turtle mode for sure, but it's working. And so, Regardless of the fact that SpaceX is still ahead of Blue Origin, you have to give him credit for operating that company efficiently and actually delivering a product that we all saw go up and come back. And so the rockets did go up and land and come back, and he's delivered people to space and back. And so he's achieved the goal and not lost all the money and not put the company in jeopardy. Clearly, there's some operational—

That's a narrative violation.

It is a narrative violation. What makes Bezos uniquely equipped to take on this type of project is the nature of his business over the last few decades. So he's never really had the zero marginal cost luxuries afforded to other pure-play internet founders. You think about Mark Zuckerberg, it's always been—

Except for this ads business.

Yeah, he has that now, but that's very new. For a long time.

New, but very real contributor.

Yeah, very real contributor now, but even then, it is much thinner margin than a Google or a Meta. And it always has been because they've always operated in the real world and they've been competing with Walmart and Barnes Noble and Romans and all the other big orders.

Yeah, he always said, "Big gross margins scare me." Yeah, he's always had thin margins until AWS, really.

AWS was the first like, okay, This is like a monster cash cow. And even AWS doesn't have 80% gross margins. It is a cash cow, but it's a different shape. And of course, you go back to the Jeff Bezos lore, and he had the door desk, and the whole business was very scrappy for a long time. And Amazon's just always interfaced with the physical world. So it's been, even though it's a software company and a tech company, it's always interfaced with the physical world. And they've had to focus on operational efficiencies to scale. So in 2012, Amazon bought Kiva Systems, which turned into Amazon Robotics. And that's a big risk to buying a big company. And we think that if this $100 billion fund turns into sort of a private equity roll-up, you're going to ask the question, can Bezos buy a company for $1 billion, $5 billion, $10 billion, $50 billion, who knows? And operate it and scale it and actually continue to deliver value. And he certainly did that with Kiva Robotics, which turned into Amazon Robotics, which delivered over a million robots that are deployed across the operations network. So from fulfillment centers, warehouse automation, inventory flow, last-mile delivery, all of these require tight integration between hardware and software to run smoothly. Bezos clearly loves this stuff. And you can actually see it when you look at his face whenever he gives a tour of an Amazon warehouse or a Blue Origin facility, he's just beaming. And he's absolutely obsessed with this technology and the physical world. And so the question I had was like, what should he buy?

Yeah, the other thing is like seeing all the different volume across Amazon and being like, well, it'd be pretty convenient if that was made here or this was made here or at least some of the components were made here and it wasn't, you know, you didn't just have assembly here.

Yeah, I mean, yeah, there is an— I hadn't considered that narrative, but there is a different— flow where you're—

He's not sitting there being like, great, I love that so much of our volume is having to be shipped or flown across the ocean.

A ton of that, yeah. And then also he's seen attacks from Temu and Shein and Alibaba and AliExpress and a whole bunch of direct from manufacturing plant to customer. And how do you respond to that? Maybe getting into the deeper in the supply chain makes sense. Although I'm not entirely sure that there's so little reporting at this point. It's the only line is that he'll be buying manufacturing companies. So there's a lot in the supply chain. So I wanted to dig through some stuff that could potentially be in the crosshairs for Bezos. The first is Lear. They manufacture seats and electronic systems for the automotive industry. And a lot of these companies have huge revenues and small market caps. So Lear did $23 billion in 2025 revenue. The market cap is around $5 billion. And so the business is operationally complex. I mean, on a PE ratio, it trades at like, I think, 13 price to earnings. And so, like, you have these high revenues because you're in the manufacturing business, but you have low margin.

But— Yeah, and this has been some of the concern around VCs investing in manufacturing businesses.

Yep, other use cases.

And just a general concern that, you know, they might today be getting valued on a revenue multiple, But as they get to the public markets, people will expect them to be valued on earnings. And that is a very different story.

Yeah. Lear, interestingly, not the maker of the Learjet, which sold to the Gates Company, which is not affiliated with Bill Gates. I was like, is there a link here? But there's not. But Learjet and Lear are two different companies, although they both make physical things. But there's an opportunity here for the Bezos thesis that we are are sort of gesturing towards. The business is operationally complex and AI can be applied across plant scheduling, supplier forecasting, visual inspections, and more. This whole thesis of like the collars get less blue, there's upskilling that happens in the facility and you're able to produce more at scale. BorgWarner is another example. They're a scaled auto supplier, $14 billion in 2025 revenue. Market cap is $9.5 billion. They're more focused on propulsion and power-related systems. They literally make turbochargers for internal combustion engines, although they have shifted their business a lot to be focused on electronic components, both in terms of making battery packs, motors, but they're actually already getting into selling turbine generator systems for data centers. They're part of the AI boom already. And there's more opportunities to grow into different industrial power applications. Hexcel is another company. They're a leading producer of carbon fiber reinforcements and resin systems for aerospace. They're guiding to $2 billion in revenue, but the market cap is $5 billion. So a little pricier on a price-to-sales ratio. Here's a fun one, Goodyear. Not only do you get blimps, which has got to be important if you're trying to go blimp for blimp with Sergey Brin, you just jump to the front of the line. It's only a 1.8x multiple. This revenue multiple is criminal. Yeah. So the market cap for Goodyear is $1.8 billion, and revenue was $18 billion. So they're trading at a 1.1x multiple.

Somewhere out there, there's a VC-funded tire company.

It's like 100x revenue? And they're like, we're going to go through 1,000x multiple compression over the next decade? Yes, you will. But Goodyear's sort of the classic AI for manufacturing case.

They gotta focus, they gotta get investors thinking about their advertising blimp monopoly and scaling that business.

Yep. So quality control is really critical. There's opportunities for downtime optimization. So if you can have better systems that limit downtime, you can produce more tires. The tire market is extremely competitive. You see Chinese tires on cars more and more. These are thin, thin, thin margin companies, and that's why the multiple is so low.

Scariest moment of my life in a car, Chinese tires.

Yeah, Chinese tires, old Chinese tires. Old Chinese tires. But a lot of people go in, they say, I need new tires, just give me whatever's cheapest. If you're competing on price and you're a low-cost supplier, every dollar counts. On the bigger side, you have Rockwell Automation. I don't know if you've ever seen those videos for the Rockwell Turbo Encabulator. Have you ever seen these? We should pull this up. Let me find it. It's one of the best, like, you know, marketing videos. Turbo Encabulator. Let me find this. The Rockwell Retro Encabulator. I got it here. We'll watch this. It's a 2-minute video. And we'll see if you can— yeah, yeah. Any of these. Yeah. We got these. If we can pull this up and let's play. Here at Chrysler Motors Automotive Operations, research has been proceeding to develop a line of heavy-duty transmissions that establishes new standards for reliability, durability, and quality. With customer needs as our primary focus, work is proceeding on the crudely conceived idea of an instrument that would not only supply inverse reaction— I think this is Chrysler's like joke. Let's play especially the one that I shared from Rockwell. This is the good folks at Rockwell Automation who—

that was a cool video. Look at this. Research has been proceeding to develop a line of automation products that establishes new standards for quality, technological leadership, and operating excellence. With customer success as our primary focus, work has been proceeding on the prudently conceived idea of an instrument that would not only provide inverse reactive current for use in unilateral phase detractors, but would also be capable of automatically synchronizing Cardinal gram meters. Such an instrument, comprised of Dodge gears and bearings, Reliant electric motors, Allen-Bradley controls, and all monitored by Rockwell software, is Rockwell Automation's RetroEncabulator. Now basically, the only new principle involved is that instead of power being generated by the relative motion of conductors and fluxes. It's produced by the modial interaction of magnetoreluctance and capacitive directance.

Passive interaction.

The original machine had a base plate of pre-famulated ammolite surmounted by a malleable logarithmic casing in such a way that the two spurving bearings were in a direct line with a panametric fan. The lineup consisted simply—

it's just, it's just like such a funny in-joke for electrical engineers because It's just a whole bunch of slop, basically. None of those terms mean anything. But it really— I don't know.

Good startup launch video concept.

Yeah, yeah, yeah. I mean, it helps you detect, are you in group, out group, basically. Very, very fun. But they are— What? Well, were you able to clock it, or were you like, oh, that sounds good?

I mean, it sounded silly.

It sounds silly. So Rockwell Automation's on the bigger side. but they sit right at the intersection of factory automation, software, and control. So although that video's a joke, Rockwell does work on factory automation already. So the thesis with this one is that the business is already dedicated to industrial automation. And so it's less of a turnaround, but it's a control point for pushing AI to thousands of other factories here deeper in the supply chain. It's expensive at $40 billion, but potentially in the budget.

Do you think there's a world where he just buys Ford Motor Company? Obviously. I mean, yeah, wouldn't he? Would have to lever up.

Yeah, but what is Ford now?

I thought it's $45 billion, but I'm not— but he's not going to just drop like raise $100 billion and then, and then buy a company entirely with equity.

I feel like it will be deeper in the supply chain than a brand, but I don't know. It's possible. It's possible.

But he might say—

Isn't he already a big investor in Rivian?

Yeah, sure.

So looking deeper into that supply chain, and there's a few other car companies that he's been involved in. I just don't know that that's where the big opportunity is. It might be deeper in the supply chain, like bending the metal that goes into the bumper, this and that. And it's 12 steps deep, it's more boring, but it's even thinner margins. Less brand risk and brand value, but really focused on applying that actual commercial excellence. But I mean, truly, your guess is as good as mine. There's very little to go off of. Also, there's a link to Project Prometheus, which is the AI project. And so there's a chance that this whole project is more aligned with the software singularity, software-only singularity, and you wind up with buying a lot of things, but everything's in line with like more data centers. And this is more of like, let's put the $100 billion to work towards building compute capacity for Project Prometheus. That feels like not what we're hearing, but it is one possible outcome.

I wonder if he would do anything on like the solar side, right?

Yeah, yeah, yeah, that'd be interesting. And then there's also the option that we get some sort of Elon-style mega corp with Blue Origin. Involved at some point. Like, you know, XAI and SpaceX merged. It's possible that Blue Origin and this vehicle come together with Project Prometheus in some way, and then that goes public as a new entity. A lot of these, he's actually mentioned the whole space data center thing before, I think a while ago. Wasn't it like more than a year ago?

I don't think so.

Yeah, I think they also just filed something with the FCC.

Okay, yeah, for the new cloud.

Yeah, it was like 50,000 Satellites, I think.

Yeah, yeah, yeah. So he wants like a direct Starlink competitor. And so when you think about the full AI stack, if he's really off planet, you know, pilled and thinks he can get there even if it's like a year or two behind, or I guess what's the timeline for his Starlink competitor? That would be like 3 years behind, 4 years?

Yeah, I mean, for the moon, he's supposed to be ahead, right?

Oh yeah. Maybe this will all go on the moon. Well, either way, it's a big move from an experienced operator with lots of opportunity ahead, and it'll be fun to follow along.

And I'm glad he's back in the game in a big way. Yeah, not just doing random, you know, side projects.

Bottle service.

Or bottle service.

Yes.

He's got to be spending more time and energy securing America's future.

Yeah.

Securing bottle service.

Well, you know, I mean, The guy can take a weekend off, but one photo from his weekend flies a lot further than, you know, he could be completely locked in and just grinding 80 hours. And then he goes, gets bottle service just one time. He was just there for 30 minutes.

Riding on the yacht.

Maybe, you never know. Anyway, let's first tell you about Gusto, the unified platform for payroll benefits and HR built to evolve with modern small and medium-sized businesses. And let's also tell you about Figma. No matter where your idea starts, Figma Make, Cloud Code, Codex, or a sketch, the Figma canvas is where ideas connect and products take shape. Build in the right direction with Figma. And let's go over to Jensen on the All In podcast.

Engineer at the end of the year, I'm going to ask them, how many tokens— how much did you spend in tokens? And that person said $5,000, I will go ape something else.

Yeah, right.

If that, if that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed. Okay. And this is no different than one of our chip designers who says, guess what, I'm just going to use paper and pencil. I don't think I'm going to need any CAD tools.

This is a real paradigm shift.

Yeah.

Thinking about these all-star employees, it almost reminds me of what we learned in the NBA when LeBron James started spending $1 million a year just on his health of his body, like maintaining it.

That's right.

Here he is at age 41, still playing. It really is, hey, if these are incredible knowledge workers, why wouldn't we give them superhuman abilities?

That's exactly where does that go.

If we, if we extrapolate out 2 or 3 years from now, what is the efficiency of that all-star at NVIDIA and what they're able to accomplish?

They look like—

well, first of all, things that, that, that, wow, this is too hard. That thought is gone. This is going to take a long time. That thought is gone. We're going to need a lot of people. That thought is gone. This is no different than in the last industrial revolution. Somebody goes, "Boy, that building really looks heavy." Nobody says that. "Wow, that mountain looks too big." Nobody says that. Everything that's too big, too heavy, takes too long, those ideas are all gone.

You're reduced to creativity.

That's right.

What can you come up with?

Exactly. Which means now the question is, How do you work with these agents? Well, it's just a new way of doing computer programming. In the past, we code. In the future, we're going to write ideas, architectures, specifications. We're going to organize teams. We're going to help them define how to evaluate the definition of good versus bad. What does it look like when something is a great outcome? How to iterate with you, how to brainstorm. That's really what we're looking for.

The internet is having a lot of fun with this. Nuno says, guy that sells shovels says you should spend 50% of your salary on shovels. But of course, Jensen's point is generally correct, which is that you should give your best people a lot of leverage.

Yeah, I do wonder what the leverage ratio is in other industries. Like if you're a crane operator and you're making, I don't know.

If you should give him the best possible crane.

Yeah, like is it possible that if you're operating a $10 million crane or something, I don't know how much a crane costs, or a cargo ship, cargo ship that's ferrying oil across the world, how much does that crew cost? And then how much does the ship cost? And then what's the depreciation on that ship? Because it might be like a $100 million ship that will last 20 years. So you're looking at $5 million a year. And if the crew, total cost of the crew is like $1 million, well then you're actually spending more on the capital asset than the underlying talent. And you're getting leverage on top of that. And for a long time, software engineering has not been that. It's been $100 worth of internet and a couple thousand dollars on a MacBook and a lot of open source software 'cause you're using Python and Linux and these things. And I mean, I guess you could maybe burden like your cloud budget once you actually deploy the app. Like if you think about if you're a Facebook engineer or Instagram engineer, you could push a feature that consumes a ton of compute. But this is also going back to like the AI talent war where you get back into, if you're an elite software engineer and you're like, "Look, I'm gonna be in charge, I'm gonna be managing half a million dollars worth of token budget over the next year. Well, I want comp that's higher than that," or something like that. It'll be very interesting to see how the value works there. A reasonable ballpark for the annual cost of a fully loaded cargo ship is around $4 to $18 million. So including the cash operating costs plus the depreciation. So I don't know. Yeah, this certainly isn't like the first industry to be hit with something like that, but it'll be interesting to see how it changes.

In other news, the US Department of Justice announced yesterday 3 charged with conspiring to unlawfully divert cutting-edge US AI technology to China. The indictment unsealed today details alleged efforts to evade US export laws through false documents, staged dummy servers to mislead inspectors and convoluted transshipment schemes in order to obfuscate the true destination of restricted AI technology. China, said John Eisenberg, assistant, assistant attorney general for national security. These chips are the product of American ingenuity, and NSD will continue to enforce our export control laws to protect that advantage. Uh, so the company, uh, SMIC, Super Microcomputer Inc., uh, co-founder, they caught him, they caught him red-handed.

His name's Wally. He's arrested today or yesterday. He personally holds half a billion dollars of Super Micro stock, still risked it all, and now he's facing 30 years in federal prison. That is crazy. Uh, he was charged with smuggling billions in Nvidia servers to China, used Southeastern Southeast Asian shell company to funnel $2.5 billion in servers to Chinese buyers. $500 million worth shipped in just 3 weeks in spring of 2025. That's a lot. $2.5 billion in servers feels like enough for like a Frontier training run. Like that's a big, big, that's a big push. Built thousands of fake dummy servers to fool US compliance auditors caught on surveillance camera. Using a hairdryer to swap serial number stickers. And so OXGigi says, this man is a billionaire and was removing labels with a hairdryer personally. You're simply not grinding hard enough.

There's always, there's always a grindset lesson in any story like that.

Let's take this over to LinkedIn.

It's also notable because there's the export tax or tip that, you know, you're trying to bring chips out of the country, the the US government kind of flips over the Square terminal and they say tip, please, 25%. So you're looking at hundreds of millions of dollars if these were, even if these were chips that were able to be exported.

We were debating the flipping the iPad around, asking for a tip earlier today. And I think our joint stance was if the iPad is turned around, you got to tip. Yeah, it's just the etiquette is that you got to tip. But is there anywhere where that line crosses and you say, I can't, I can't possibly push something?

Funny thing is, Erewhon asks for tips.

Really?

On online orders.

Online orders? Wait, but there will be a human delivering it or?

No, there's two separate tips.

Okay.

There's the person in the store who puts the things in the bag and then there's the delivery person.

Yeah.

And they flip it over twice.

Yeah. The game theory of this stuff is always hard, because I feel like— I don't know. There's regulations on taxes on tips, but it's very unclear if companies actually funnel— how they distribute the tips. Are you tipping the person? Are the tips grouped together? Were your tips pooled when you were a valet?

Yeah.

They were pooled. So even if you did a great job—

Yeah, there's 2 jobs at the hotel, the valets.

Pool.

Yeah, Bell staff, the people that like take you down to your room, do not—

interesting.

Cool. Huh. And that was just the rules.

Oh, it's because I might be the one who parks the car, you might be the one who gets—

well, in a dynamic, in a dynamic where you have a bunch of valets that are waiting for cars, it would be a bad experience for the guest if the valets are like, they see a nice car pull in and they're like fighting over themselves.

Sure.

To deliver a service, right? It should just be like whoever's available to deliver the best possible service. Whereas Bell work is much more one-on-one. It's like I'm kind of your guide on the property. I'm going to like, you might have the person's cell phone, they might be texting you, et cetera.

Interesting.

And so it makes more sense to not pool.

Well, tip your export control, tip your federal—

I did have a funny story where there was a guy, a billionaire who used to come to the property and everyone knew that every Every single thing.

That's not just a word.

We still have— we haven't successfully kind of welded it into existence.

Not at all.

I think we're the only people.

We didn't come up with this term. It was delivered to us, and then we appropriated it.

There was a guy who would come to the property. Any single thing he got help with on the property—

He would tip?

He would tip $100. And so it did cause chaos. Okay, because then people are like, oh, I'd love to refill your Diet Coke. They'd be like waiting around, like trying to be in the right spot to open a door, 'cause they knew it was a bit, it got a bit out of hand.

I saw a piece of alpha on Instagram Reels that suggested that you tip the front desk person when you're checking in before you check in.

Yeah, that always feels like a bribe, but it is the right move.

This is a big thing in Vegas, right?

It is.

You're supposed to put it in between when you give them the card.

