H200s in China, Apple Blocks Vibe Coding, Peptide Debates | Andy Fang, Matt Jayson, Dr. Cameron Sepah, Chris Gadek, Chris Hladczuk, Georgios Konstantopoulos, Matt Huang

You're watching TBN.

Today is Wednesday, March 18th, 2026.

We are live from the TBN UltraDome, the temple of technology, the fortress of finance.

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. Let's also pull up the Linear lineup. Of course, it's the system for modern software development. 70% of enterprise workspaces on Linear are using agents. We have a great show lined up. We have Matt Wang. Paradigm joining at the end of the show. Cameron from Maximus is coming in person. We got Andy Fang from DoorDash, Matt, new acquisition. We got a, we got a great lightning round. Chris from AdQuick, two Chrises. We're going back-to-back Chrises today. Anyway, thank you for tuning in. Thank you for watching TVPN. As always, the big news of the day, more news out of NVIDIA GTC, lots of NVIDIA announcements. The stock is up up, it's a $4.44 trillion company last time I checked. That is big. NVIDIA's been on an absolute tear and some really promising things with the Groq acquisition already, or like pseudo acquisition, the deal, the Groq partnership. They're already starting to explain a little bit more about how those two technologies fit together. And there's a good, there's some good coverage in Strategic about that. But the big news out of Nvidia yesterday was that Nvidia says it's restarting production of AI chips for sale in China specifically. So Jensen Huang says the company's supply chain is fired up after months of mixed signals from the Chinese market. We've been tracking this for a long time. Of course, chips were banned first from sale, for sale to China in 2022 by Joe Biden under the CHIPS Act that also unlock billions of dollars in incentives for American chip manufacturing. We began following the Intel story. We began following the Huawei story. There were a number of different initiatives. And then the narrative flipped back and forth, back and forth on what are the risks and what are the costs and benefits of actually selling chips to China. Back in 2022, after the CHIPS Act, which if you wanna read up more on it, we've interviewed Chris Miller, the author of Chip War, It's a great book. Highly recommend it.

We should have him back on.

That'd be amazing.

About that time.

Yeah. Ben Thompson's also covered the story throughout. There's a whole bunch of good stuff in "Stratechery" on chips in China. And the calculus has always been pretty clear, like on the first pass, the first order of facts, AI is an important technology. America wants an advantage in the AI race, the AI buildout. So less chips for China means more chips for America, stronger US economic engine.

Massive chip shortage right now. You see old chips being valued basically more than they were when they launched, which implies there's plenty of demand in America. So you'd want to keep those here.

And plenty of demand at TSMC from American chipmakers specifically. I mean, even Apple is sort of getting crowded down there on the 2-nanometer, which I think is better for phones than for GPUs, but they're grappling with what the compute boom, what the AI buildout boom will be for their business. But the second order effects are what are more complicated. So in November of 2022, I was writing about this and I said on a YouTube video, in the past, China's dependence on foreign technology companies has been seen as a key bargaining chip. Why would China invade Taiwan when they need to keep TSMC's manufacturing facilities online? Chip manufacturing is extremely precise. Of these factories isn't gonna withstand a rocket hitting it. So if there is any sort of broad military action in Taiwan, TSMC probably gets a little bit damaged. I mean, even the tiniest earthquake.

Right, they have to track the weather.

Yes.

Outside of the facility.

Exactly.

If the weather outside of the facilities is fluctuating at all.

It can affect you inside. Earthquakes, Taiwan is famous, or TSMC is famous where they don't even need apparently a, like a detection system or a push notification to their employees if there's an earthquake. If the employees sense that there's an earthquake, they just get up and go to the factory and start working on things. It's not like they need to like, oh, we need to email all the employees. The employees just know because that's the level of sensitivity over at the TSMC fab. So if you want to keep TSMC producing chips, you can't invade Taiwan. And so by banning the export of chips to China, the cost to China of a Taiwan invasion decreases. And so if China can't access TSMC chips anyway, it's a lot less risky to go to war. And that was always the risk with hardcore chip bans, that you'd create an increased risk of geopolitical conflict. And in 2022, the Russia-Ukraine war was about 6 months old. Global conflicts have grown significantly since then. Obviously, we've been tracking the Iran war and America's military is, is potentially stretched thin. So the risks of a Taiwan conflict are higher than ever. And so you add to the fact that everyone agrees that we will be in chip constrained, a chip shortage through at at least 2030. And the need to keep TSMC supplying chips to American companies is extremely important. It's always been difficult to parse the various arguments around selling chips to China because there's an insane amount of money at stake and many, many people whose basically their full-time job is to advocate for a particular position. I was talking to a friend who has interfaced with a lot of the different groups in Washington, and there are like huge think tanks, huge lobbying organizations on both sides. Because there's so many different interests going on.