Oh, I saw this. Yeah, you stack the credit card and the ID and you put the $100 bill in there and deliver that, then they have plausible deniability. And they can be like, oh, here you go. Or they can be like, OK, thank you, and I'm upgrading you. And I think, at least this person on Instagram was making the case, that if you average out you're tipping $100 for every room, and you're buying a $400 room, but you get a $1,000 room upgrade every 5 times.

The only durable alpha is you you want to tip before the service occurs.

Yes.

If you tip the valet when they bring you your car—

Yes.

Yes.

How do they know what kind of service—

So if you're ordering at Starbucks and you want to make sure that they're going to enter your order correctly and get your name right, you got to tip up front and then say what your order is.

Walk up to the counter, say, flip it over.

Flip it over. I'm tipping you first. Now let's start the transaction. Anyway, when a broker who had bought Nvidia-powered servers from the Southeast Asian company—

Walk up to the counter and say, what's your tempo address?

Oh, yes, yes.

Send him a machine payment.

Yes, automatically. When a broker who had bought Nvidia-powered servers from the Southeast Asian company sent Sunny a text message containing a link to an announcement about Chinese nationals being arrested, for smuggling AI chips into China, Sonny allegedly responded with sobbing emojis.

Scroll up, scroll up so people can see.

Morning, Bruce. Hey, man, you're probably going to jail. Supermicro co-founder, crying emoji. And scroll down and you'll see Nick Carter, literally this: Your son has passed away. It's the warthog explosion. Oh, okay. I guess we can't find it.

It's weird.

My ex is different.

Jordan Schneider apparently was all over this back in August of last year. He said a super micro ad in the middle of a Reuters piece talking about how tracking devices and AI servers, just priceless.

Whoa.

What did Jordan know? What did Jordan know?

Deep alpha in China Talk. Head over to China Talk and subscribe. He's got some amazing content, some amazing interviews, and is a good friend of the the show. And so there's, there's clearly secrets to be unveiled through the China Talk archive. Let me tell you about CrowdStrike. Your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches. And let me also talk to you about Cisco. Critical infrastructure for the AI era unlocks seamless real-time experiences and new value with Cisco.

Thank you to the Cisco team for sending us these mini McLaren helmets. They're sponsored by McLaren.

Very cool.

And we are honored.

Yeah, we love F1. I'm sorry, but the Supermicro thing is awful, but parts of it are genuinely hilarious. They literally used a hairdryer to move serial numbers from real servers to dummy servers to throw in a warehouse and got caught on camera. And the pictures have actually leaked.

Literally caught red-handed.

Yes, it's a red— it's a red hairdryer. That is remarkable. Trung Phan says, "Do things that don't scale. Classic founder energy." Yeah, definitely, definitely that. What else is going on?

Let's pull up this video. I don't know if Bone—

It's 6 minutes long.

The legend Bone.

Chinese entrepreneur boasts receipt of 200 NVIDIA H200 GPUs in Beijing despite the US export ban and then explains how he circumvents the export ban.

Wow.

In January of 2020.

Beijing-based entrepreneur Su Di on the Chinese social media platform Douyin has sparked—

Wait, it's literally— you can see— watch this, watch this. Look at that. Rewind for a second. You can see a Supermicro logo.

No way.

Su Di on the Chinese social media platform Douyin has sparked significant discussion.

In the video, Su boasts—

Oh my God. That's like the biggest logo for Supermicro ever.

Wow, Bone is absolutely correct.

And this is from January. No, I'm telling you, like a lot of— I mean, I'm sure the folks in the Justice Department were working on this like earlier than this, but a lot of times, like legal action will happen like downstream of like YouTube drama videos. Like there will be some like, you know, fake guru who's out there like running a Ponzi scheme and people will be dunking on it on social media and then it will get picked up by the—

So almost a year and a quarter ago, somebody responded to Bone and says it's SMIC packaged. And Bone says, do you think SMIC is smuggling chips?

Wow.

And the— this is possible. It's a family-owned business in Taiwan. I won't be surprised if there are some family members who want to make this money by setting up fake shell entities and smuggling chips for business partners in China.

Wow.

It's good money.

Wow.

Wild.

This is crazy.

Wild.

Kevin Kwok has a good meme that everyone enjoyed. They expect one of us in the wreckage, brother. And it's not Jensen. The Bane meme. It's very, very funny.

And yeah, the SMIC founder, I believe, was at GTC just a few days ago.

Oh, really?

Hanging out. No way. I saw a picture of him with Jensen.

Yeah, until the music stops playing, I suppose. Let me tell you about Labelbox. RL environments, voice robotics, evals, and expert human data. Labelbox is the data factory behind the world's leading AI teams. And let me also tell you about Public, investing for those who take it too seriously. Stocks, options, bonds, crypto, treasuries, and more with great customer service. Speaking of going insanely hard, says Joe Weisenthal, I'm obsessed with the name of the rocket that will take the first Chinese astronauts to the moon. It's called the Long March 10. The next generation crewed launch vehicle. Long March is a great, great name for just locking in, the lock-in rocket. Lock-in rocket.

And it's time.

What are we talking about?

For Kimmigate.

Kimmigate. What's going on here?

Thin was.

So Cursor yesterday announced Composer 2, which apparently was composed of other models. Did very well on Cursor Bench. A little bit odd with the charts on the higher cost, lower cost, but seemed like it did well for Cursor's use case. Cursor fans were satisfied, I suppose. Finn says that he was messing around with the OpenAI base URL in Cursor and caught this, account/Anysphere/model/kimmyk2p5rl, and then some other numbers. So Composer 2 is just Kimmy K2 with RL. At least rename the model ID. Well, the co-founder, I believe, or someone from Cursor chimed in and said, and said it is in fact RL'd KIMI-K2, and this is why open weight rules. We the consumers get new improved models with minimal cost. And it's Lee Robinson who is teaching developers at Cursor. Lee Robinson says, yep, Composer 2 started from an open source base. We will do full pre-training in the future. That's an interesting claim. I'm wondering if they even need to. I don't know that there's any problem with using an open source model here. He says only one quarter of the compute spent on the final model came from the base. The rest is from our training. So they took whatever compute went into K2 and then they 4x'd that or 3x'd it and dropped that into RL. This is why evals are very different. And yes, we are following the license through our inference partner terms. I don't know. They're staking out their claim. They're making their strategy known. Jack Morris says, "People are taking this the wrong way. Look at it like this: Composer 1, DeepSeek probably. Composer 2, KIMI certainly. Composer 3 likely will be the first frontier base model pre-trained from scratch for a single domain: coding." It's a scary thought, isn't it? People are taking both sides. Tyler, do you have a take on this?

Yeah, I mean, I think it's very reasonable to do RL on some open-source pre-train. That makes a lot of sense, right? Because you can kind of imagine that there's this dynamic where— there's a post about this we can talk about. But so there's all this stuff about how, okay, Claude, Anthropic trains a model, and then Chinese open-source labs kind of distill on it. The actual implementation of how they distill it maybe doesn't matter that much, but you're kind of getting some form of Claude a little bit, that's open source. And then you take that and then you can do RL to do a specific hill climb on some task, like just coding. And that's a very good way to make a good model, right? So PrimeIntelect, their most recent release was, I think it was GLM 4.5. And then so they take that base model and then you do RL on that and you get this really great model because RL is where you can get a lot of these small gains on specific tasks, which can build up over time. I think it makes a lot of sense. It's not like a big dunk to be like, oh my gosh, they were just doing RL on this open source model.

I think it's a very logical strategy.

Yeah, that makes a lot of sense.

Yeah, I think the criticism of— I think they probably should have come out and said they did it and not tried to position it as like, we did a— they said what? This was our first continual pre-trained.

Yeah, continual pre-trained.

So they kind of like using that word. It would have been easy to be like, we took— We took world-class open-source models and improved them in order to create things for our specific problem.

Yeah. And I think people would've been like totally fine with that, right?

Yeah.

It's probably meant to like, I mean, a lot of, you know, American labs have done this and said it and everyone's like, wow, this is great. This is a great model.

Yeah.

I'm, you know, excited to use it.

Yeah. Didn't Perplexity at one point create like DeepSeq-1776 or something like that where they were like, we specifically made it more American.

After, I will say after I actually created it.

Oh, after you did that. Yeah. Okay. Yeah. That's so funny that you did that.

Zephyrus says, "Cursor Bros are treating Kimmy like Voldemort." I don't want to say the name. He did in fact put it in a post.

Yeah, yeah, yeah. Accelerate Harder says, "The funny bit is training from scratch impresses investors and Twitter, but it makes little sense for most teams while strong open source options exist. So Cursor actually did the smart thing, but pretended they did the dumb thing, and it nearly worked." So more of maybe a comms issue than a business strategy issue. But we'll—

Yeah, I think there are some concerns maybe about the actual license of the model. Because technically, you are supposed to say that it is like KIMI.

Yeah, usually you have to pass through these things. Android is built on Linux. And that's in the— There is release notes.

Likely to be some deal behind the scenes that actually means it's fine.

Yeah.

We really have no idea.

Also, I would not be surprised if people have just not read the Composer 2 license doc yet or something. And it's actually just fully disclosed there. And then people are like, I found in the URL. It's like it was also in the terms of service. Like it was just in plain text somewhere that you just didn't find. Who knows? Let's see. So 3x— Ellie is giving some comps on this. So 3x the training compute gets you a 1% improvement on SWE Bench Multilingual and 21% on Terminal Bench 2.0. But K2.5 is in non-thinking mode. If those benchmarks are useless, it's weird that they're the ones reported in the Cursor blog, then something is wrong. So people are going back and forth, but ultimately I think this, you know, where the rubber meets the road is ARR. Like, Cursor has been growing, Cursor continues to grow, even while they've, you know, been like steamrolled and the vibes have shifted. If people love this and people are using the product and it's a growing business and the market is growing so fast that they have a great business, that seems like, you know, business continues as usual. Let me tell you about the New York Stock Exchange. Wanna change the world? Raise capital at the New York Stock Exchange. Let me also tell you about Vibe.co, where DTC brands, B2B startups, AI companies advertise on streaming TV, pick channels, target audiences, and measure sales just like on Meta. So should we move on?

Delvegate.

Delvegate, another gate, another gate, Delvegate. So what happened here? Where should we start?

Anonymous Substack pops up.

Yeah, these anonymous Substacks are interesting. DeepDelver clearly created just to share this particular story. No idea who's behind it.

Yeah, incredible, incredibly professional.

Okay.

In their presentation, right? They're putting backups of the article. Yeah, on day one. They're saying reporters who want to get in touch.

Yeah.

Whistleblower.

They're sharing a Proton. And so, yeah, they make an argument that Delve has had a bunch of kind of nefarious activity, kind of duplicating reports, not basically just like a lot of their value prop I think was on the fact that they were using AI and speed. And yeah, it doesn't look good at all.

I don't believe we've had Delve on the show, but I searched the calendar and I got one hit for Delve and it's in a summary, it's in a summary of, of Bill Gurley's book that says, "It's a page-turner that delves deep into the surprising, often zigzagging career paths of exceptional performers. There's something inspiring on every page," is a quote from the New York Times bestselling author that he gave to Bill Gurley's book and just sort of wound up in our notes. Anyway, so what actually happened? What is Delve? Let's start there.

Delve does SOC 2 compliance and HIPAA compliance. They've worked for a number of companies like Lovable, Clue.ly, et cetera, et cetera. And yeah, you can kind of, I mean, the article is like very, very, very long and detailed, but it just goes into—

Erin Griffith shares like the key claim. Erin Griffith, of course, is over at the New York Times. And she quotes this, Delve built a machine designed to make clients complicit without their knowledge to manufacture plausible deniability while producing exactly the opposite. Then a lot of people are having fun with the fact that the Dell founders were in the 30 Under 30, which of course has surfaced a lot of people. I think there's, are there really 1,800 30 Under 30 every year? That's true? Or is that like DocuSign?

I thought you were saying all in.

Oh, maybe overall. I know that there's multiple categories.

So there's more than 1,800. I think it used to be that there was It's just there was actually 30, and then Forbes realized, like, what if we scaled this?

Yeah, this is the plan. Scale the Midas List. 100 Midas Lists, so you get 10,000 VCs paying you top dollar to rank. It should be an auction. So Austin Petersmith says, not going to lie, as we've been going through a grueling SOC 2 process with Vanta, one of our sponsors, I have felt a lot of FOMO reading about Delve customers getting it done in 3 weeks. If all this is true, then no more FOMO. So he's—

Milo—

there are some things that you just can't short.

Milo pulled up a post from the CEO from April of last year that says— the CEO says, when you step into the office and your founding AI engineer is on his third all-nighter, his team never stopped shipping. Milo says, third all-nighter of what? Compliance? I don't want my security company shipping 3 AM code. So, very unfortunate situation. Not gonna make any real assumptions, but this story's definitely not over.

Well, let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI, own the data platform that powers it. And let me also tell you about Lambda. Lambda is the superintelligence cloud, building AI supercomputers for training and inference that scale from 1 GPU to hundreds of thousands.

Okay, so the snake post.

We're not doing the snake post. We're going to the snake? No, I hate snakes. We're not doing it.

Whoa.

I refuse to pull up a snake post. We're skipping straight to Bruce Lee.

Okay.

We're skipping way down.

Okay, I'm not gonna make John watch a snake video, even though you wanna close your eyes.

I hate snakes. I saw it, I saw it.

This video is definitely, It's like terrible to watch.

It's also fake. It doesn't exist. I'm Irish. No snakes in Ireland.

We got snakes on TV. We gotta do that.

Snakes are banned.

We gotta do that bit where Snake Man catches like a rubber snake.

No, no, you will not do that to me. Under no circumstances will you do this. We will be watching Bruce Lee fight Chuck Norris in The Way of the Dragon because this is the way that we pay our respects to Chuck Norris, who of course passed away at the Age of, I think, 86. Norris. He was 86, and this is one of the most iconic fight scenes in movie history. Have you seen The Way of the Dragon?

Of course not.

Of course not. This scene was unique. It was set inside Rome's Colosseum, and it was often cited as a turning point for how martial arts were portrayed on screen, particularly because they didn't use actors. So both Bruce Lee and Chuck Norris were legitimately high-level fighters. Norris was world karate champion at the time. They cast him in part because of that, and it has a much more like grounded, credible feel. As opposed to like previously more stylized kung fu films. And you can see this with the— Bruce Lee actually directed and choreographed this himself, so Lee had full control. So the director didn't say anything. And they do these cool like punch-in shots. Look at this. So you're seeing nice wide, you can see exactly where he is. There's no body double. And then we're going to zoom in. And then what are we going to do again? We're going to zoom in again. And so we see his face and he's just sitting there waiting. Brilliant shot. So, so much time to breathe and understand and feel what he's feeling. The music, footsteps, breathing, clothing movement, the cat, of course.

Any snakes?

Yeah, no snakes, I don't think. I haven't watched the whole thing. The camera often stays wide. This is— this stands in contrast to something like The Bourne Identity, very like fast cuts, close-ups. Much easier to get something that feels intense when you're cutting and chopping. It's much harder to make something like this where you see the full body. You see now we're in slow-mo. They're, they're, they're shotting back and forth. The empty arena gives a gladiatorial vibe, a sense of ritual combat. Visually frames the fight as mythic, not just physical. Bruce Lee is learning mid-fight. Norris has the edge with his rigid karate form. But Lee adapts. He loosens his stance, increases his mobility, switches his rhythm. We got to talk over it or else we get demonetized and banned. So I got to figure it out. Anyway, we're very sad that he is no longer with us, but he leaves behind a massive legacy and a portfolio of films, including the TV shows as well, Walker, Texas Ranger, which was constantly on TV. When I was a kid.

Well, rest in peace to a legend.

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. And let me also tell you about 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.

So Bernie Sanders had a conversation with Claude. Video was posted yesterday. Yes. Justine Moore says he did the meme.

He did the meme.

Bernie says pretend to be a scary robot. I'm a scary robot.

Yeah, yeah. Taylor Lorenz chimed in on this too. This reminds me of that tweet that was like typing I'm evil into a Google Doc and getting scared. Yeah, there's a little bit of that. Will Manidas had another take. He said a trillion dollars was spent on AI risk teams and no one thought to hardcode into the model to not tell Bernie Sanders that you do crimes. Very funny. But this is the correct way to disclose that you're talking to an AI model. Yeah.

Do you think he was inspired by the Vanity Fair piece and then immediately shipped this?

The Vanity Fair piece, like, buried the lead and like it was surfaced in a bunch of screenshots later that made it look like the entire article was AI-generated, which it wasn't. Like, the Vanity Fair article did have access to a bunch of people and there are some interesting vignettes in there. But this is very clear from day one what's happening. Like, he's talking to Claude and he's investigating and, you know, sharing his fears, and I think this is a great way to actually share what you're doing and make it very clear. And so people are mixed on this.

Miles Brundage says, worst part of the Bernie video is using Sonnet. Not sure why folks are so mad about the other stuff.

That's very funny. And then Delicious Tacos. Delicious Tacos. I also smoked a grog. It's Bernie going like this. No, no, don't click that. But you can imagine what happens there. And then, yeah, Hope's Revenge did—

made the card for us.

Thank you.

Signed.

Signed in there. It's great. Anyway, let's move over to Montana. One and only Moonlight Basin. There's a property that made it into the timeline.

Ooh, indoor pool.

New.

Okay.

This is a new hotel. Okay. Got to visit the property a few weekends ago with Paul.

I saw some promotional material for this.

It is absolutely stunning. I want to have Sam, the founder of Cross Harbor, who's the developer, on the show because he is, he's basically, in my view, felt like he was the mayor of Big Sky. He's developed and all these incredible properties, donated a hospital to Big Sky too. Just said, here's the hospital for the local community, which is amazing.

Does it have zebras?

Not— no zebras.

Does it have tortoises?

For $5.1 million, this safari-like estate comes with zebras.

Oh, you got a solution to my problem. I need zebras.

Animals roam free at California's 137 Oak Hill Preserve, which has a 5-bedroom lodge and a helipad. I'm loving this.

This is crazy. Okay, wait, I, I can't believe, uh, the, the value here. Uh, I feel like anything that has zebras and tortoises should be up in the double digits.

You should get a zebra. I don't know how much it costs to get a zebra, but you should get like a 10x multiplier on it.

Yeah, for sure. So, Ron Kraft visited Africa for the first time in the '80s, he said, but his heart never left. He proposed to his now wife, Bernadette Kraft, in a hot air balloon. That's elite. Above the Serengeti National Park. The couple got married in Nairobi, Kenya in 1988. And 2 decades after, 2 decades ago after taking their 3 sons to Victoria Falls in Zimbabwe. Brian started wondering how he could bring some of the African experience closer to home. Can I recreate a mini Africa that's a short distance away from our house? Thus began a years-long process of developing a roughly 137-acre safari-like sanctuary in Ione, California. Is that Ione or Iony? I don't know. With a helipad and a collection of animals including tortoises, alpacas, pacas, East African plains zebras, and—

I think it's aony.