Yeah, not to mention how much of everyone's retirement accounts NVIDIA actually makes up.

Holding up the world economy, right? That's the meme. And that's certainly true. So there are good arguments on both sides. One that keeps getting trotted out is, you know, it's important to keep China dependent on the American AI stack and it's reasonable. The better argument might just be dependent on a functional TSMC fab, but there are benefits to, to the, the idea that whatever models get built there will be applicable here. We'll be able to transfer those, that research and development that happens over there very quickly. But the much more important thing is just that the more the economies are interlinked, the less likely there is a conflict. So all of this underscores the importance of TSMC Arizona, Samsung, Intel broadly, as well as startup fabs startup fab projects like the TeraFab. And there's a whole bunch of other companies that are doing, that are at least talking about toolmaking, talking about getting a fab up and running. Nothing that's really taken off. I mean, even Samsung is not fabbing any of the leading edge NVIDIA chips at this point. But there's incredible economic incentive and People have known that this is an important kind of sort of cash cow business for a few years now at the very least. And so there's certainly been a lot of incentive to put plans in motion that might play out before 2030 because there's such a shortage. And so in general, we're traveling this like long and narrow road, but I'm coming around to the idea that selling some chips to China is the best possible move at this particular moment in time. It was easy to stick with like the first order the first order logic of just, we want the chips because we can use them to power our economy, so we should have them. But going forward, it'll be really interesting to track how China's indigenous, indigenization project of their domestic supply chain evolves. They of course have—

Yeah, my take has always been like, even if they are getting chips, it's not like the party is gonna say, actually, our domestic supply chain is no longer important because we're getting a drip of H200s. They're still going to keep the momentum that they have. It just wouldn't be like them.

There's a lot of debate over that, because momentum comes from—

You can take some of the wind out of the sails, but they can just spend more dollars.

You can slow the momentum potentially. That's the argument, is that by limiting demand, like there's a local fab in China that just says, well— A chipmaker. A chipmaker that says, okay, well, our demand is half as much, so we can't afford to scale. Sure, we have the money, but we don't really need to deliver this because there's no buyer. So you don't get the process level of execution. You don't get the excellence that comes from actually needing to run the real business. It becomes more of like, NASA than SpaceX. That's always the risk with like, you know, you're just throwing government money after it. Before we move back to The Wall Street Journal, let me tell you about Figma. No matter where your idea starts, Figma, Make, Cloud Code, or Codex or Sketch, the Figma canvas is where ideas take shape and products— ideas connect and products take shape. Build in the right direction with Figma. And let me also tell you about Public investing for those who take it seriously. Stocks, options, bonds, crypto, treasuries, and more with great customer service. So NVIDIA and the Trump administration have been involved in a complicated tango over the sales of its advanced artificial intelligence chips in China. Last April, the Commerce Department halted exports of the H20 processor NVIDIA designed explicitly for the China market. That was the nerfed H200 that was not supposed to be able to train as advanced of models, but it, It was reported that the high flyer team behind Deepseek sort of figured out how to use those chips effectively. So there was a debate over, is the H20 actually just as useful as the H200 or close to it or closer? Well, it's sort of a moot question now because H200 is coming to China, which is the more advanced version, the not nerfed version. So the company's fortunes turned again in December when the US said it would allow NVIDIA to sell its H200 processor, a chip that is a generation behind its most powerful series of GPUs. So this is not the B200, not the Blackwell, this is the Hopper, but it's still very advanced. In China, as long as the company shared 25% of its sales with the US government. So there's basically an export tariff. GPUs or graphics processing units are powerful chips used in training AI models. I think everyone knows this. In late January, after Huang visited China, officials there signaled they would approve H200 sales as well. So there was a big debate over whether or not Beijing would block the imports of H200s in order to stimulate the indigenization of the domestic semiconductor supply chain. It seems like these will be going through and they will be purchased.

They inspire local industry to grind harder.

Basically. But you have a whole bunch of AI labs that are not in the fab industry, that are not involved in the verticalization process, and they just want to train great models because they have real businesses. Like they're running they're running Alibaba or Baidu or any, any number of companies. And, and ByteDance wants good AI and they want the best chips and they're not really interested in getting into the supply chain at a deep level, at least in the short term. So, but until Tuesday, the status of the B200 or the H200 in China was unclear. In Nvidia's most recent earnings report, the company said that although it had received approval to ship We have shipped small amounts of H200 products to China. To date, we have not generated any revenue from those sales. Speaking at the company's GTC event Tuesday, Huang said that in recent weeks, demand signals out of China have strengthened. We have been, we have been licensed for many customers in China. We've received purchase orders from many customers and we're in the process of restarting our manufacturing. Our supply chain is getting fired up. Nvidia didn't comment on how much it expected to earn from H200 sales in China, but in the past, The company has said that the Chinese market for its AI processors could be worth tens of billions of dollars a year. That makes perfect sense.

And Christina over at CNBC got the original scoop near the bathroom line at GTC.

Really?

Huang happened to walk by security.

No way.

And asked if he'd answered her earlier question. Well, she said she had one more. You said you got China purchase orders. That means you got the green light from both sides. He said yes.

Wow.

That is a scoop.

You never know. As a scoop athlete, you never know where your next scoop is going to come from.

Seriously. You got to be ready. The sticking a microphone in someone's face as they're walking by really does make sense.

With the camera flashes going.

Yes. Jensen loves answering some questions on a microphone line. Chinese authorities approve NVIDIA's H200 chip sales, source says. Tior Taxes says Xi Jinping has given up. The situation with the H200 sale in China summarized, and it's everyone jumping off and Xi Jinping saying, stop, domestic chips can work maybe. And everyone at the lab saying, must get CUDA, dude, NVIDIA is the future. And they're falling into the CUDA dependency moat. Will be interesting to see the evolution of the, of the CUDA moat as other chips come online. I mean, we've seen TPU become functional. So there are questions about the CUDA moat long term, but it certainly seems like not a stopper to demand right now as labs place orders from NVIDIA.

In other news, J.R.R. Tolkien used Gen Z brain rot slang over 70 years ago. That's how ahead he was. And the quote, maggots jeered the end cigars, you're cooked. Whiteskins will catch you and eat you, they're coming.

Oh, he's like using it literally, like you will be cooked by some villain, I suppose. A cry from the Grishnakh showed that this was not mere jest. Horsemen riding very slowly had in— this is really hard to read with these hyphens— had indeed been sighted still far behind, but gaining on the orcs. So the orcs will cook you if you, if you fall behind, I suppose. Let me tell you 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 with Okta. And let me also tell you about Lambda. Lambda is the superintelligence cloud building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. Nvidia's biggest GTC announcement was a $20 billion bet on the same problem that Cerebras solved 6 years ago, says Andrew Feldman, the CEO and founder of Cerebras.

Shots fired.

Shots fired indeed. He says their next-gen inference chip, not available yet, has 140 times less memory and less memory bandwidth than Cerebras. To run a single 2 trillion parameter model, you need 2,000 Groq chips. On Cerebras, that's just over 20 wafers. Even paired with GPUs, Groq maxes out at 1,000 tokens per second. We run at thousands of tokens per second today and every day in production now. Why? When you connect 2,000 chips together, every interconnect has latency, every cable has overhead. It doesn't matter what your memory bandwidth is on paper if you are bottlenecked by the wiring between the thousands of tiny, tiny chips. We solve this with wafer scale, one integrated system, little interconnect tax. Jensen told the world that fast inference is where the value is. He's right. It's why the world's leading AI companies and hyperscalers are choosing Cerebras. And so he puts up a little graphic of Cerebras versus Rubin plus Groq together on one system, and he is touting 90 times the amount of memory, 90 times the number of chips needed to run a 2 trillion parameter model. I don't know how relevant— I don't know, is there— I wonder how this scales relative to, well, what if you're not running a $2 trillion— a 2 trillion parameter model? If you're 5x lower, does this number get much more manageable? It does feel like he's talking his book here. But I don't know.