Aony, okay. So now approaching the age of 70, Kraft, who is a commercial real estate broker, entrepreneur, and aspiring screenwriter, is listing the vacation property for $5.1 million. Approximately 15 animals are included in the offering. That's only less than a 10x multiple per animal. Known as Oak Hill Preserve, the estate is about 40 miles from Sacramento, Where the Crafts live, the animals roam freely on the property with access to plenty of water and grass. Kraft likes to drive around the estate in an old pickup truck he outfitted with benches in the back and lights in the front and rear. The idea is to go looking for animals like you would on a safari. Kraft assembled the property by buying 3 separate parcels over a 5-year stretch starting in 2007. It had the right amount of open space like a savanna and the right amount of trees like you're in a thicket. Kraft improved the land, of former cattle ranch bit by bit, starting with no-climb fences, roads, and underground power lines. Throughout the property, Kraft said he tried to evoke Africa with curved roads and buildings with rounded edges. One of the first structures he built was a bunkhouse for a group of friends. "I've got a group of knuckleheads," he said. "I did it with them in mind. It was a guy thing.

The wives put up with all of us." Someone says knuckleheads and you don't immediately think of a specific group?

This is amazing.

This is California.

Just driving around this California estate and just seeing zebras roaming is amazing. Ostriches. This is, this is a treat. This is a treat. A few years ago, Kraft said his friends chipped in to buy him a Jeep with animal print paint. We were all having breakfast. All of a sudden I hear the theme of Out of Africa, he recalled. Then one of the guys pulled up in the Jeep. It was quite a surprise. He completed the main lodge on the property around 2018, spanning around 4,000 200 square feet. It has 5 bedrooms. There's a cafe for dining near a creek on the property, as well as entertaining space that Kraft calls the cigar bar.

John, you have to get this place.

This is extremely me.

This is so you.

This is extremely me. Kraft said that he is at the estate every week. It's basically just a family getaway. His children and their friends spend time there, too. And his oldest son got married there.

Wow.

Why?

Why is he selling?

Why is he selling? What's wrong?

This is—

this has got to be some Jurassic Park style thing where, like, he's been doing some rare zebra crossbreeds that are about to take over. He's like, yeah, the electric fence works well. It's no climb. You're like, but they've learned to fly. I'm not— I'm not exactly sure. The property has many outdoor amenities, including a basketball court. That's going to be big. Basketball is making a comeback.

Craft is selling his bedroom suite, but he promised his wife to list the property when he turns 70.

Oh, okay.

That's when, that's when you should be selling your primary home.

Yeah.

Moving to the, to your, to your ranch.

Okay. I'm, how far is it? Ione?

Ione in Amador County is home to wineries and equestrian properties and it has become a second home market for people from the Sacramento area. The local luxury market is strong. Yeah, it's really far, but it has a helipad.

Yeah, 6 hours, 6.5 hours straight shot, maybe 7 hours straight shot in a car. But if my plan to ban speed limits happens and you can drive there 200 miles an hour instead of 65, you'll get there in 2 hours.

It's a flat 2 hours right now from San Francisco. But get this, by train, it'll only take you 6 hours.

Wait, did you guys miss this line? It says one of his friends discovered what he believes to be a century-old gold mine.

Wait, really?

Okay, he's just pranking. He's pranking everyone right now.

This is crazy.

He just— I think he wanted to just show off his— I think he just wanted to show off his ranch. So he's like, yeah, I'm selling.

I'm selling. He's not selling this thing.

There's no way. Zero chance. He just wanted the Journal feature.

We do need to get this guy on the show, though. He seems like he's just had a great life and a great time. Seems like a good person to learn the ins and outs of commercial real estate from. Quickly. Let me tell you about Plaid. Plaid powers the apps you use to spend, save, borrow, and invest, securely connecting bank accounts to move money, fight fraud, and improve lending now with AI. And somebody who is selling is Orlando Bloom. He put his longtime Malibu home on the market for $12 million. He spent millions redoing the 4-bedroom house. Now it's for sale. It can be you. It is above El Matador State Beach. You're going to have to tell me if that's a good beach.

Is this a good beach?

It is a fantastic beach.

OK.

It was the actor's primary home for a full decade.

A lot of tourists there.

A lot of tourists. Okay.

Which is somewhat of a downside.

Somewhat of a downside. Okay. So it's in a gated community. Stunning. It's in a gated community. Only Jordy can tell us if it's a goated community, though. The Malibu house has interiors by the artist and furniture designer Roy McMackin. Bloom did an extensive renovation that cost more than twice what he paid for the house. He bought the house in— where he bought it in— sold. He sold it separate, Malibu house, in 2024 for $4.1 million. How much did he actually buy the house for?

I forget, but $2.5.

$2.5. And then I guess he spent more than $5 renovating it, and now he's looking for $12. So he said he's reluctant to let go of the Malibu house, but he rented it out last year and realized that he likely was not going back there. It felt like I think it might be the right time to let it go, he said. And he's somebody that picks up the phone when the Journal's calling, which we love to hear.

Listing agent is Chris Cortazzo.

No way. Again, Cortazzo.

Basically, Chris has almost half of all the high-end listings in Malibu.

Do you know how he described the ground floor when Orlando Bloom bought this house? He described the ground floor as a bit higgledy-piggledy, which is a great line. It was a bit higgledy-piggledy, but I needed it for my knuckleheads. I needed it for my knuckleheads. He said he's got the most amazing drives through the Malibu canyons. He's very—

that is true.

It has a pool, a Jacuzzi, and a sauna, I believe. And so you can go check that out. There are, there are some other amazing properties in the mansion section, but we We'll have to get to those later. First, let me tell you about Vanta, automate compliance and security. Vanta is the leading AI trust management platform. And let me also tell you about Graphite, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. And without further ado, let's bring in our first guest of the show. RF Kenmore is in the Restream waiting room. We'll bring him into the TVPN UltraDome. He's a menswear insider. And we can give him a couple minutes because there is A tequila billionaire who is selling a Balinese-style Hawaii home for $32.5 million. It's designed to look like a temple. The house on Kiholo Bay was built in Indonesia, disassembled, and then rebuilt in Hawaii. That is crazy. Billionaire John Paul DeJoria's Hawaii home has traveled almost as much as he has.

I like this home. Just take the home and bring it to another island.

That's pretty remarkable. So in the '80s, he had the Balinese style— Bali, from Bali. That home designed and constructed by artisans in Indonesia, then disassembled and shipped to Hawaii's Big Island, where it was rebuilt on a waterfront lot. The process cost a couple million dollars. He's a co-founder of the tequila brand Patron and the hair care products company John Paul Mitchell Systems. Now the property with shingled roof designed to resemble dragon skin is coming on the market for $32.5 million. What a remarkable property.

You really haven't lived until you've gotten food poisoning in Bali.

I've never been. Have you been to Bali?

A couple of times.

Surfing?

Yes.

A couple of times. Isn't it a really, really long flight?

It's a terrible flight.

It's a terrible flight, right? It's very inconvenient.

Especially when you're a college kid and you're broke and you're just like— I used to, with surf trips, I'd go and I would just search the rough dates that I'd want. And then I would just buy the cheapest possible flight, which would usually involve a 17-hour layover in the worst place on Earth.

Amazing.

But I would do it because my time was basically worth nothing.

Lovely.

And it was worth it.

But shouldn't you have just gone to an island that has similar features closer? We have the Caribbean. Is there not good surfing in the Caribbean? No.

No? You can't surf in the Caribbean? No, no, it's not to say there isn't good surfing in the Caribbean, but the Caribbean, all areas are seasonal, right? So different times of year, it's going to be better conditions. But the Caribbean, you have situations where like it's really good for 6 hours and then it's flat.

Okay.

Whereas Bali, you're gonna have like more consistent surf and it's an island, so you can like go around it depending on the conditions and whatever swells.

Okay.

Well, I believe we have our next guest in the reshroom waiting room. So let's bring in RFK Moore into the TBP and UltraDrum. How are you doing? Good to meet you. Thank you for taking the time.

How are you?

What's going on, boys? How's it going?

It's fantastic. And looking sharp of the weekend. You are looking sharp. We love having anonymous commentators on the show. What can you tell us by way of an introduction? What is your focus? How would you introduce yourself without doxxing yourself?

Yeah, big fan of menswear, big fan of the clothing industry, brands, apparel, products. I've worked in men's brands in the industry for over 10 years now.

Overnight success. It also says here you're a hotel bar evangelist. Explain that.

That's right. Yeah.

What's so special about a hotel bar versus just a normal bar?

It's an elevated experience. Why?

It doesn't necessarily need to be an elevated experience.

You keep the riffraff out. Why?

You can drink around your hotel bar.

They don't know about it. They don't follow RF Kenmore yet.

Okay.

Okay. It's a bit of a filter. You don't always have people wandering in off the street. Sometimes the bar, you have to go through the lobby to a room to a room. Just kind of filters people out, you know.

I wanted to get the update on the menswear industry broadly. Where the— like, what's happening in clothing?

Who's the biggest winner right now?

Yeah, who's the— yeah, who's, who's the biggest winner?

Yeah, who's on an absolute, uh, like, generational run?

Uh, I mean On the largest level, I don't really know in terms of luxury or those like major players, but on the more like specialty sort of scale, probably Buck Mason, you know, kind of big in the conversation. They're opening tons of stores. I think they're up to like 50 or 60 stores now.

Wow.

Men's From what I last heard, men's still drives a pretty sizable portion of the business, which you don't really see. Usually women's overtakes as soon as you introduce it.

Oh, interesting.

So yeah, probably.

So even— so, so would you expect at some point women's to overtake men's at Buck Mason, or is it just the name is like so— it's just incredibly masculine, but maybe there's still an opportunity there?

Yeah, I would probably, I would probably expect it to. From what I hear, it's doing really well. They have sort of a good like foundational business going with t-shirts and sweats and sort of like coastal, you know, linen and good fabrications.

Well, I would say a large part of their success is that they put a vintage Porsche in their stores, right?

Yeah.

Have you guys ever seen their Porsches they have in the stores?

Do they actually?

We went to that suit fitting with, uh, yeah, uh, it was at a Buck Mason.

Oh, that was—

that's a Buck Mason.

And they did have a vintage Porsche. Yeah. Yeah, yeah, yeah, yeah. Where did that trend come from? Is there like a clear lineage, or is it just like a broad trend?

We need— we need—

you guys know Know Your Meme? Yeah, where it like traces the origins of every meme. We need to know your Porsche.

Yeah.

Um, I don't know. I, I, I know that ALD had the, the Porsche in their little SoHo pop-up that they did for their Porsche collaboration, and that might have predated any, any Buck Mason Yeah.

Yeah.

I mean, it makes total— I mean, it's funny because I've always appreciated— I really love your posts when you're just kind of joking around, be like, no, they couldn't possibly put a car in their retail store because it is so cliché now. But I think it's a way for— it's just a way for a brand to signal like who they are because cars are extremely personal.

Yeah.

It says if you're a car person and you choose to invest your money in cars, your car does say something about you, whether you like it or not.

Yeah, I would love to walk into a menswear store that had like a Dodge Viper sitting there, like an SRT, or yeah, maybe like a Dodge Durango SRT.

That'd be ideal.

I'm sure we can find—

What?

The three-wheeler?

What's that one that they always play the loud music out of?

Oh, is that the Ariel Atom? No, the Slingshot or something. I forget.

Yeah. Who's going to be the first? Who's going to stand up and be the first?

That would be good.

I like it. What is, what is actually going into— how does a brand thrive today? Everyone knows that apparel brands, you know, you're put at— you're having to spend so much money on inventory. You don't necessarily know what's going to be hot. But what is— what are the brands that are doing the best in menswear doing from a supply chain standpoint? Are they, you know, leaning into US manufacturing? Are they, are they getting, or are they still like dependent on, on Asia or, or Latin America? What's, what's the approach that's working?

Yeah, I mean, there's no one in terms of supply chain or sort of like the operational side of the business. I don't think that there's one model that, that you have to adhere to. I know that COVID really scared a lot of brands into diversifying their supply chains. There were some brands that were just super, super heavily concentrated in China, even with just one vendor base. And so that's really forced people to look into Latin America, India, different areas like that. But as far as like what drives success, I think community is more important than it's ever been. You know, you look at the—

what does that actually mean though? Because usually when people, like, for example, like, I've bought a bunch of stuff from Buck Mason over the years, but I've never been like identifying as like, I'm in the Buck Mason community.

Community to me would mean like, you walk in and there's, you know, you got your Dodge Viper there, and then they hand you a White Monster, and there's like a mosh pit that that you can like go and like thrash to like Limp Bizkit with in. And like, you're going to be fighting guys in the mosh pit, but you're going to be building community through that.

Got it.

That's what it means to me. Okay.

But maybe you have a future in menswear.

I think so. It's possible.

You look like it. You're looking, you're looking dialed. You're kind of Mog and Jordy right now.

Yeah.

Casual Friday over here has crept into every day.

Well, you almost wore jeans today.

I was very close to wearing jeans. But I did throw on the dress pants. Walk me through the guide to visiting New York. We head to New York every once in a while. We're usually at the New York Stock Exchange downtown. I believe it's in downtown. Great approach there. Where's the best place to live? Where's the best place to hang out? What's the guide for the West Coast fellow who's visiting or planning to move to New York?

Yeah, everyone's got different— that's kind of the cool thing about about Manhattan, that there's kind of different flavors depending on which neighborhood you go to. Um, I think the Upper East Side is coming back. People are rediscovering the Upper East Side, like, kind of like they're the first ones to do it, which is funny. But my favorite neighborhood still is like the Nomad, Flatiron, uh, sort of area. It's central, but it's, it's not It's not in the middle of anything, if that makes kind of sense.

Does the city have a feel of like it's back? Because there was a lot of like doom and gloom over like, Mom, Donnie, everyone's going to move to Florida, everyone's going to leave. And but what's the actual vibe on the ground been?

I—

yeah, I think it's the same as it's always been. I think, you know, you walk around anywhere, anywhere busy, you know, you go to any of the shopping areas, whether it's Fifth Avenue on Flatiron, or you go to a nice restaurant somewhere that's popular, they're slammed and they're impossible to get into. So, I don't think—

What is your pitch for tech people who want to dress more, want to elevate their wardrobes in New York specifically? I feel like the finance guy, has some sort of shape to them that's been established going back. There's lots of great reference points. You can say I'm a garden gecko guy, or I'm a Patrick Bateman guy. There's plenty of iconic cinematic portrayals of the Wall Street guy. Less so in the New York tech scene. The New York tech scene is certainly growing, and it feels like it has an opportunity to distinguish itself from the Silicon Valley scene.

I don't know. I think you need to establish like a tribal council of tech leaders to really hone in on the personal aesthetic. But if you're an individual, I mean, I think a uniform is always a great place to start. Just pick a couple of good things, a top, a bottom, and then buy really good versions of those things that are obviously well made and that fit you really well. If you need to take them to a tailor or whatever to make that happen, that's good too. And then just buy multiple And then from there, you can kind of introduce maybe a stylish piece that you found really interesting that just happened to be kind of like a one and done.

What's going on with various lifestyle brands in different kind of subcultures, whether that's running or golf? Are they all going to converge on a long enough time horizon because they need to scale? What are you seeing there?

Well, I don't know if you saw my tweet about Malvern. I guess they're converging. They're just grabbing at whatever hype they can find transitioning from golf into run clubs. But I don't know. I mean, it's hard. If you look at the hardcore running brands that are kind of driving the culture right now—Bandit, Satisfye, there's a couple smaller ones. Like, uh, OOVOO, or I don't even know how to say it, um, in the UK, uh, or SORE. Could you, could you imagine a world in which they, they pivot into golf, or they expand, I guess, better phrased, into golf? I don't know. I think that would feel kind of disingenuous, and you would end up with this, like, disparate feeling brand identity and assortment. So I don't know. I think lifestyle is probably the more the more natural.

Yeah, it's an interesting dynamic where it's easier to break out and establish a foothold as a brand by really focusing on like a subculture like running. But then can you— is running, you know, what are you running from? Right. Is it somewhat of a— is it an Instagram kind of, you know, fad? Right. Are people— can running support 10 different, you know, microbrands over time? Probably. But at the same time, like, it feels like very, like, on trend right now, too. And yeah, we'll see. Yeah, we'll see how durable it is.

So outside of following you, what's a good way for folks to get into menswear? If I'm talking to somebody, they're interested in cars, I might recommend Doug DeMuro, go down that rabbit hole. you know, there's a whole— you could pick up a copy of Car and Driver, you could start watching Drive to Survive, or just watch F1, you'll learn about cars broadly. What are the different entry points? What are the great, you know, either creators or publications that you think help people get up to speed on menswear?

Yeah, that's kind of why I started tweeting, honestly. It's hard. I don't know that there is that there is one. And I think the offering in terms of voices and publications and more of like the institutional side of it, they're, I don't know, I feel like kind of aimed a little high. Like they're looking at temporary brands, they're looking at runway, they're looking at things that may not be super useful to most people. And so I found a really hard time trying to find relatable, digestible content or advice. Sure, sure. Yeah, for normal people. Who you got?

Jeremy Kirkland, Lammo. Okay, he's a buddy. Cool. But, uh, he's had a— he's had a great podcast forever that kind of goes through and interviews a lot of the great kind of founders and creative leaders in the industry.

Yeah, he's cool. There's, um, there's that dude, uh Now I'm blanking on his name. I don't know.

Well, last question. Do you think surf brands can ever make a comeback? Do you think that any people will be like, I got to get the new Billabong drop. I got to get the new Quicksilver. Is there a chance? They've fallen off so incredibly hard. What else would you want? The Warped Tour. Yeah, the fall off is just insane. Like young teenagers, they want to wear Chrome Hearts and, you know, and streetwear. They don't want to be caught dead in Billabong, but everything, all these things come back around eventually.

Yeah, it's kind of funny, I was just talking to a buddy in the industry about this, how Billabong, Quicksilver, whatever, kind of, you know, they were tapped into culture, they were kind of running things for a little bit. But now that, I mean, that vibe is probably running brands or, or golf brands maybe a little bit. Um, there is the one brand, uh, Florence Marine X, which Jon Jon Florence is the face of. Yeah. Which is backed by the Hurley family, and he's having a go at this sort of like modern, uh, take on, on a surf brand. You have Outerknown as well, uh, Kelly Slater's Outerknown brand that he co-founded. It's definitely more of like a lifestyle.

Yeah. The, the question though is can any of these brands like break back into culture?

Can Billabong or Quicksilver break back specifically? No, no.

I'm just saying like even Outerknown and Florence Marine X, like you don't see like Instagram kids in a GT3 RS, you know, rocking like, you know, Outerknown. And my question is like, could you ever see one of those people, one of those archetypes, Rocking a surf brand, it seems like a stretch.

But it's hard. Yeah, it's hard to imagine. But I don't know. I mean, crazy things happen.

Billabong's got to sign ASAP Rocky as creative director.

Most of the GT3 RS owners seem like the type to put the surfboard on the roof, which is ridiculous. Obviously, no one would ever do that. But I mean, this is Guns of Baldur. Anyway, thank you so much for doing this.