There are probably—

But it's a big book. It is a lot to talk about. It's a big chip.

I don't know. I mean, have you, Tyler, have you ever run into any Groq chips in the wild? I remember the original demo was inferencing Llama. That was pretty fast and everyone was like, yeah, this is amazing. I've hit the API a few times. Okay. Because yeah, you can like, there's just like consumer API where you can run like local smaller models. I had heard it's very fast, a lot of chips, but that's not on its face like a huge, a huge problem necessarily. But it did seem like they had made some design decisions that maybe painted them into some particular corners, but those corners might be still economically advantageous in certain cases. Yeah, I mean, it was definitely still slower than— there was like the ChatGPT thing, right? Because that's very much like it's literally the model.

No one can beat ChatGPT.

I mean, do we know that that's what's happening with the ChatGPT demo? Because Like the Groq demo was very much like, we have a chip, here's a demo. And it was actually 2,000 chips behind the scene or a bunch of chips together behind the scenes. And so it's possible that the ChatGPT AI is doing something similar where in order to speed things up. Sure. Yeah, it's just, yeah, we don't know anything.

I just asked ChatGPT, what's your favorite thing about Tyler Cosgrove from TVPN?

What is it?

Couldn't find any information about a person named Tyler, especially in the context of TBN, which might refer to The Big Bang Theory, a popular TV show, or maybe The Bachelor Paradise, a reality TV show.

TBP?

TBP.

The Bachelor Paradise Network. TBN. Okay. Yeah, that works.

It's fast though.

It's fast, but it doesn't— I don't think ChatGPT has search. Yeah.

That would slow things down.

Let's play this video of— I want to play this video of Jensen back in September 2009. Bubble Boy is having some criticism of GTC because apparently people are getting up and asking questions that are intended to pump bags. He says, "GTC has turned into a conference less about tech and innovation and more about pumping your bags by getting a Jensen soundbite on some niche supply chain player. And so someone stood up and asked, hey, why'd you invest $4 billion in these companies? I mean, that's a reasonable—

Well, let's take a walk through history.

I make content for retail investors so they can have the average American and people around the world can have a stake. Let's play this Jensen clip from 2009. September 2009, Jensen walks onto a small stage at the Fairmont Hotel in San Jose, about 1,500 people in the room. "In computer science, if Dave Patterson and John Hennessy, and I know that Pat Hanrahan is sitting here watching me, you guys are gonna go nuts." You need to put a mic on him. "This is called CEO math. This is not computer architect math. This is called CEO math." That stage. So much smaller. Why could it make a lot of sense? This is an illustration of coprocessing.

Yeah, Carlos says no jacket.

No jacket.

He is looking pretty jacked.

There'll be a straw test at the end of this presentation. So this is a simple map. The first section on top is an example of a program that is parallel intensive.

And on the bottom is an example of a program What is he doing? Is that a camera?

And the way that I simply made it is the serial code part of the parallel intensive program takes about 1 second to run, whereas the parallelizable part of the code takes 199 seconds to run, which makes sense. It's an enormous amount of data. NVIDIA was a $7 billion company when this happened. Jensen kept saying the GPU wasn't just a gaming chip, it was a computing platform. Kept saying parallel processing would reshape every industry from medicine.

They're like, yeah, yeah, yeah, just put the video games in the bag.

Seriously. Yeah. This was, yeah, before training neural networks. Wow. Pretty, pretty remarkable. 17,000 people are at GTC 2026 now, packing a hockey stadium. Watching the same guy explain what comes next. Very, very interesting. Anyway, 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.

LinkedIn Semi Analysis is over there slopping it up. They say LinkedIn is the only social platform where you can post a sloppy reaction poll and it will get tons of engagement.

We need to do this.