Great to meet you. Thank you for all of your commentary. Enjoy your weekend.

And we'll chat with you soon.

Go find a Porsche store somewhere. We will.

You're the man. Let me tell you about Turbo Puffer, serverless vector and full-text search built from first principles and object storage, fast, 10x cheaper, and extremely scalable. Before we bring in our— Oh, is our next guest available? Yep. Fantastic.

Let's bring in Mitch Lee from Arcboats.

That's the sound of the boat going. You need to play that constantly. Let's bring in Mitch Lee from ArcBoats. How you doing, Mitch?

What's going on, Mitch?

I'm doing well. Thanks for having me.

Thanks for hopping on. I'm super familiar with you. I think we talked on the phone a couple of years ago, but since folks out there might not be, please introduce yourself and the company.

Yeah, my name is Mitch Lee, co-founder, CEO of Arc. Arc is on a mission to electrify the marine industry, rebuild this very outdated industry around more modern electric powertrains. This is similar to what happened in rail many decades ago. Every train today is electrically powered. They're diesel electric, hydrogen electric, battery electric, but they're all electric. Marine industry is gonna do the same thing and we're out to a big lead there.

Yeah, we were talking to Travis Kalanick about how I think he got like a tour of Saronix factory and was like, I want to water ski behind that. But I think you can technically—

He's so addicted to water skiing. He wants a drone to help him water ski because he's like, I can't get anybody to drive the boat.

Yeah. So talk to us about what can you do on an Arc boat? Fish and wakeboard and water ski? Is everything on the table? And then, yeah, is there an autonomous vision in the future?

Yeah, uh, so briefly on the business, where we started was we, we had this core thesis: the entire industry is going to go electric over the next decade or so. We started by hardening that technology in the consumer space, so our first boats to market were consumer boats, a speedboat that you could water ski behind. Now we're ramping production of a wake sport boat, so something that you wakeboard behind.

John's a lake guy, I'm more of an ocean guy.

This is— it's lake boats.

Lake boats. That's my— that's what I grew up with. I love it. Uh, so we're ramping production of that.

What's the goal? Are you trying to— is there anything you can do on the electric side to make like the biggest wake ever, or is it more just kind of like, yeah, we want kind of something like that?

Like the Plaid was like, for the price, like the fastest 0 to 60 time. Is there an equivalent like Oh, electric unlocks this, because obviously everyone knows about the environmental benefits, the cost benefits, like maintenance, but the maintenance. But like, is there going to be like a killer feature from electrification? There is.

I mean, there are multiple killer features. One of the biggest ones is actually sound. Okay. The— it's quiet when you're on the boat and like operating. So you can just talk.

That's amazing. That's amazing.

Uh, you don't smell the fumes of it.

Um, the torque on it. So that's a— that's a downside for John. He loves the smell of, you know, fumes in the morning.

Yeah, seeing the oil pool as I jump into the water.

Exhausting down into the water and cancer that you're— yeah, it's great. Exactly. No, but the torque at low speeds is very high, so you can— when you punch the throttle, it's immediate acceleration. The way you corner, it handles like a jet boat. So, you could whip this— it's electronic steering, electronic throttle. Okay. So, you could whip this boat around with like, you know, steering with one finger and throw people off the boat because it's so powerful. And then it's software-defined, so every part of this is, yeah, powered by software. You've got in-house firmware, in-house software, in-house telemetry system, over-the-air updates. You hit one button, the wake is perfectly configured for whether you're wakeboarding or wake surfing on the left or right side of the boat. We can set that all up in software, and it's just a punch of a button.

That's amazing. So talk to us about where you're going next, how much money you've raised, what the plan is going forward, because obviously you found this beachhead metaphorically and you're expanding.

Yeah, we just closed a $50 million Series C. That brings our total amount fundraised—

That's great. Love the energy of this.

Play the boat sound. I want to hear the boat sound again. That's good.

That's good. Gets me fired up in the morning.

I love it. That should be your alarm clock though. So 50 miles.

That's my, anytime somebody calls me, that's the sound that it makes. That's great. We just closed that.

Sorry, it's Friday. We're having fun.

Sorry, don't worry, I'm hyped up on Celsius. We're good to go. Good, good. We— so we just closed that round. It brings our total funding up to $160 million. And the— again, we harden this technology on the consumer side. We're now— we're at a— call it a $40 million run rate on deliveries of those boats every month. It's going up every month. Our margin is expanding. Yeah, we are now expanding that core technology, the platform that we built to commercial applications and defense applications. We signed a $160 million deal for ship assist tugboats, so giant tugboats out on the water that push and pull cargo ships. Oh, wow.

So that rapid acceleration and that like turning capability that you mentioned, I imagine is huge for if you're like, makes a tugboat potentially a lot more like accurate. Or is that—

You definitely get a lot of torque benefits from it. But a big part of it is they care about profit. They care about— like, tugboat operators want to make more money. And the way that you make more money is you cut your OpEx in half by reducing the fuel costs, by reducing the maintenance costs, by keeping that vessel, which is a $20 million vessel, online and running more— like, a greater percent of the time. And then you also avoid a bunch of compliance burdens. These things are cancer-spewing machines. So by going electric, you avoid a lot of the compliance burdens for like, hey, how do you EPA more cleanly burn bunker fuel? Sure. Instead, it's like this thing doesn't even have fuel tanks on board. You can put a diesel generator on the back of it, but it's a, it's a much cleaner—

what about, what about autonomy?

The— in the commercial space, that's not as big of a deal today. Uh, on the consumer side, I mean, we have a fair amount of autonomy. You could hit a button and the boat will just hold its position in the water. Sure.

Will allow you to wait for a plane. Without an anchor, you're saying?

Yeah, yeah, anchorless anchoring.

That's cool. That's awesome.

So if you're out in Lake Tahoe, it's pretty deep there. You can't drop an anchor, but you can just hit a button and say, I'm gonna jump into the water because my boat stays still.

And the boat just stays there. That's amazing.

That's really cool.

I mean, we, as kind of a little side project on a weekend, took the boat out and made it remote controllable. So you could just kind of remote control a $300,000 Arc Sport. So if you want to, I mean, I don't think this is legal in maybe any state but Texas, but you can like hop on.

That's where Travis lives. I think you gotta give him a call.

Yeah, you get one of our boats and just remote control it.

What about the elephant in the room? Jet skis. When are we going to see jet skis?

Where are we going with this?

I don't—

jet skis are a tough problem. It's like, it's kind of like motorcycles. There's a reason that they haven't gone mainstream electric yet. I think we might see that one day, but the cost— I mean, they're low margin vehicles. And so getting the cost down to be competitive is tough.

What about— you're saying so it's hard and low margins, you don't care about me having fun? Love of the game.

What about the charging infrastructure at different marinas? You want a jet ski really badly?

I mean, the SaaS—

Well, you just want this 'cause it'll tow you into the big waves, right?

I just, jet skiing is euphoric.

Yeah, if you're at Nazaré, you need an electric jet ski that's autonomous to tow you into the 80-foot waves.

But the environmental, you know, it's hard with a jet ski, you're thinking about, you know, The dolphins are not having a good time. Oh, they don't like them? You need to do it for the dolphins. Okay, okay.

For the love of the dolphins.

For the dolphins. That's our motto at Arc. For the dolphins.

Mark is a huge dolphin fan. Talk to me about the charging infrastructure at marinas. Is this something that you partner with docks to get up to speed? Is there infrastructure that's pretty easy to just plug into? How has that been?

The way I describe this is, Electric boats make way more sense than electric cars. That's not to say electric cars don't make sense. They just have hard problems to solve. You have to solve for road trips. Gas cars are really good at what they do. Like, going electric on the water is a no-brainer. Sure. Charging infrastructure is a good example of this. Marinas are already wired for power because big boats use shore power to power refrigerators or, um, charge up batteries because you're gonna have batteries for a long time.

So you need a lot of charging infrastructure.

So, and if you have a dock, then your dock is usually wired for power for lighting, or if you have a boat lift, that boat lift already runs 240.

Oh yeah, that makes sense.

Charging is a pretty tractable problem. It's not perfectly solved, but it's— we're out to a huge head start there. When you get into commercial and defense applications, we just designed the vessels so that they don't need charging. We're installing charging, but you don't need it. So Everything is hybrid electric. It's diesel electric or hydrogen electric or nuclear electric. Like aircraft carriers today are nuclear electric, for example. They have an electric propulsion system, but then where the power source is coming from is nuclear energy.

Interesting. Yeah, it's fascinating. Well, thank you so much for taking the time to come chat with us. Congrats on the round and see how—

We got to get out on the water. We got to do it.

There's nothing like experience. This boat, come demo it, sign up for a demo. I mean, to anyone that's watching this too, come out for a demo. It will shift your perspective on the industry.

Amazing. Amazing.

Great to have you on the show, man.

Super fun. And we'll talk to you soon. All right. Cheers. Our next guest is here in person. We have Bucky Moore from Lightspeed. Let me first tell you about Shopify. Shopify is the commerce platform that grows with your business, lets you sell in seconds online, in-store, on mobile, on social, on marketplaces, and now with AI agents. And without further ado, Bucky Moore. How you doing?

It's great to be in the temple today, guys.

First time in person.

Yeah, I think the last time we did this, I think we were both kind of just getting started on our journeys, right? It was early last year. And so you guys have just ascended. More sponsors, bigger studio, better hair.

I didn't know you were jacked.

Yeah, it's a great day today. We had leg day today. Beautiful day.

In LA, what's your split? Leg day once a week or twice a week?

Monday, Friday.

Monday, Friday. Yeah. Yeah. You guys, serious business.

We maybe need to adopt that because we don't work 5 days a week in the gym with the team, but we do like we've been doing a Technology Brothers split, which is a more traditional upper body. So it's like, you know, leg day would be Wednesday, but then you don't get the second leg day.

It's like, oh, Saturday. Oh, I missed it. Recovery though. How has Lightspeed been?

It has been fantastic. I mean, I think we're all just lucky to be in the center of the industry right now, given all the amazing stuff that's happening. I think since we last spoke, we closed $9 billion in new funds and we have been—

Don't hit the gong for us. All right.

Smash that thing. Congratulations. There we go. Very good. So $9 billion in funding, and that's across venture and growth?

That is correct, yeah. So Lightspeed is a generalist firm, and I think something that's, I think that's very unique about us is we have produced like really meaningful returns in just about every sector from healthcare and life sciences to enterprise to consumer to hard tech. And financial services. It actually is though. And I think it took me a while to realize that it's the secret, I think, of the platform is that we were willing to take deep, deep technical risk and very high conviction bets early. And I think that's the secret.

Yeah. So what have you been focused on? More on the venture side?

Absolutely. So I've spent all my time kind of in and around AI on the early stage side. Obviously, we've been spending a lot of time kind of looking for these like really, really transformative opportunities in the infrastructure layer. So new inference platforms, new model providers, and then in the application layer, I mean, I just feel like I'm seeing the future every day. There are companies that are just bringing this intelligence to bear in unique ways with these industries that I think is really gonna be the tip of the spear in how the world adopts this technology. So when we talk about the SaaS apocalypse, I think a lot of that is somewhat overblown. Like I would agree with what Pat and Carl said yesterday in the sense that people do business with people. And ultimately, I think it is so much harder than a lot of these blog posts make it out to be to actually get this technology into these industries, into these end users' hands in a way that actually makes it actionable. And so we've been very actively investing in both layers. And I think there are places where the model providers are gonna do a good job of that, but they are still gonna have to shoulder responsibility of that last mile. And then I think there's gonna be places where app providers are gonna be really prolific and build very valuable companies.

Yeah, let's work through the supply chain a little bit or the stack 'cause AI means so many things right now. Starting with like, it feels like the ship has sailed a little bit on like training new models for the most part. The big labs are really big now, huge, maybe going IPO this year. Is that true? Neo Labs are popular.

What? I mean, there's enough. There are a lot of good products. We get a new Neo Lab at least.

Is that interesting to you?

So I would say as a firm, we have been very active in investing in a lot of these Neo Labs. So we're an investor in Mistral, SSI, Thinking Machines. I would also concede that I think the bar to reach the frontier today is higher than it's ever been. And so I think for us, it's really just a question of whether—

Yeah, we've now seen like you can have unlimited effectively unlimited resources and not get to the frontier. And so as an investor, you have to see that the team actually has some— capital is not enough.

I think there's an argument to be made that there are just no secrets in this industry anymore, and scale is all you have. And I think there are really 3, maybe 4 labs that are able to really play that game.

I'm more asking, yeah, you've made bets on Neo Labs. Those make sense. Great teams. Are you expecting to make a whole bunch more NeoLab bets that are like truly, we're not on top of any open source model, we're gonna do the full thing in 2026, 2027, or has that ship sailed more or less?

Yeah, I wanna be careful with what I say here, obviously to remind myself that this is all about exceptions, but I think what we're seeing is the bar is way, way higher. Yeah, yeah, that's true. And so everyone talking about sample efficiency, everyone talking about continual learning, I mean, these are things that the large labs are very, very focused on. On as well. But I also think that there's a pretty valid argument to be made that the bureaucracy of these large labs is real. Like in the same way that OpenAI was able to exploit Google's bureaucracy by scaling a transformer before they could, I think you can argue that such opportunities will exist in the future for, you know, for OpenAI and Anthropic being distracted by making the products that earn them all this money better.

Sure, sure, yeah, that makes sense. What about like much deeper in the supply chain? There was this article in The Information a few days ago about a company called Giga that has raised no venture funding. It's a small team that got into like building electrical infrastructure and it's in the AI narrative. They were initially selling transformer equipment, like physical transformers, not the transformer architecture to Bitcoin miners. Then they pivoted into the AI data center boom. They are seeing a lot of success. Would you ever dip down that deep or is that sort of out of your purview? Obviously exceptions to everything, but— Oh, absolutely.

I mean, I think my— a few of my partners have been spending a lot of time on just the power problem. Yeah. And I think from my perspective, as someone who tends to invest more in the software layer, what I'm seeing is that a lot of these neo clouds are actually blocked by their ability to raise the credit they need to actually bring this power online by the quality of the counterparties. And so if your counterparty is OpenAI, for example, the debt providers are going to be very, very aggressive and open to flowing you the funds that you need to bring all this online. But on the other hand, these neolabs who aren't generating revenue, may not have a near-term path to revenue, they're looked at very differently as counterparties. And so that's, I think, a big blocker for these neoclouds to bring a lot of the capacity that they have online. And so how do you unblock that? Obviously, we need new and more efficient means to generate electricity. And so we've been very actively investing in nuclear. Oh, really? We were investors in a company called Base Power that's doing some very interesting work, right? Go Zach. And I think we will continue to be very active in that area. And my My partners on the hard tech side are very busy trying to figure that out.

Then on the other side of the stack, the application layer, what advice do you give to founders? What are the green flags for, okay, you're building in the application layer, you're never gonna compete in training a foundation model, maybe you do some fine tuning, but there's other secrets in the business, there's other sources of moats and strengths that you will see compounding and you don't think you're going to get steamrolled by an OpenAI or an Anthropic anytime soon.

Yeah, what's, what's interesting right in this moment, I mean, this week it's been top of mind or the last couple of weeks is like, I expect that the dynamic we're seeing right now between Cursor and the labs is what we'll see for pretty much every other vertical, right? I expect to see that in legal, like, you know, Claude has come out with like plugins for legal, but you can imagine them just saying like, yeah, we now have 100 people working on just the legal application of our technology. And so I think that every across— that doesn't mean, you know, doesn't mean these companies can't still do well. But I'm trying to think through like the timing where OpenAI and Anthropic just like basically go all in on some of these bigger categories.

Yeah. So I'm glad you brought up Cursor and like the software engineering space because it's obviously the most mature. So I I think it's very illustrative, as is legal. And interestingly, today, you guys— I'm not sure if you guys covered this in the beginning of the show, but like the model that Cursor launched, right? So first off, you know, it's very obvious now that these companies are going to be training their own models. And it's also the case that they're going to be starting with one of these open weight models, which today are Chinese, which as an American citizen doesn't sit super well with me. We are investors in a company called Reflection AI that aims to change that. But on the other hand, I do think what Cursor is out to prove is that the user data that they're capturing is going to allow them to set up this reinforcement learning loop that does allow them to build something that is like frontier quality. At the same time, it appears as though these companies are still sending a lot of their most complex queries to Codex, to rather GPT into cloud. And there's obviously a margin problem there that is pretty hard to compete with when you think about the competitive products that OpenAI and Anthropic have. And then when you shift gears to legal, my assumption would be that like Harvey and Lagora will have to do the same thing that Cursor's doing. And so I think this is kind of an elephant in the room right now as to whether doing that, whether post-training open models allows you to, combined with like unique user feedback that you get from being an application provider, is defensible enough. And so when I think about like the companies that we're looking to invest in, I would actually say that I think that is just gonna be an inevitable challenge for any of these industries that hit a maturation point of AI adoption like legal and software engineering have. But on the other hand, I think there's some industries where they're very large, they're far enough afield from where the model providers are today and probably will continue to be, and the context engineering to actually get the customer data into the model is just so messy. It requires going across different business functions. It requires a lot of like hands-on forward deployed engineering. I think those are the kind of companies that we get really excited about because I think just being really good at that is not only defensible, but it also allows you to start to, generate this feedback loop with your customers where you hear a lot of their secrets. And those secrets allow you to feed that back into how you make your product better at the expense of anyone else playing in the space. Because if you're serving the customer, they're only telling you those secrets, right? And I think Palantir is a good example of this in sort of the pre-AI era. And I think we're going to see many companies ascend in that same way. And so the companies that we're investing in the application layer, I think, have a really clear story around how that is hard and defensible enough that they can make it a core competency.

Let me throw one at you that sort of riffs off of that, and you can tell me if it seems like a fit. Like Amazon, obviously the goal was to be the everything store, but there were certain categories—regulated medical products, Hims, Roman—those companies were able to break through because it was out of the core competency. Is that a reasonable thought where it's not just about the flywheel of what's happening with those customer secrets, but it's also just such a sticky, messy problem that the big one-size-fits-all solutions can't necessarily attack them?

I think there's sort of a B2C version of this analogy, and then there's like the B2B one. So the B2C one is where you started. What you've seen with Amazon is they've obviously like kind of done what Costco does, which is when they see a category that's sort of commodity enough, they can create a generic version of it, and they can have sort of margin advantages there that allow them to make a lot more money and compete. But I think that the B2B analogy is the one one that I think of a little bit more, which is to say, like, if you look at Amazon Web Services, they've let companies like Snowflake and Databricks build really, really large businesses on top of them because it's driving consumption of their underlying infrastructure. Oh, interesting. That's a good one. And while AWS still has data warehouse and data infrastructure products, they're probably 80% as good. So if you're in the business of just kind of simplicity and you just want to kind of buy everything from Costco or buy everything from Amazon Web Services in this case, like, you're going to go that way. And many large enterprises do. But on the other hand, Databricks and Snowflake have built really, really healthy businesses, and the health of those businesses accretes to the success of Amazon Web Services. I think you're going to see something very similar with the model providers, which is they're perfectly happy letting 75% of the large industries out there just kind of sit on top of them. And then I think the 25% of those industries that they're going to play in, those are the ones that I think you want to be very mindful of and avoid. And I think what the sort of hysteria around the SaaS apocalypse is right now is that we're starting to see the aperture of that 25% 25% like rear its head. And I think what that means is you naturally take a step back and say, wait, like it's not just gonna be software and legal. And so I think that's the conversation that we're having every day is like what's next? And I think Cloud Cowork is sort of an example of where you could also just imagine that there's sort of this new class of AI app company that essentially connects to something like Cloud Cowork via MCP and actually is in a really interesting position. So like we're seeing companies emerge where for example, in the coding paradigm, there may be places where you want to jump out of the code and jump to a design canvas, for example, right? And so there's a company, for example, that's trying to just be called Paper that we think is really interesting, that is trying to provide a canvas that literally just allows you to go from like the code you generate in Claude to the canvas and back, right? And today that's something that Figma hasn't focused on as much, but I'm sure will. Granola, which is a company in our portfolio, like they've launched an MCP server and people are using it very aggressively in Claude. And I think what you're going to start to see happening is that the monetization model of a lot of software companies is is gonna shift to actually monitoring, or sorry, metering the rate at which the MCP server is consumed. So for example, every time I call Granola's MCP, Granola should make a little bit of money, right? And I think that's actually gonna be a very healthy business model and one that's probably durable going forward.

Jordy, sorry, cut you off.

How do you think about opportunities at the super early stage where a lab could do it but wouldn't because it's only, in the near term, like a $300 million, $500 billion revenue opportunity because I think there's like lots of, you know, you look at companies like Open Evidence or Suno that have built like, you know, pretty big businesses. And like, I think like Suno seems to be a beneficiary of like the battle between Anthropic and OpenAI right now specifically 'cause OpenAI is like, well, we could put a bunch of resources into a, you know, a music product, but that's not the most important. Like if we just, even if we knock it out of the park, it's not gonna necessarily change the dynamic at the frontier. Are you thinking about, are you seeing opportunities at the early stage where you're like, yes, the labs could do this, they would be in a good position to win, but it's just not even a big enough opportunity. And so a small focus team can get to a $5, $10 billion valuation and then have the opportunity to like, scale from there?

Yeah, so I'll start with Suno. So we're also one of the largest investors in Suno, love Mikey. I was honestly very skeptical of the company as a music aficionado. Like it was just difficult for me to conceive that music could be generated by AI that I'd actually wanna listen to. And I've been proven wrong and really changed my mind on the company. And so I think the sky's the limit for them. And like what's interesting about Suno and more broadly audio models is they're just not as big as language models, meaning like you don't need as much compute to build really good audio models models. And so I think the scale advantage that the labs have is just a little bit less of a thing in that area.

Yeah, interesting. It's more taste-driven.

That's right, that's right. But I think coming back to this, like, hey, there's gonna be 75% of the industries that they sort of rule out as like just immaterial, given how big they get. And obviously they're growing so nonlinearly that like what's material for them changes by the week. And so I think there is gonna be a pocket of really, really venture-scale industries that the labs just do not service adequately. And I think that's, What we spend every day thinking about is like, what are those spaces and how do we find the best teams working on them? And so I think that's again why I sort of take the other side of this like, you know, B2B app businesses is sort of overlooked.

What's an overlooked category right now?

I mean, I wouldn't say it's underrated, but I think like power is still overlooked. It's just so important and such a bottleneck.

So big and such a bottleneck. And the fact that you can name some nuclear companies, you can name, you know, we've had BasePower on a bunch of times, but you know, you would think there would be a bunch of like BasePower-style opportunities.

I think we're probably rate limited by capital going into that industry right now. That's just my intuition. The other area that I think is like more core to where I spend my time that I still think is underrated is inference, right? So like the GPU supply crunch that we're seeing right now is largely in part as Dylan has said on this show before, like due to the fact that not only these consumer products but the B2B products like Cloud Code and Codex are just really taking off and creating like insane demand for inference. I actually think that in the end, inference, if you look at it as a market, will be much, much bigger than cloud computing was pre-ChatGPT. And if you think about like, I mean, we're talking like hundreds and hundreds of billions of spend every year. And if that's true, like I think there will be very, very large inference platforms built built in each modality. So there will be an inference platform for real-time video models. There will be an inference platform for open source and custom language models. There will be an inference platform built specifically for long-running agents. So I think we're just going to see that industry, which today looks like one industry, break up into many because of how big it is and how much room for specialization.

Yeah, I mean, there was some leaks around like Fall this week, right? Who's just been on an insane trajectory and kind of they've done a bunch of rounds back to back, but going that quickly from a zero to an $8 billion valuation while staying relatively out of the— 10 years ago, there would have been a bunch of beat reporters just covering Fall, and they are just happily being like, yeah, we're just building a huge business over here.

And you can sort of imagine inference specialization, if that's the thesis that plays out happens both on like the model level, the data level, the taste level, the harness, but also down to the hardware. Like there's certain configurations, certain chips. I mean, we see this through InferenceX from Semi Analysis, like AMD chips will outperform NVIDIA chips with certain models. So if you're in some particular inference vertical, you might have a very different semiconductor supply chain.

Yeah, and I think one interesting thread to pull on there is, is if you start to relax the latency constraint when it comes to inference. So for example, like if I'm just telling my software agent to go do work and come back to me in a couple days when it's done, I probably don't care as much about latency. I care about throughput. And if that's true, then you start to look at these chips that Intel and AMD have and they look really, really good. They're cheaper and really stand up.

And you can imagine that for like, if I'm doing like blood work or something and I have to physically mail blood and there's like no AI that can short circuit that, So I'm going to be waiting a day or two to get the data back. And the model that processes the data or sequences it or something takes an hour instead of a minute. Like that's not going to be mission critical to—

You don't need Blackwells for that type of thing.

Exactly, exactly. Whereas if I'm doing a Google search or I'm on my computer like actively doing work, throwing that on Cerebras or something or Groq is going to speed things up. And specifically in the developer workflow too.

Yeah, I think where this starts to get really interesting is like if you ask yourself like, will Will there be more of those types of workloads or less? Like, I think it's quite obviously more if we're moving to these long horizon sort of more autonomous agents. And so I think there's a world in which the sort of latency constraint that drives a lot of the hardware purchase and design decisions today starts to be relaxed and we see a much more heterogeneous ecosystem there. That makes sense.

Fundraising timelines today, faster than ever?

For venture or for startups? Early stage.

No, no, no. Yeah, I'm talking about startups. Like if you back a company at seed, how long—

I think there are 8 companies in the current YC batch that are raising at over $100 million valuation, which is crazy because Demo Day is next week. And so how do we even know that this is happening? Usually you wait until after Demo Day. Not if you have the kind of momentum. And I think the average price is around $40 million. And clearly deals are getting done very, very quickly.

Yeah. Is the global conflict that we've entered into, is that slowing things down at all yet? Is there any kind of like pullback that you're seeing?

So I'm not seeing the geopolitical impact early stage fundraising in any way, but coming back to kind of the velocity question, look, I think right now we, so first of all, what I've learned is that you don't really get to decide how much of the best companies you get to own, and you don't really get to decide the valuation. They decide it, and you have to decide whether you want to be involved with them or not, right? And right now, I think early stage is in this tricky place where, like, for a seed round, like, you're really generally getting around 10% ownership, right, for the most competitive rounds. And as a result, if you go back to kind of like what made the early stage model work, it was like owning somewhere between like 15% and 20% of companies that become really, really valuable. So the way you kind of have to play play this now as a multi-stage firm is you get the 10% in the seed, and then you effectively have to be in a position to buy up in the next round, the Series A. But at the same time, the Series As are also ballooning in price, and there's really not a lot of de-risking because the valuation— or because the velocity is so fast, right, by comparison to even like a couple of years ago. So right now what you see a lot is like you get 10% in the seed, the Series A happens somewhere between like $150 million and in some cases like $300 million post, and then you end up kind of trying to I'll co-lead that or lead that, and you end up in like the low to mid-teens. And so it's just a different time, but like that's what it takes nowadays. And there are obviously exceptions to that, and I'll gladly buy more than 10% of a company at seed if I'm excited about it and the founders are open to it. But that is the reality that we live in, and I think what that means is you're just, you're building a basket of companies that tend to be kind of in that low to mid-teens ownership more often than not.

Are ghost ship zombie acquisitions affecting venture economics? Economics at all? Or has it been—

So these are like legacy software companies that are slow growers and not going to—

No, more like licensing deals.

I'm thinking of super fast growth startup with a bunch of venture capitalists that then a bigger company comes along and says, we want to buy you, but we're public. And so we can't just acquire the whole company and go through FTC approval. So instead, we're going to hire the whole team, do a licensing deal, pay a dividend to the cap table, But there's always this moment where you're like, did everyone get comped what they expected? Did the employees get what they deserve? Did the venture investors get what they deserve? Did the founders get what they deserve? And it's sort of this renegotiation moment. And there was a fear, at least last summer, that like the Silicon Valley social contract might be violated. And it felt like at the moment maybe that was like a fear that didn't come true, because in every case most people seemed like they wound up with jobs at cool companies, at bigger companies, and they did get paid out. But has the fear calmed down, or is there still a low-level ambient fear in venture?

So first of all, I would say that the idea that an acquirer will try to structure deals in a way that disproportionately rewards the founders is nothing new. That's been happening. That's a tale as old as time. And there is sort of this term used in the industry called the founder bribe, where to get a deal done, the acquirer may in fact try to route as much of the proceeds toward the founders and the core team as possible. It's also perfectly rational because like, why would they wanna give me any money as the investor, right? They'd much rather give money to the people that are actually doing the real work.

Yeah, your job is done at that point.

And so it also depends on, you know, is it an asset sale or a talent acquisition or somewhere in between? And everything's sort of a, you know, a blurry line.

Yeah, and so with respect to these like creative structures that you're referring to, right? Like the Scale AI, Windsurfs, and so on. Like, without commenting on specifics, what I would say is this is obviously an artifact of trying to get around a lot of the regulatory environment that makes buying companies difficult, for one. Two, the bankers out there today are just pitching these left and right because they're out there trying to transact, and they think this is a much more realistic way to transact. I didn't realize that. And three, I think the jury is still out on whether the acquirer really gets, like the return on these that they're looking for, right? Like a lot of these are still in flight. Totally. And I just don't know. Yeah. So, but what I can tell you is that like, it's becoming more common. I told you why it's becoming more common. Yeah. And the bankers are very, very actively trying to consummate these types of transactions every day.

How introspective should a new founder be as they embark on their entrepreneurial journey? Yeah. Because there's two sides to it. Like, I totally get the Travis Kalanick was sitting here telling us that he just runs through walls. Everyone was super pumped up about that. I love that attitude. I think that chopping wood, putting your head down, incredibly valuable. At the same time, I love the Founders Fund question, the Peter Thiel question. What do you believe that very few people agree with you on? That's introspection. Like, you're actually just reflecting on like, where do I disagree with everyone, maybe there's opportunity there, and maybe there's a combination, but I'm wondering how you deal with it. Like, how much of the time are you just probing founders' market size, progress, revenue, versus where do you want to be in a decade? What are you doing with your life? Is this actually the right job, or do you really just want to get back to working at Google and this is just a way to get a promotion?

So my experience has been it is very hard to be an effective leader without some form of self-awareness, which requires introspection. Let's just get that out of the way. At the same time, I think this position that Mark is taking or the retard maxing article that has been going around the internet is— there is absolutely merit to just moving forward and putting one foot in front of the other. And so I think it's healthy.

Avoiding analysis paralysis. Exactly.

And so I think it requires a healthy balance of the two. But I would disagree with Mark that I have not found it's it's impossible to be an effective leader without some form of self-awareness. But at the same time, in order to inspire people, you also have to demonstrate an ability to move forward through difficult moments. And I think that's where the— I won't say it again.

And I think that's actually closer to his real belief. But in the context of that conversation with David Senra, it's better to distill the overall thesis, which is that right now, most people would benefit from just having a little bit more forward forward motion, so let's collapse that into somewhat more of an exaggerated take that then can fly and it gets exaggerated in many ways.

Yeah, exactly. What advice, if any, have you given portfolio founders that may be on the earlier stage side that seem like they're gonna need to raise over $100 million in the next 12 to 24 months? How concerned should they be around everything happening in geopolitics right now? Does the conflict in the Middle East kind of like turn off the tap or slow down any LP dynamics for some of these mega funds?

Should startups keep their corporate treasury in oil right now? Probably not.

Well, probably not. You don't want volatility. I think the, you know, what's happening in the Middle East right now is a tragedy. And it goes without saying that many of the countries that are most adversely affected by that have been very, very active players as LPs. I think the impact that this could have on the way they behave in that sense is like very unclear, but like, it seems likely that there will be some impact.

And so that has a consideration on venture capital. Yeah, my example was like, if somebody's, doing a lot of angel investing, and they're working at Google making $3 million a year in a cushy job, and then they lose their job, do they keep ripping a bunch of checks?

MAXIOUS was talking about this. And he was saying, well, the LPs that write checks from Gulf sovereign funds, they live in New York, and they operate in New York, and they take meetings in New York. But there's just a matter of that sovereign wealth fund is zero-sum, and if it's is going towards something else, it can't go into a venture capital firm and work its way through the—

And ultimately, the royals get to decide.

Exactly, the priorities.

And there's a whole bunch of other political considerations. Priorities could be shifting right now. Totally. So I do not know what's going to happen. I think those are the considerations. And I think when we advise the companies that we're working with that are in a position where they actually can raise that kind of money, we have not yet seen the market indicate that there isn't robust interest for companies that are really scaling rapidly and have earned the right to go out and raise money that way.

Yeah, it seems like we're pre, like, some sort of correction where we're like, oh yeah, like, you can't raise the same round. It's just you might raise $80 instead of $100, and your competitor's in the same boat, so you're probably gonna operate very similarly with just a little more.

Yeah, I would say that I do think the bar for these new for the neo labs to raise that second round of funding is much higher coming back to what we were talking about. And so I think to be able to get one of these off the ground, it's possible to do so as someone who has been a part of doing really novel research at one of the big labs and is someone who can probably build a really outstanding research team. On the other hand, I think to raise that second round of funding where you get to go really big and build out that frontier cluster, I do think it's gonna require some mix of, hey, the work we're doing unlocking new commercial opportunities, and the work we're doing is truly, truly novel at the level of the research. And I think absent those things, what I've seen is that it's a lot harder to pull those second financing rounds together. And I think this is kind of a dead zone for these neo labs where if they can't go big with compute, it's really, really hard for them to be on the frontier. Yeah. Did you have a conventional path to venture? I would say I had a pretty unconventional path. So I grew up down here. Started my career in lower middle market private equity. I moved up to the Bay Area to work at Corp Dev at Cisco, and then I think that's when things started to get a little bit more— yeah, proud sponsor, right? We love Cisco. And it turns out that the founding leader of the Corp Dev team at Cisco was actually one of the co-founders of Lightspeed, Barry Eggers. No way. And that team has definitely produced a number of VCs that remain in the industry today, and so that's kind of where I sort of got more conventional, but I think where I started was fairly unconventional.

So Corp Dev at Cisco, were you working on a lot of M&A deals? What was the M&A environment like back then?

Yeah, so when I was there in 2011, this was when we were sort of litigating whether there would be cloud computing or on-premise computing and whether cloud computing was something that would be more than like a test and dev environment for the average company. Sure. That definitely taught me that it's very, very easy to underestimate the future when you're one of the incumbents. And the rest is history, obviously, with respect to cloud. So at that time, we were trying to execute on a strategy that would put Cisco in a position to own the right pieces where they could both serve enterprises building their own clouds and the cloud providers themselves. So we were buying storage companies. We were buying new next-gen networking companies. In fact, one of the projects I worked on was trying to acquire, unsuccessfully, that is, Martín Casado's company, Nicira Networks, which ended up going to VMware and really did transform the industry. So shout out to Martine on that. That's crazy. And so, yeah, we were working on a lot of deals like that. We were working on venture investments as well. So sometimes the companies maybe, or an industry, is not fit to enter via M&A. So you would invest in the right companies and sort of learn, right? Sometimes you would partner. So yeah, that's the kind of stuff that we were working on. It was an amazing job, amazing way to get into Silicon Valley. Totally. It helped me gain deep appreciation for how much impact these infrastructure companies can make across every industry at a time where everyone was kind of focused on what the next Uber or Snap was on your phone.

Yeah, yeah, yeah, yeah. It actually shows you like the breadth of the industry because every day you're looking at some company that is not a household name, is not getting a puff piece in whatever legacy journal is out there, but their business is fantastic, they have an incredible team, and there's a path to just absolutely massive growth.

Also being able to level check with founders other founders that might be exploring M&A and just being like, well, I've actually bought a bunch of companies. And like, you're actually helping them qualify themselves and say, are you ready for this?

And I carry that with me to this day. And I wouldn't have that experience had it not been for the time I spent over there.

Well, Bucky, thank you so much for taking the time to come chat with us. Have a good rest of your show. Have a good rest of the day.

I wish we could hang around.

Tell everyone about wewillconsole.com. Console helps AI agents build— Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. And I'm also going to tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. And without further ado, we have Steve Huffman, the co-founder and CEO of Reddit. Steve, how are you doing? What's going on? I'm doing great. Thank you so much. How are you? For taking the time to come chat with us. You gave a fantastic speech at my YC batch back in 2018, and I really enjoyed that. And yeah, it's just great to catch up with you. I would love for you to sort of just set the table for us for how 2026 is going over at Reddit. What are the biggest initiatives that you're working on? How are you spending your day? Just give us, get us up to speed.

Okay, 2026 is going, going fast. Yeah, actually, man, I can't— it's like I'm looking at the date now, it's March 20th. It's insane. So right now we have a, I think, single-minded focus on making Reddit more successful for new users. I think that's the part when I think about the Reddit machine and the Reddit business, that's the part that I think has the most potential or the biggest gap to close. Yeah. So like taking a step back, right? Reddit's got content communities for literally everyone. And I think we've proven that Reddit can work for anyone, you know, men, women, young, old, nerds, normies. And so now our task is can we make it work for literally everyone? Yeah. And making Reddit successful for those new users in that first session. That's the thing that I spend almost all my time thinking about right now.

What, what, what are the qualities? Like, what is the, what is the experience that someone has when they join Reddit that gets them to stick? Like, what kind of journey are you trying to take them on? Yeah, that's a great question.

Yeah, the classic like Facebook one was like, if they add 5 friends, they'll stick around forever or something like that, right?

Yeah. And I'm, I'm not even sure that was true for Facebook, although it's like a tale that's told in Silicon Valley. But I mean, there's probably some truth to it. And the Reddit equivalent of that would be you find 5 communities on Reddit that you love. Yeah. But do you need 5?

Because like, you know, growing up, yeah, just because for me it was like hip-hop heads in high school, like if I typed in like R in my browser, it would immediately just like, you know, like populate r/hiphopheads.

And I think that's actually the answer.

You need to find one of those things that is uniquely Reddit, touches you, and provides that experience that you can't get anywhere else online. And then once you do, you find that, you know, then we've got you.

Then we've got you. Yeah, so I remember the first time people were starting to use the TikTok algorithm, like the feedback was like, it's creepily good basically within like 5 swipes. If you stopped and paused over a car video or a watch video for just 1 millisecond longer, your entire feed would be cars or watches or whatever you were into. And I'm wondering of that customer journey, that you're trying to funnel people towards the right content that they'll love, is it more of a user interface design problem or UX design problem, or is it actually just a compute problem and you wanna apply more machine learning, more modern AI tools to bring that content faster, learn what the user's interacting with or observing, and then help them along?

I actually think it's both. We've seen this over the years. One of the things we can do most consistently to grow is make the product easier. So that would be like the UI. You make it a little bit easier, a little bit faster, easier on the eyes, maybe less chaotic or overwhelming. And we just grow because we have people trying out Reddit every day. And it's just for some of them, it just doesn't work. But also, right, the machine learning, the feeds, those early recommendations and insights, have almost unlimited upside as well. But actually, Reddit is probably in that era of ML. Like once you're a core user and you're in your feed, if you express an interest, yeah, we'll go really heavy on it. Yeah, almost too heavy sometimes. Yeah. And it's a problem because I don't know if this is like a normal Reddit user experience, certainly mine, like people send me weird shit all the time. And then, you know, if I view it and Reddit, I'm like, oh, okay.

Oh no, my whole feed's gonna be this thing now. That's my life. Yeah, yeah, you need a little bit of— So there's a lot of finesse required there. Yeah. What has the shift to mobile been like? Like, what is the story that you tell about making Reddit more mobile-focused? Obviously there's a huge shift in user behavior across all internet platforms, but what was your, individual experience?

Well, so if we look at Reddit's story, like arc, yeah, the shift to mobile for Reddit was slower because Reddit, like, we have so much text. Yeah. And so that was like a key— we were like a keyboard and screen platform for longer than most. Yeah. Now, like, desktop, uh, desktop is by far our smallest platform. And so we don't really think about the shift to mobile anymore. Like the shift happened. Yeah. Maybe for Reddit, it happened despite ourselves in some ways, but it's happened. And so now it's more of the shift to the app. I think our, one of our largest platforms, I think our largest single platform is actually probably mobile web. And then you've got iOS and Android. And so for us, we get so much mobile web traffic because of the search referrers from like Google. So people are, if you're just using the internet on your phone, you're probably using Google. If you're probably using Google, you're probably using Reddit. But so getting, helping people kind of cross that chasm between mobile web into the app and doing so in a way that feels good, right? Where we're not like coercing a user to download the app, right? Which is annoying. You know, finding that balance and doing that in the right way. I think that, again, there's a lot of, there's many ways to do it. And I think there's a good way somewhere in the middle with a lot of finesse. But that's actually where we spend a lot of time thinking. Yeah.

How does the average Reddit user feel about AI?

Hmm. Well, I think the average Reddit user is basically the average person. Yeah. And I think AI right now is polling less than politicians. Yeah. And last I checked, nobody likes politicians. So, but at the same time—

But the difference there that I think is funny is like, people like, you know, may have a negative experience of politicians, but then, but with AI, like people like say they don't like it, But then they have all these magical experiences in their life, and they use it like crazy, and it makes their life better in even a bunch of different small ways. That's exactly— In ways that they're aware of, in ways that they aren't aware of, right? Like it's happening under the hood. And so I think at some point it will actually flip, right? Just because it's a very valuable magical tool that anybody can benefit from.

We're going through that transition. Yeah, exactly. AI. Do you like AI? People overwhelmingly negative. Do you use AI every day? Every— it's like everybody. Yeah. And you can see this kind of— you can see this relationship with Reddit as well. AI makes not just Reddit, but the whole internet safer, right? Things like violence, harassment, bullying, we can use LLMs now to find and to just filter out, to moderate out at platform-wide scale in seconds. It's incredibly powerful, right? The worst job in the internet used to be looking at the worst content on the internet. People don't have to do that anymore. So that's really powerful. It also happens kind of behind the scenes. But then AI writing, like on Reddit and elsewhere, it's just kind of annoying, right? And we're going through this, you see this on Reddit, It's not prohibited behavior. Like a human being with a human account uses ChatGPT to make a post. Yeah. Like that's allowed. But then you see all the comments just flaming them, right? Thank you, bot. Thank you, bot is like the new okay, boomer. And then of course, there's just like the maybe spam and other bad behavior, which is definitely not allowed. So, but I think like on Reddit, in emails, in school, we're going through this societal calibration about where AI is helpful and not helpful.

Yeah. How did you process Maltbook?

Oh, yeah. Oh, I think I clocked that one within 5 minutes. I said, this will be a meme for about a week, and the world will move on. And it was. It was like a total insider AI meme, like fun thing. You know, if the accusation is on the internet, the internet's dead, everything's turning into bots, it's bots pretending to be humans. On Maltbook, it was a bunch of humans pretending to be bots. And so it was like, to me, it was like AI fan fiction. Yeah, it was like so, so goofy, funny, but goofy and otherwise insignificant.

Yeah. What has been What is— walk us through the history of bots on Reddit, how you've dealt with the problem, when they've been bugs, when they've been features. Features. Like, that's a big— I mean, even on X right now, it very much feels like a bug. There's certain accounts, if you look in the comments, you can just tell every single comment was written by a bot. It doesn't add anything to to the conversation at all.

And yet 5 years ago, everyone was tagging those bots, remind me or bookmark this or other cool things.

Yeah, I would say that like Grok, like the ask Grok, is like 50% of the time it's annoying, 50% of the time it like adds relevant context that improves the experience.

Yeah, yeah, yeah. So I think the solution to this issue was and always has been through transparency and intentionality. We've always had on Reddit helpful bots, right? The remind me bots and the haiku bot and maybe less helpful but accepted, like the grammar correcting bot.

Sure.

And then we've had bad bots, but bad bots we just call spammers. Yeah. Right there. They're submitting spam or they're otherwise manipulating Reddit. That's always been forbidden and we do our best to remove those. I mean, we do it at an absolutely massive scale. It's like 100,000 to 1 million accounts a day that we ban. And then with AI, it's not a new problem. There's just new technology. And so again, if you're manipulating Reddit, if you're spamming, if you're pretending to be a human but you're not, that's not allowed. You get banned. But we also want to leave room for helpful bots or even agents. But I think they need to be labeled as such. And then there's that gray area, which is what we were talking about before, which is a human being using AI to write. That's not forbidden, but it is kind of annoying. But we'll let that one play out. And right now, what we're seeing is people downvote it, they complain about it, communities maybe ban it.

Yeah, but at the same time, for every post there, when a human with AI was writing something, posting it, and it's getting negative feedback, there has to be like a number of them that used AI that people don't even know they used AI and it is just being additive, right? Like it's not a—

It's possible. It's possible. And in which case, it's probably okay. I think the question is, is there a human behind the prompt? Is it a human's idea? Like I'll give you an example of AI content that's totally acceptable. Translated content. Like, I'm a non-native speaker, like, help me, help me write better. Like, that's just the world we live in now. Yeah. And so for us, like, Reddit is for humans, like, that is our platform, that is our product. It's human connection and community. So we're going to start actually talking about this more. So this is a really apt time to have this conversation. This idea of humanness and human verification and what I call ass in seat. I guess they're actually human using Reddit right now, regardless of the tooling that you're using. Yeah.

How do you think about delivering on that?

I think the most lightweight— I think there's various technologies. The most lightweight way is with something like Face ID. Face ID or Touch ID, broadly, it's in the family of technology called passkeys, which I actually didn't appreciate about these like a year ago, they actually require human presence, like a human has to touch or do or look at something. And so that actually just proves that there's like a person there or gets you pretty far. And so I think that's very lightweight and accepted. I think there's heavier versions like, you know, the ID checking services, which we have to use, you know, for regulations here or there. And I think there's in-between technologies that the internet really needs, like third-party, individual, no ID required, decentralized, like, I don't want to say identity, but like information providers that there's a need for on the internet because not just Reddit, every platform wants to know, is this a person? Now Reddit's version is, is this a person, but we don't want to know which person this is. Right, this part of our promise to our users is we don't know your name, but we do want to know that you are a person. And so it'll be an evolution for us, I think, for a while, and probably every platform, I think, to find the right kind of middle ground here.

Seed stage or Series A startup wants to unlock revenue on Reddit, what advice do you give them outside of giving you ad dollars?

Build a great product. Build a great product. I mean, that's my advice for every startup. Yeah, like build something.

They shouldn't— they shouldn't do a hostile takeover of a subreddit. Look, because I know you guys— I mean, I know that's like an ongoing issue. Look, I think—

I think that is a very difficult thing to pull off. You know, there's always that startup that they build something really great and the founders are super authentic and they can show up on Reddit and say, hey, we built this, what do you think? And everybody loves it. And then the other 99% of companies get like chased away for self-promotion. Yeah. Or they're manipulating Reddit in some way. Yeah. And you know, we end up banning them. So it starts with building a great product. And then from there, you got a lot of options on Reddit. Yeah, of course, you can be an advertiser. But you can also talk to your customers, you can provide customer service, you can do AMAs, there's all sorts of things you can do. But I think the only ones who are successful are ones who, I think, have an earnest desire to contribute to a community and not take advantage of it, because people are very sensitive to being taken advantage of.

Yeah. What are the unique dynamics on Reddit when it comes to creating ad products? You know, Meta, you know, and Instagram seems to be the ultimate state that a user can be in, just like pure consumption. They get shown cool things and they're kind of whatever state of mind they're in, there's an openness to like, you know, immediately just buying something. I would say X is maybe the opposite of that. I think, did Ben Thompson describe it as like you go on Reddit to like get a knife fight? Yeah, you go on X to get in like a knife, you know, you're like, you're there to like—

When you're on Instagram, you're shopping, you're shopping for everything.

I mean, Instagram's products would be better if it was only ads. Yeah, I know.

So, well, in some way it is. I mean, for me, like my experience forepage is just cars and watches.

Yeah, cars, watches, clothes, houses.

Not directly, they're not directly ads, but they're like functionally, I'm like being marketed to, but it's just UGC.

Yeah.

So, so on Reddit, you know, what's interesting about Reddit, and so people talk about their interests and hobbies and passions. What I, what I didn't see coming when we started Reddit is that it turns out within people's interests, hobbies, and passions, it's a lot of commercial conversation. Yeah, they're basically asking the question of what should I buy? Yeah, but they don't say what should I buy, they say what should I wear, what should I watch, what's the best gear, like where should I go? And so it turns out that Reddit, which has this kind of anti-commercial vibe, is actually extremely— it's a hub for commercial activity because that's what people talk about. Yeah, totally. So that's why I say for brands, like You can stick the landing. Yeah, like it's there. The conversation is happening. But again, transparency and intentionality, that's what, that's what really matters. And so yeah, ads is the easiest way to do it. But also you can be helpful to your customers, which are definitely on Reddit talking about your product. Yeah.

Can you help me understand the shape of the advertising business? Is it, is it the entire revenue stream? I know that there's 'Cause there's deals with other companies and there's other pieces of the business throughout history. And then how power law driven is it? A lot of ad platforms have like a super long tail of small advertisers. We've talked to other folks with ad platforms where they're selling really big ticket packages to Fortune 500 brands and it's all shaking hands and kissing babies. They're basically a politician on the road all day.

The story arc of any ad platform is basically the same. You start with big companies and brand advertising, because those are the companies that will test new platforms. And then you simultaneously move down market into mid and small businesses and more performance-oriented. What's great about Reddit is we do all of this. So our fastest growing segments were mid and small business last year. We have a nice balance between brand and performance. And so we're one of the few platforms that whatever your objective is, and whatever the size of your company is, Reddit can work for you. Now, almost all of our work in the ads platform, it comes down to, you know, better measurement and better targeting, right? It's really just a game of inches, just constantly grinding that out. And making sure that our ads actually work, actually deliver a return. But Reddit, it's so funny to me because when we started Reddit, I didn't like ads. Now, this was 20 years ago. So, you know, the pressures of the real world have made an honest man out of me. But so it's almost surprising to me how good Reddit is for this. And so it's a never-ending journey. But to answer your question before, About 95% of our revenue is ads. It's actually a great business model because it just scales.

Yeah, it makes perfect sense. If an investor came to you looking for alpha, what subreddit do you think there's a $10, $100 billion business brewing, business opportunity brewing in? I think like Chris Dixon has talked a lot about, like, you'd find a subreddit that was, you know, a hive of activity.

Yeah, this idea of what people on the weekend for free we'll all be doing during the week in a few years? Oh, well, that's an interesting question.

There's so— we're in this like funny state that like we're in this like vibe coding takeoff. Yeah. So I, I look at kind of the arc of, of my career over the last 20 years, and I was lucky to catch a couple of waves. The first wave was just the internet. And I think of each of these waves as a lowering of the bar of what it takes to start a business. So first is the internet. Then that was the access to capital. So like Y Combinator and the more democratization of investing. And then you got mobile. And then you got cloud. So now, like, to start a successful company, you can build an app and host it in the cloud. But the equivalent version of that 20 years ago or 30 years ago was you had to know enough people to raise $10 million to even get your foot in the door.

You had to go to the computer store.

Yeah, but you still had to be technical. And now we're in this era with, with vibe coding. Vibe coding is such a small word, I think, for such a big concept, which is like, basically, if you have an idea you can basically bring it to life. And that's becoming more and more true every day. And so one of the things I've been thinking a lot about is, like, the headlines are all like such and such company lets go of all of their engineers, like engineers are being replaced by AI. I think that's missing the next step. Like these engineers are builders. What do you think they're going to lose their jobs and just go home and sit on their hands? No, now they will probably start companies of their own. Yeah. And, and so I think there's this transition, I think it's made very, very fun where everybody's a builder. And there's likely to be this like Cambrian explosion of creativity and creation.

Yeah, but what I think about in the context of Reddit, there's like all these subreddits that might only have like, 5,000 active members, and historically they maybe had an idea for like a little vertical software product or a mobile app. And like, it would never make sense for a team of 5 brilliant engineers to like raise millions of dollars and then go and build that product because the TAM is like the 5,000 people. But then you actually can do that now because it's just one person that's like, hey, if I can even get it like a few hundred of the people in this subreddit to pay me to use this product, I can have a real business and I can create a life. So it's like—

It looks more like niche podcasting or having a Patreon.

It's like, oh yeah. Same exact thing is what's happened in media, right?

1,000 people paying you $5 a month?

Yeah, this show wouldn't have been able to exist with cable because you wouldn't— No product. Yeah, it's too niche. And so, yeah, I think every subreddit, there is a software product to be built, and it's not necessarily like a venture scale opportunity, but you just do the math. I'm like, can I get 1,000 people that pay me $20 a month? And like, great, that replaces the salary that that person may have gotten.

Yeah, it's the whole story of the internet, which is the transition from centralized traditional power to decentralized. And so it's this really enabling and uplifting force. Yeah, podcasts are such a great example of the decentralization of media. And so now I think with AI, there's just going to be this whole new era of creation, which I think is the optimistic side of the story I'd love to hear more of. Because I think the pessimistic side is the transition and the change. But I don't think the outcome is all bad. I think it actually could be quite empowering. I completely agree.

What are the— give us a guide to watches on Reddit. What are the best sub-communities?

You're talking about watch and bezel?

I don't know if I can, like, with a clear conscience recommend anybody get into watches on Reddit. It's like— I've fallen into that for better or for worse. So you can start with watches. Okay. And then maybe you should start with like watch— I forgot it's called like watch hot takes or something, but just to see like people who are coming out on the other side of this. Okay. But yeah, I mean, in ruins, in shambles.

It's just— gosh, it's a— it's a— it's a pit.

I don't know where to describe it. Like valueless consumerism. Like just It's like telling somebody, it's like, what's the best subreddit about smoking? It's like, I don't know.

That's great. Is, is Vacheron underrated by the tech community? For sure.

That's their whole strategy is to be underrated. Yeah, I think like high quality and, and unhyped. I think, I think their CEO just had an interview where he basically said that's our strategy.

Yep. I know, but people are going to figure that out. This is, you know, it's— It's going to be properly hyped. They're going to prop— the internet will do its thing, and it will hype it properly. Slowly, though. I think it will.

Yeah, I know. Look, it's just the lifecycle of hipsters, right? I was there before it was cool. Yeah, yeah, totally.

Maybe over 10 years ago, 2012, you taught me Python on Udacity. I took that actual course, Introduction to Backend Development. Why did you teach that course? Oh, okay.

So the very specific reason is one of my favorite professors at UVA was at the company. And so he invited me and said, Steve, do you want to teach a class? And so, and I did. It was, it was Intro to Web Development. That's right. It ended up being in that era. So it's like 2010, 11, somewhere, 12, I guess. Ended up being one of the more popular classes. It was so fun. It was so fun. You know, it's actually funny. Still, I get recognized in public. From that? From that much more than Reddit. That's crazy. I was on a plane once. This guy behind me pops over my seat. He's like, Steve Huffman? I was like, yeah. I was like, you know, and it's like, he's like, Steve, I heard your voice. I learned to program. I listened to your voice for like 40 hours. It's like in my brain. Yeah.

And honestly, it's the greatest feeling.

I would love to do that again. I should make time for it. You should. Because I just meet people for whom it helped make their life better. It's truly, it was the greatest honor.

Yeah. It was so fun. It was fascinating because I mean, the content was good, but there was something about the fact that it was coming from you, from someone who'd actually built something. And the story of Reddit was so intertwined with Y Combinator at the time and building something sort of without the permission of the big machine that was, you know, traditional, you know, software development or some megacorp. And, yeah, it was just way easier to pay attention, I think, because it was like, "Oh, yeah." Like, you could imagine yourself building something similar. And it was, yeah, it was inspiring. I really appreciate it. Well, thank you so much for taking the time. Have a great rest of your day. Have a great weekend, and we'll talk to you soon. Yeah, great.

Thanks for sharing that story. And guys, thanks so much for having me on. This was a real pleasure.

Yeah, this was fantastic. Awesome.

We'll talk to you soon. Talk soon. Take care.

Bye. Let me tell you all about Okta. Okta helps you assign every agent a trusted identity so you get the power of AI without the risk. Secure every agent, secure any agent. And let me also tell you about Restream. One live stream, 30+ destinations. If you want to multistream, go to restream.com. And we have our second in-person guest. We have Quaid Walker from Bezel.

How you doing, Quaid? Doing great. Thanks for having me.

Welcome back to the show.

Thanks so much for taking the time. Look at this mogger right here. Look at this mogger. It's bodybuilding day on TV. Dude, get a— you need a bigger shirt.

I'm just trying to keep up with you guys over here. It's good. It's good. How's life? It's great.

You guys let me hear Steve talk about watches.

Yeah, yeah, yeah, yeah, it was good.

He's like, don't get into watches.

Yeah, yeah, yeah, yeah, we'll let you react to everything he has to say. Uh, Vacheron, underrated, overrated? Underrated. Underrated. Why? Explain their strategy. Explain how they fit into the landscape of strategies that are run by other brands. How do they differentiate?

Why? I mean, from a history perspective, been around for obviously an insanely long period of time. I think they have the hype models. They have the Overseas, obviously the 222. You're seeing a lot of celebrities wear it. I think like the famous Brad Pitt wearing it in kind of the pre of the F1 movie made it go crazy. But it was the vintage example of it. They released the new one. But I think still it doesn't feel like it's at the same hype levels as like an Audemars Piguet Piguet or Rolex or things like that. So it's in the top 3, but I think from a cultural perspective, it's not totally up there yet. So it's super cool to have that mix where it's got the heritage but not necessarily the hype quite yet, despite being very hyped.

He mentioned a Reddit watch hot takes. I don't know if you've spent time there, but what hot takes do you have about watches?

Oh, that's a good question.

Or what is a— even what is a popular hot take that you see trotted out that maybe you disagree with, something like that?

I think like the hot take in the watch world right now is like this quest for independent brands. Like, I think we have a lot of clients that are diving into like the Journe market, for example, and it's like absolutely going crazy. And like the Journe Elegante, for example, like the—

Dude, I added that to my bezel like watch list, and at the time when I was adding it, you could get one for like $90,000, and now they're like $120,000.

I was talking to two clients yesterday. Really? In 2024, we got each of them an Elegante. One paid $42,000, one paid $53,000 for it. Wow. And now they're like $170,000. Wow. And that's a quartz example. It's very cool. You put it on their wrist and it'll kind of go to the time it is once you put it on. Oh, interesting. So it kind of stacks itself.

Yeah, yeah, yeah. And tell the story of FP Journe. Like why is that company interesting right now?

It's just like a scarcity thing right now in the sense that it's like the— there's a few of the brands, Journe being one, like Akrivia being one, that there's so much storytelling around the mastery of the watches that they're creating. But then also there's so few of them made, and so they get wiped up in this kind of hype cycle, and then they end up being so heavily demanded by the same type of collector that would have maybe started their collection buying Rolexes, then moved to Pateks, and now this is like the upper echelon. One thing that Journe does really well is whether you buy a Journe on the secondary market or the primary market, they kind of welcome you into the community. Oh, interesting. Which is super awesome.

Yeah, it feels like a black mark when you go into some of these stores.

Yeah, and so like they're excited that you have access to it.

You have to sheepishly admit. Yeah. Is there— should I— can I mentally comp Journe to Koenigsegg? In the sense that Koenigsegg is, you know, this very— Well, the watches work. OK, the watches work. But a lot of people will say, you know, Koenigseggs, they're expensive. Maybe there's downsides. But Christian von Koenigsegg is alive. You can get on the phone with him. You can see him at an event. And you can't do that with Ferrari. You can't do that with Lamborghini. But you can with a Koenigsegg. And similar to Patek, AP, Vacheron, the founders, the real creators are long gone, but not so with FP Journe.

Yeah, I think the community aspect of Journe is really vibrant in that sense. Like, there's a lot of events that are happening in LA. Sure. In Geneva, there's a Michelin-starred Journe restaurant. Whoa! Yeah, so there's a lot of access to the brand in ways that I think builds that mystique. But ultimately, it comes down to the watches. If you're a Journe collector, you know that there's a period of time in the year where everyone gets emails that lets you know if you get an allocation that year or not. It's not as sporadic as collecting other pieces. So like everyone knows to check their email around that period of time in the year. And like we're in the collector groups, they're talking about—

Acceptances forever.

Exactly. Yes. What— I want a recommendation for you for a tech person getting into watches. In terms of budget, let's assume that they're just like a mid-range AI engineer. They have like $100 million to spend. What would you do?

Oh, I was just— we were talking about this backstage with very different budgets than $100 million.

No, no, we can ladder it up. Yeah. Uh, what's a great watch that's just a good starting mechanical watch to actually get you into the world of horology?

Yeah, so that's a— that's the most fun part of my job is we sell the vast majority of the kind of whole spectrum. Q4, cheapest watch we sold was $750. Most expensive was $1.6 million. And we just hit a billion in supply 2 days ago, so a lot of access.

Gong, gong, gong! Hell yeah!

But yeah, it really depends on price point. Obviously there's a lot of interest there. Like, you know, I, for the sub-$5,000, I always go like Tudors and Omegas, things like that. I love the like kind of Tudor Black Bay offerings. That's a great one. So fun. Yeah, it's really grown on me.

I wasn't super into the hands.

I think the Black Bay 54 is like one of the best kind of vintage reissues that's happened in kind of modern era. Yeah, love that watch. Uh, sub-10, uh, I think vintage Rolex or like, you know, neo-vintage, like '90s Submariners, like a 16610.

And what's reliability like with a '90s Rolex?

Is that something you're going to be able to daily drive?

I So I have a couple watches. My daily is a 16610 '90s Submariner. It's probably the most vintage you can go while you're still trusting that it's like pressure tested and everything's working really good. I surfed in it this morning. No way. I work out in it, I run in it, I love it. I think I'm a big fan of using your watches. Yeah, yeah. As you scale up from tech—

Working out with a watch on, what hap— like, what are the downsides?

I think maybe curling is a little bit inhibited. Overhead press maybe on the wrist. But yeah, I don't know. I think it's very James Bond-ified for watches, where you put the watch on in the morning, and then you go work out in it, and then you go have your day in it, and then you go to dinner in it, and then you throw on a tux and you're wearing it. I think there's something so cool about Sean Connery-era black Submariners that I think Growing up seeing that, I want to recreate that.

And James Bond only wore Rolex in the books and maybe one movie, and then Omega got the—

I think it switched to Omega in the Pierce Brosnan era. Prior to that, it was—

Good deal or something.

Yeah, I think the Acquired podcast did a whole deal on Omega winning that deal. That's right.

That's right. What are you wearing today?

I'm wearing a Rolex Daytona in platinum. Heavy hitter.

Platinum's a heavier material, right?

Quite heavy. Yeah, technically, you should— we should— afterwards, you should try it out, but it's very heavy. It's like a workout.

Oh, watch salesman comes by.

He's like, you can take it off my wrist.

Take it for a spin. No, no. Jordy, what else? How is the watch industry navigating tariffs at this point? There's a lot of confusion. There's new tariffs. What is the approach from the manufacturer side? I know on the Secondary side, if you owned a bunch of watches, it was great because they just are going up in value even though there's not necessarily like more demand. There's just like kind of a leveling.

It's funny to talk about it with you guys because the last time I was here, I think I had just gotten back from Geneva in like April when they announced it.

Yeah. And so it's— yeah, that was— what was that? What event were you? Watches and Wonders. Yeah, yeah.

So the initial I think it was like 10% or 15%, whatever your initial tariff was, was announced while I was in Geneva. And then I came back and chatted with you all about like the overall experience. And then it shot all the way up to like 39, or high 30, 39%, and then came back down to 15. And so throughout that cycle, the first thing that happened is exactly what you were describing, where folks who had watches in the US immediately shot up in value and there was much more demand, 'cause two things happened. One, the pricing goes up, but also the access goes down. If you're a dealer in the US that gets a portion of your supply from Europe, now it's way more expensive to get the watches in. And so the first order of fact is pricing, like you were mentioning. The second is our authentication reports, like the number of watches that we reject. In H1 of last year, we rejected 27% of the watches. In H2, we rejected 38% of the watches. And my thesis on that is just if you were a dealer in the US and you had less access to inventory and all of a sudden it became more expensive, you were incentivized to get creative and try to pass things through the system. So maybe you're swapping dials, maybe a watch that's been pre-owned is polished and you're marking it as unworn. Like, you're doing what you can to get the throughput through that you would expect given the fact that the inventory is limited. And so that's been kind of a second order effect. Where we are now, I feel like prices are still increasing. There's less tariff talk than there was. I think people are starting to accept the reality. Brands increased prices pretty broadly. Some brands then pulled back the price increases. Shareholders. But like Patek, for example, pulled back the pricing increase. Some brands have not pulled back the pricing increase, and so it's kind of a per-brand thing. The reality is a lot of these brands can charge twice as what they charge right now, and everyone's going to still buy their watches. And so I think brands have different opinions.

Yeah, price increase from Patek doesn't affect that many people.

But it's even surprising collectors and groups felt slighted if they, you know, they just felt opportunistic if the tariffs went down. And so I think a lot of these brands are optimizing for building a relationship for hundreds of years with their clients, not just, you know, doing better in a short term.

I want to talk Oscars reactions. Kevin O'Leary was spotted wearing two watches, a Cartier Crash Skeleton and a ruby-set Rolex Daytona. Yeah. Potentially doubling the market size for watches. Are you behind this? Are you pushing this two watches on one watch on each wrist?

You got the Whoop on.

Yeah, I'm wearing a Whoop on my other side, which I think I even feel like I'm kind of a clown for doing that in some capacity. I'm a one-wrist man. Yeah, yeah, yeah. But, you know, it's good for our business if anyone wants to wear it on two wrists. And I think if you are passionate about watch collecting, if you're doing it for that reason. Amazing.

I've been seeing a lot of jump hours. Those seem to be popular at the Oscars this year. Is there a broader trend there that you're picking up on?

I think as people dive deeper into watches, the symptom of that is that I think people are more interested in, like, higher horology pieces, more complications. And so it's kind of that initial sweep that I was talking about with folks getting into independent pieces, where it's no longer just like— to talk to myself— like, I want to wear the platinum Daytona. It's now I want something that is particularly an independent brand that other collectors will raise their eyebrow about and get interested about. And so jump hour being a really fun complication. It's also just a cool thing to look at on the dial.

I saw a guy on Instagram who basically just does 3D renders of ideas that he has for watches. And he had one, the whole thesis was the watches are often photographed at 10:10. So the hour hand this way and the minute hand this way, because that's the most aesthetic presentation. And so he designed a set of watches that were— the hands would be fixed at 10:10, and then you would have a jump hour over here and a minute wheel over here, and there'd be a whole variety. One was like a flyback where you push a button and then it goes to the actual time, but mostly it's a 10:10. A lot of fun ideas.

Predictions for AP Rolex? Rolex came out with the Land Dweller. Oh yeah. A bunch of different watches that people love. AP, you know, a lot of the watches are, for better or worse, not super desirable outside of the Royal Oak. Yeah. I'm assuming they want to change that.

I think they're doing a lot to make the Code 1159, like, as interesting as possible. And so, like, the star wheel is a great example, or It's like a really awesome complication. It's really the most exciting 1159 right now. I think I'm hoping we get something in not a 41. Would excite me. Like smaller?

But what about an entirely new silhouette? I think it's far away.

It'd be awesome. I just think it's less likely.

When did the code come out? I don't know.

And is there— do they have a real reason for that? Like, they're just like, we don't need another one?

Well, I think it's quite hard to replicate the heritage and the excitement associated with the Royal Oak, right? It's like, you know, it's the Genta design, 39mm size, you know, the kind of the silhouette of the current 16202. There's only 4 watches, you know, that kind of feel like they have that degree of a mystique and history history around them, and it's probably like, you know, the Nautilus, the Overseas, maybe the Royal Oak, and then like maybe the Daytona, I think, are like the steel sports examples that fall into that category of like hype cycle. And I think it's very, very hard to recreate that. Bootstrap a new one.

And you think Rolex is more like Rolex being like, we're going to create the Land Dweller. Yeah. And just like take a risk, or— Well, that's—

if you think about it, it's like Rolex.

Like, I'm saying, like, you have, like, Land Dweller, you have Cubitus, like, why not take the risk?

Rolex doing the Cubitus was like a crazy thing.

You mean the Patek?

Yeah. Rolex doing the Land Dweller was a crazy thing. Like, it's— these companies move really slow and intentional, and typically a new model is like the same reference, but, you know, upgraded by one number, a new caliber, slight different perspectives on it, but ultimately the same model. So maybe we're moving into a world where watchmakers are starting to be more ambitious, but I think it would be like a kind of change. And so I would expect Rolex to not come out with a new model. I would expect AP to do the same. But we'll maybe get some interesting dial variations, interesting metal variations, things like that. I'm hoping for, as I was saying, We both wear the 15300. I like the 39 size. I think if we got more access to that, I think that would be really exciting. Patek, it's the anniversary for the Nautilus, so there's some conversation around what that's going to be and what's going to come out there. If it's going to be a steel reissue, if it's going to be another precious metal example. Certainly we'll get something in the Nautilus side. There's a lot of rumors around the Rolex Pepsi, and we've had those rumors for a number of years. But, you know, there's some speculation on what's gonna happen with that. It's gonna get discontinued and things like that.

Are there less leaks in the watch world? Because in the car world, you have to actually like go and road test the car. Like, people are posted up in Maranello. Like, you know, like, just it doesn't matter that the car is camouflaged, they're gonna get a picture and be able to clock it if they pay enough attention. Whereas in watchmaking, it's like you might have like a hint, like only like 3 people that actually have knowledge and they're working on it.

I can't profess to know the car example, but I would imagine that logic makes sense. The way the leak cycle typically works is like everything builds up to Watches and Wonders in April. And we're in the stage right now where you get like all the fake leaks and AI has just made that even easier because anyone now has the ability to like Photoshop whatever they think is and it looks legitimate.

It looks like it might look like a render, but they'll use a render. Yeah, exactly.

But it's like I could go and cook on any of these models build out what I think is going to be cool, and it looks convincing enough. It's like 5 years ago that was challenging. But then as you get closer, you'll get real leaks, and those are typically like they're uploading something on the website for a new banner and they miss something, or there's a production video that's made and someone has the wrong file somewhere and they leak it. And so as you get closer, sometimes historically those leaks have been correct, and then Rolex will start teasing it themselves. And so for the Land Dweller, they teased a hint of the integrated bracelet. And everyone was like, "Oh my god, it's the Oysterquartz, or what is that? I don't know." And so that's when they start to stoke it. And then Federer was seen wearing the Land Dweller a week before they announced it somewhere, and everyone was zooming in on his wrist like, "Is that a Datejust? What is that?" And so they play into it, but that's typically how it goes.

That's awesome. You launched a partnership with Kalschi. Watch price prediction markets, I saw some funny criticism, people being like, this is dumb, just buy the watch, which I thought was like a really dumb criticism because like obviously like not everyone wants to invest like $20,000, $30,000, $50,000, $60,000, you know, whatever it is. And there's a lot of other dynamics that, that make it pretty interesting. How does that product actually work under the hood? Like, what does the partnership look like? I'm assuming you guys have all of the real-time kind of pricing data, but tell us about it.

Yeah, so we built the price engine internally, like, very nerdy, true machine learning model, mathematical consensus. Our thesis was you can scrape public data, so we'll go out and we'll scrape the entire market, but there's only a few players that are at our scale that have enough of actual data. And what I mean by that is like, you know, a watch might be listed for $10,000 publicly, and that'll be what everyone scrapes. But then it might actually be offered back and forth and sell for $8,700, and that is data that is only private to like us and other marketplaces that have our scale.

Yeah. Because that data doesn't need to be public.

Exactly. It's certainly not public. And so our thinking was, you know, we can build what we feel is the most accurate data available and built out that model. And it's designed to be predictive. So it uses different inputs and values different inputs based on different models. And it's truly predictive. And so the thought was to use that internally for us to badge listings as good price, for us to offer some price guide rails for retail sellers, whatever it is. And then saw Calshi was kind of moving into collectibles. And so we connected. We were jamming on building some markets, and so for us, we're just the data kind of provider there. We have no economics from it. For us, it's like just an exciting thing for the watch world. The watch world loves it, we're gonna double down on it. If the watch world hates it, like, we're gonna learn from that. So far, the response has been largely very positive. We have two types of markets. We have the qualitative ones, so that's like, is the Rolex Pepsi going to be discontinued? And the source data for that is like really Rolex, if they're gonna roll it out. And then we have quantitative markets. So like you can bet on the Rolex index, or it can be model specific, like, you know, will the Starbucks be above this price threshold by the end of March or things like that.

That's cool. Last question, will AI ever be able to catch fake watches? Like, could you have a process when somebody's trying to list a watch to just be like, yeah, just don't even send this to us. Like, we don't, you know, we don't want it.

The blanket large, or like quick answer is no. We believe in every watch getting in the hands of like the artisans on our authentication team. They're smelling the papers, they're feeling the weight, they're looking at all aspects of it, they're opening up all aspects. However, we use AI and like computer vision at the listing process. So if you wanted to list the watches on your wrist, it would need to get approved by our team. It needs to go through our model to make sure there's no, like, obvious problems with it. And then once it gets sold, it ships to us, and then we use AI to optimize their flow. And so I don't want my team wasting any time doing anything they shouldn't do. I want the product to be really optimized for them. But ultimately, you have a watchmaker and an authentication specialist. It's in their hands, and they're feeling the watches. And I believe that is forever how they will be authenticated, at least for the, you know, medium term. I love it.

Well, thank you so much for taking the time. Thanks, guys. Great to see you. And we will move on to our next guest. But first, let me tell you about Cognition. They're the makers of Devin the AI Software Engineer. Crush your backlog with your personal AI engineering team. And I will also tell you about Finn, the number one AI agent for customer service. If you want AI to handle your customer support, go to finn.ai. And our next guest is Ankur Jain. And this is part of our Lambda Lightning Round. Let's kick off the Lambda Lightning Round. Ankur Jain is the founder of Bitnami. Built, and we will let him into the TVB Ultra Room in just a minute. Let's bring in Ankur.

How are you doing? What's going on?

Good to see you again. It's been like a decade.

I was about to say, I saw your name and I was, I was just trying to make sure I get to the dot. It has been a long time.

It has been a long time. Uh, I think the last time we hung out was maybe in Silicon Valley, maybe in New York. Lots of good times. You've been on an absolute tear. Uh, get me back up to speed. Where's the business? Uh, what's life like now? How's your 2026 going?

2026 has been, uh, it's been a crazy year so far. I can't believe it's already the end of March. Yeah, um, we—

well, first of all, we're based in New York now.

We got to get you guys out to the city. That's number one. Yeah, we— today, this morning, I don't know if you saw, we just announced built hospitality for restaurants. Um, so we— and we— look, we spent the last 5 years taking housing, which is the biggest expense for people in this in this country. Yeah. And how do we bring hospitality to the housing market? How do you make your $2,000, $3,000 a month rent expense feel the same as when you stay at a luxury hotel? So you get rewarded. Your experience is seamless. They know who you are. When you sign a lease, you're not sending in paperwork, all these types of things. We're now bringing that to neighborhood merchants. So today we announced with Chef Thomas Keller, Daniel Boulud, Major Food Group, Tao Group, Boca Group, you know, a bunch of these major restaurant groups across the country, a way for them to design guest experiences end to end in the same way we did for housing and allow you to do things like, you know, John, you go out to a restaurant and you can say, charge it to my apartment just like you would at a hotel.

And everything now becomes seamless. Amazing. Um, is the business entirely, uh, consumer focused right now in the in the sense that I think people think of Bilt as a fintech company, but also a credit card company with rewards component. Is there anything else to the business? And then I wanna know like, why have previous attempts struggled? Because it feels like there's been needed innovation in this space for so long.

Yeah, for sure. So our business is actually primarily B2B, but we are a consumer brand. Yeah, that's right. So we provide the hospitality side software for apartment managers across the United States. One in four apartment buildings run on Built. Okay. And that's our primary business. Interesting. We then have a marketplace with now over 45,000 merchants who are looking to connect with people who live in these apartment buildings. Yeah. And we facilitate that connectivity. Yeah. So that's our core business today. I think people often, often misunderstand this. It's about 11% of the cards linked to your Bilt Wallet. So when you move into an apartment building, you can link your Amex card, Visa card, get a Bilt co-brand card, and use that to pay your rent, get rewards, use it at your local merchants. About 11% of the cards linked to Bilt members' accounts are our Bilt co-brand card, but 89% of the business still comes from the broader ecosystem beyond the the card.

Okay. And what's the, what's the difference in the flow? Are people coming in for the built co-brand card and then expanding out, or are they, are they, or are you selling customers who are already in the ecosystem through other touchpoints the co-brand card downstream?

I mean, look, to your last question, it is so hard to break into housing. It is this crazy fragmented world. Yeah, that people, like, people have been happy to take paper checks up until 6, 7 years ago, which is crazy for your biggest expense. And so we started this company trying to pitch property owners, fell flat on our face, and then pivoted to say, what if we went direct to consumers to build some momentum? That's how we launched the credit card. Yeah, the card then got us momentum, which then allowed us to go back to property managers and say, look, we have hundreds of thousands of people at the time that are now using this card to pay their rent and they want to pay with the card. Why aren't you using our software for all of your renters? Mm-hmm. And then we kind of went back and then grew the business. So today almost all of our growth comes from our housing partners and people who move into the building that want extra rewards get the card. Sure. So it looks more like a traditional, like Gap or, you know, Sephora or— Yeah.

But just, just to be clear, so you could be living in a building that runs on or with a landlord that runs on BILT, you can still get rewards just through BILT, but then you can add the card. And so it's like B2B to C. Yeah, that's exactly right.

And again, most of our revenues come from the housing platform and the merchant platform. Sure. So if you think of it, property managers and merchants, our business model is similar to Shopify, if you're familiar with it. Sure. We provide the infrastructure.

They pay us for Shopify.

Sorry, their business model.

No, no, I'm kidding. I'm kidding.

So, so, so these property managers and merchants, we want to connect neighborhoods like a hotel, right? So they run on our experience platform, they pay us for the software, they pay us for the transactions, or the customer pays the transactions depending on the model. And then we get paid when we send customers to local businesses. And that's how we're able to to fund these rewards across the ecosystem.

Yeah. What were people getting wrong during that era where there was a whole bunch of skepticism around, like, is it ever going to be economically rational to give points for something as big as a rent purchase? It felt like there must be some sort of narrative violation there because you have Ken from Amex on the board, or it seems like you're a close investor. Like, it didn't seem like you went into that, like, blindly. There was clearly a strategy. So what were people getting wrong?

I think people were just missing what the business was. Sure. I mean, the reality is, as you know, it's not sexy and fun to talk about B2B software, but we think it is.

We love B2B software. But yes, I completely understand that.

People talk about the co-brand card, which we love. Yeah, man, sound effects are great. This is like, this is fun. So we— so a lot of people talk about the card and the reality, they're right. Yeah. Having a card in isolation would never make sense in this business. Yeah. But when the card is part of facilitating an ecosystem, which for us, we believe, you know, every other major commerce platform started with how do you sell car service rides? How do you sell books online? How do you sell restaurants, you know, food delivery? We said if you start with where you live, which is the home, it's the biggest expense for consumers. So right away, If you can monetize $2 trillion a year of spend on housing by being the go-to platform and the brand for housing, you have Uber for rideshare, you have DoorDash for food delivery. There was no brand in housing. That's where we started.

Yeah.

And then you start to think about every major decision you make when you move in. You pick your new restaurants that are your go-to. You pick your go-to bar, you pick your new pharmacy, you pick your new gym, you pick your school district. All of that stuff. Starts with where you live. And we are just piece by piece looking to connect that for customers. So if you saw, we just launched our concierge service in our buildings now, like a hotel. You move into a building, you sign your lease through Built, you pay your rent through Built, you get rewarded, and then you can ask the concierge, hey, I need a reservation for 3 tonight or 4 tonight in our neighborhood. Book me a dinner, schedule me a car, and use my rewards to to pay for it all, and it's all seamlessly done through the built app experience. Yeah. Last question for me.

How are you thinking about stablecoins and the future of transferring money? It feels like that's a very interesting place if you're moving a lot of money that is potentially on top of Visa rails, you're potentially paying a fee. How are you seeing that play out?

So look, we're today, just to give you some sense, we'll do over 100 billion $1 billion of housing payments to the platform. Sure. And that's now we just added mortgage. So that's going to grow quite significantly too. Yeah, I would say our general perspective is we're agnostic on how people want to pay. Okay. We, we just launched a partner with Venmo recently where people can pay with their Venmo on their rent payments or at the restaurant. Yeah, we obviously work closely with the big banks. We work with, you know, American Express network, Visa network, MasterCard network. As stablecoins develop, we'll do the same with them. If people want to— if merchants want to pay through stablecoin, we'll support that. If customers want to pay their rent through stablecoin, we'll support that. I think for us, we look at ourselves as a layer above the payments infrastructure. Sure, sure. Really focused on orchestrating a seamless experience between where you live and all the places around it.

Yeah, so just picking the right tool for the job.

Has a user ever given you a feature request to be able to tip their landlord? So, so they've never come from the—

we haven't gotten it like as a core user request, but what we have gotten is a lot of property owners saying, hey, can we simplify the annual holiday tip process rather than having to track a bunch of envelopes downstairs? So that's actually something that's coming out this holiday season. You're going to be able to tip your building staff.

Sure, that's very cool. That is cool.

That is very cool.

Thank you so much for taking the time to come chat with us. Great to catch up.

See you in New York next time.

Yeah, meet us at NYSE.

Yeah, that'd be great. We'll talk to you soon. Have a good one. Thank you. Goodbye. Let me tell you about Gemini 3.1 Pro. With a more capable baseline, it's great for super complex tasks like visualizing difficult concepts, synthesizing data into a single view, or bringing creative projects to Life. And let me also tell you about ElevenLabs. Build intelligent real-time conversational agents and reimagine human technology interaction with ElevenLabs. We have our next guest joining in just a few minutes, Michael Kratzios. He is the White House's science and technology advisor. They launched a new national AI framework for child safety and to standardize regulation and balance innovation with protection. We're very excited that he is joining today. To help us understand where the conversation around AI might go on Capitol Hill and within the government.

Interesting information out of some obesity drug test trials.

This is crazy.

Kermuse says this is not good. People have learned that GLP-1s are really effective. So if they're not losing weight, they know they're in the placebo group. So these people getting placebos are mad and just leave. And they drop out.

And the effect size is so big that they can tell. So it's going to be harder to do placebo-controlled trials, which is what we need the most in the peptide conversation. But I'm sure that they will figure it out. Anyway, we have Michael Kratzios in the Restream waiting room. Let's bring him into the TV panel. Great to see you again, Michael. How are you doing?

I am great. How are you guys?

Happy Friday. We are fantastic. Congratulations on the progress. I would love for you to take us through what was actually announced today. And then I have a whole bunch of questions about where we go from here. Absolutely.

Well, thank you guys so much for having me. I think today was a big day for the White House. As you know, last July that we've talked about before, we put out a big AI strategy for the country. But a big piece of it and something that we've been working on for some time is what is the actual national-level legislation that we need? What actually needs to be done by Congress to pass sort of the, the actual law of the land for, for AI? And we were tasked by the president to do this in December, and we unveiled that today. So broadly speaking, the main issue we were trying to tackle was figuring out a way to get past this patchwork of AI regulation. As you guys probably, I think, have talked about on your program pretty often, there's lots of states doing all sorts of stuff. While big companies can find a way to kind of navigate all the legal issues associated with this patchwork, a lot of small startups have a really hard time doing that. And we wanted to find a way to get past that and create one national standard for, for the country. And that's what we put out today. And in order to do that, we also included lots of other stuff on, on what we'd like to see on child safety, on the way that we protect American ratepayers that may be affected by, by electricity prices, stuff on American workers and also on education. So it's kind of the first comprehensive plan, I think, ever presented to Congress from the White House on what we should do on AI.

Yeah, we've been seeing a lot of polling about how AI is deeply unpopular. Are you looking at polling to understand where the American population wants AI to go? What are your sources of inputs to then make recommendations through?

Absolutely. We look at polling, we talk to industry, we talk to civil society, we talk to, we talk to all sorts of folks about where we should be going. But the numbers are what they are when it comes to the way that Americans Think about data centers, for example, and that's something that we tackled a couple of weeks ago. You may have seen the president brought all the executives from the top AI companies around, around the country to the White House and got them to commit something called the Ratepayer Protection Pledge. And this was them committing to build, bring, or buy their own power for any data center that they build. And it's a big step forward, essentially a commitment to every American that if you are going to build a data center in their backyard, They will not pay more for electricity. And I think that's the type of thing that Americans want to see. And it was great the president was able to push that.

I've been seeing rumblings of that with various hyperscalers sort of putting out small press releases saying that they were sort of going to do it independently. It was great to see it come together as like a unified— everyone's holding hands and, you know, jumping off the chasm together, working together. Do you think that this would eventually become a law, or is there some sort of framework that should be carried forward, or is the ratepayer protection program enough of an effort because the companies are public and they can be held account by their shareholders and their constituents?

So in the framework that released today, we urged Congress to codify the ratepayer protection pledge. And I think one of the key pieces of it that would be great to put in law is is the ability to legally build behind-the-meter power. This is something that needs exceptions that the Trump administration has allowed by executive action. But it's good to get that into law. And I think it unlocks a whole new range of possibilities around how you can actually put power online that can support these data centers. Because now, before, actually an action by the Biden administration was to ban behind-the-meter power generation. And this is something that makes it really hard for, for some of these hyperscalers and other data center builders to get the power that they need. Mm-hmm.

Jordy, timelines with Congress, what is your optimistic outlook in actually getting this into law?

Well, look, our hope is that we can do it in this calendar year. I think a lot of the provisions that were in the plan, particularly the ones around child safety, are ones that have had sort of bipartisan agreement for some time now. And we hope that, you know, bringing Congress together and back to your polling question, a lot of the issues that are covered in this framework are 80/20 issues. If we think about what are the types of bipartisan things that we potentially could drive forward, we think would make progress on things like child safety and also bring along the other issues that we bring in the framework.

How do you envision the child safety processes going out? I mean, when I grew up, you know, movies had R ratings. I couldn't really go to the theater. It makes sense to adopt some of that. People are aware of that stuff. There's also, you know, issues with depression and how the people interact with AI. Like, where do you think the conversation around child safety will go as we move towards more regulation around keeping children safe in America?

Where we're trying to take the conversation is putting parents back in control over what their children experience online. And for us, in the, in the framework, we urge Congress to move forward with legislation that puts that control back in the parents' hands. They should have the tools, and platforms should provide the tools to parents so they have an understanding of what their kids are seeing, what they're experiencing, what they're interacting with, who's interacting with them. And they should also have the right, the right features in place to allow parents to make decisions for what their kids will be seeing going forward. And I think those are just common sense, basic solutions that almost all parents around the world would want to have for their children. Yeah, yeah.

As a parent, I feel like that's something I would definitely want. What's the reaction been like from industry? It feels like there's always the, you know, the fear that, oh, they want to get them hooked young on whatever social media platform. At the same time, I feel like if everyone you know, has to, you know, YouTube's had YouTube Kids for a long time. They've segmented off different content there. The movie theaters obviously have G-rated films. It's not like the movie industry has been hurt by the ratings and the fact that kids can't go to R-rated movies. Has there been positive reception from the leaders of technology companies that you've spoken to about that concept?

I think technology companies want to see a viable path forward. Forward. I mean, they see the polls just like anyone else does, and they understand that parents want to have control over what their children experience online. And I mean, to some— to their credit, to some degree, they've put and installed a bunch of features, as you mentioned, some on YouTube. You've seen parental controls on Instagram for a while. There's things growing on Facebook and many other of these platforms. And I think they're already doing that. But I think for real certainty, what parents deserve is an actual national law that enshrines this forever. And I think it's something that I think could be a bipartisan push.

Jordy, any insights into the pressure that various industries are putting on states to try to get these sort of like state-level regulation and bans? I was seeing something out of New York that was trying to make it so that AI couldn't give legal advice, which to me is insane. We live in America. The rule of law, you know, matters. We're We're a very litigious country. If you can give anybody in the world access to legal superintelligence in an app in an ad-supported way, totally free.

Everyone should be able to sue each other.

No, no, I'm just saying like you should be able to drop a contract into an LLM. To actually understand the law. And understand what you're signing up for. Understand, you know, have transparency and not need— there's a lot of people that aren't in a position to be able to hire a lawyer. Totally. $300 to $1,000 plus an hour. Yeah. Understand contracts. Yeah. And everybody deals with this. So like, that's one example where I'm like, okay, where is this coming from? Maybe coming from some lobbying group for lawyers, which I understand their point of view, but it needs to be like some— Sure. Middle ground.

No, you're exactly right. I think laws like that, that, you know, we are very against. Are the primary reason why we're trying to push forward this framework. There needs to be a national standard that essentially says that, look, this is the law of the United States that the federal government has set. And so states then can't make laws in those particular areas. And the ones that you mentioned, sort of specific laws banning particular verticals, things like law, I mean, that's something that we want to make sure doesn't happen. And I think that's the power of having one national framework. Work. And that's why we're hoping to work with Congress.

What is the conversation like in DC right now around just the craziest Terminator AI doom scenario? The idea that we need maybe some, you know, big button that we can push to turn it all off.

There's been someone in DC creates a report around the issue, someone else with AI, then someone else summarizes it with AI, and then they create a response. With AI and then it gets summarized with AI and it just pings-pongs back and forth.

But potentially, potentially, but is there, is there anything to address like runaway AGI, superintelligence, like the doom scenario that there are at least a few Americans who do think is a serious issue?

Yeah, in the framework we talk about that a little bit. I think generally speaking, our view and what we put in the framework is that we want the relevant national security agencies in the US government to have the expertise and the skills to be able to do evaluations necessary to ensure our own national security in the United States. So, you know, it's great that there are a lot of people out there that have concerns about the future, and everyone has a right to that, and they can, you know, wave their papers or everything that you mentioned there. But the key thing, as governments, we need to have the capacity as an independent arbiter to do our own evaluations and think about it ourselves. And I think that's very, very, very different and distinct from it as a regulatory body. Regulatory body. I think in the last administration where a lot of this type of stuff was being thought at was at places that were essentially regulatory bodies and saying like, we're going to do these tests and if we don't like the answer to this test, then, you know, we may ban you or nationalize you or something. I think in our case, we think it purely from a kind of a national security standpoint where we want our experts that are thinking about the way that these models work to support our warfighters and ensure our own cybersecurity have the tools they need to do the testing to see where these models are ultimately going. And, and I think that's probably a very natural and appropriate role for the government to play.

How are you thinking about the government's role in reskilling, upskilling? We here on the show talk a lot about the economic opportunity that we see coming downstream of AI. That might be some disruptive innovation. People might have new roles. I talk about how the job that I have now didn't exist when I was a kid. Because live streaming hadn't been invented.

Or the other side of it, when, you know, Bezos— there was some news around Bezos' new project, you know, $100 billion fund focused on AI manufacturing. Our reaction was like, this is probably going to create a lot of jobs.

A lot of jobs.

We're going to make more things in America. But a lot of people, you know, maybe outside of tech responded by being like, oh, this is going to lead to a reduction in jobs.

But how are you thinking about reskilling, upskilling, just helping shape and, and help the American workforce over the next decade as we go through this amazing transition?

At our most basic level, we have a large set of programs at our Labor Department and our Education Department, our Small Business Administration, that are all geared towards this reskilling and upskilling category of activity. And what we try to pull it out in the framework we've been advocating for, for over a year, is those programs should be geared towards preparing American training workers for an AI economy. We should be creating programs and spending the resources that Congress is appropriating, appropriating to the executive branch to be able to go out there and actually prepare Americans for the next generation jobs. And we talk about that in, in, in the framework. And it's something that we're going to, we're going to continue to do. You know, we've done a lot of work on how, on how AI is affecting small business. I don't know if you guys had an opportunity to ever talk with Kelly Loeffler, who's our Small Business Administrator. Something that she talks about all the time everywhere she travels is what the uplift that AI is able to provide to America's small businesses. The vast, vast majority of Americans are employed by small business, not by all these big corporations that are signing all these like mega deals with all the hyperscalers. It's about all of these, you know, 5 to 50 person shops that power the American economy. And it's really these AI breakthroughs have happened over the last couple of years. That provide a whole new set of toolsets for these guys in a way that we've never seen before. So I think it could be really, really important for them.

Yeah, still so underappreciated is just how transformative this is gonna be for small business owners, a person that's a 1 to 10 person company. They're, you know, if you're an electrician, you're busy on a job, somebody calls, you don't pick up the phone 'cause you're in the middle of something. Well, great, AI can pick it up, qualify the client, schedule it, Like, that is going to be transformative.

Every extra challenge used to be a consultant or hiring a new person, and you're going to have so many different opportunities to expand your business. Well, thank you so much for taking the time to come chat with us, Michael.

We really appreciate it. Yeah, I love how pragmatic it is. Very pragmatic. It's addressing the concerns that everyday people have, and I think moving the conversation forward in a great way. Yeah.

Thank you. Have a great weekend.

We'll talk to you guys for having us. Talk soon.

Goodbye. Let me tell you about Phantom Cash. Fund your wallet without exchanges or middlemen and spend with the Phantom card. Is there any other news, Jordy, that we should go through today? There is a call sheet here. Which companies will officially announce an IPO this year? Jersey Mike's is running away with it at 73%. OpenAI is at 48%. And Dropbox at 42%. SpaceX at 88%. Jersey Mike's. Cerebras is at 91%.

We got to do the Jersey Mike's IPO.

Jersey Mike's IPO. If they don't get on the New York Stock Exchange, we're not doing our job. We got to have the founder on the show. Skims at 31%. Deal at 23%. Skims IPO is interesting. Databricks 22%. Shein 21%. Beast Industries at 14%. Anduril at 13%. This is an interesting— I'm really into this market. Anyway, thank you for watching. I have to get out of here. Big weekend. I will see you all on Monday. Leave us 5 stars on Apple Podcasts and Spotify. Is there anything else we gotta do? No. We'd love to hear from you. Sign up for the newsletter.