Wow. They're just really throwing shots. I give LinkedIn a chance on a long enough timeline. They will all come to X. Yeah, it might take 30, 40 years. They'll make it over to the dive bar eventually.

Maybe.

And so I like that Semi Analysis leans into the particular, like, rough edges of every platform. Like, I follow them on Instagram and they actually post just like hilarious vibe reels and brain rot. It's just like brain rot.

Yeah, yeah.

But often it's completely not.

I'm the only one that's liking them. I comment on a lot of the videos.

Yeah, I comment too. It's amazing. Like, they are not having broad success, but as far as like what Instagram—

well, it's like, you know, the fine details of, you know, Inference Max or something.

Yeah. And then it's like Minecraft parkour over it. Yeah. Yeah. I was in a few of them because They clipped our interview with Dylan Patel from Cisco AI Summit and put some videos on there. But there's a hilarious one where it's like it leads in with some text about like, you know, semiconductor analyst, like explains like this thing and then there's just no audio while he's explaining it. And then it just cuts to me and I'm like, hmm. And then it cuts back 'cause we were editing the footage. And so I don't even know if it was like a mistake upload or it's like intentional, but it's all very, very funny.

Apple cracks down on vibecoding apps. It's over for you, Tyler. Apple's moves come as vibecoding apps help people create apps.

Vibecoding apps, not vibecoded apps.

I know, it's over for him times two.

No, because he doesn't vibecode on a phone.

I'm kidding, I'm kidding, I'm kidding.

He doesn't vibecode on a phone.

I'm kidding. Apple's move comes as vibecoding apps help people create apps for Apple devices as well as web apps that aren't listed. In the App Store. Let's get into Stephanie and Aaron's reporting from The Information. Apple has quietly prevented AI vibe coding apps such as Replit and Vibe Code, which help people create games and other applications, from releasing updates to their mobile apps on the App Store unless they make modifications. Company confirmed it has told some app developers that the vibe coding capabilities violate longstanding App Store rules that say an app can't run code that changes the way it or other apps function. Apple's crackdown is happening at a time when vibecoding apps are emerging as a potential threat to the company by helping developers create web apps that aren't listed on its App Store, a key source of revenue and profits for Apple. Some of these vibecoding apps also help developers create apps for Apple devices. That ability has likely contributed to the explosion of new apps launching on the App Store in recent months, leading to a slowdown in approval process in some cases, developers say. An Apple spokesperson said the policy isn't specific to vibecoding apps. Following the information's questions about the standoffs, two of the people with knowledge of the situation said they believed Apple was on the verge of approving updates to Replit and VibeCode. Those app makers had agreed to either tweak their apps the way their apps showed customer previews of Vibe-coded apps or get rid of certain Vibe-coding capabilities entirely, like making apps for Apple devices. So it sounds like you're able to basically like generate an app with Replit and then like use a preview of it that maybe is functioning a little too much and effectively allowing Replit, the app, to do things that Apple didn't approve of?

So yeah, the App Store is always— I've never heard of Vibe Code. That's what I was looking up. The App Store is always so fascinating. So if you search Vibe Code, you get an ad for Replit as the first response. Replit is number 3 in DevTools, has 14,000 5-star reviews or reviews. Then VibeCode is listed as VibeCode Website Builder, has 3.3K. Pretty solid. It doesn't look like it's charting, but it says learn how to VibeCode. No experience needed. Build websites with professional designs. Much more focused, I think, on static content, but there's been a number of these website builders in the App Store for a very long time. Then Replit ranks number 2 when you search for Vibe Code because it says Replit Vibe Code apps. Then you get Sticky, an AI game maker, Vibe Coder, AI app creator, Anything, the AI app builder, Vibe Code, Claude and Codex AI with 21 stars. There's a lot of people that are like just jumping in. There's one that's gone viral.

Replit's argument is that they're helping the user generate an app that's just opening it up in a web view. So it's effectively just a mobile— an app built for mobile web. And Apple's arguing that no, that's effectively in the Replit app.

Would you download this app, Jordy? I don't know.

That's insane.

I don't know if you can see, but it's definitely vibe-coded the App Store preview.

It's like a graphic, like an AI image of the gigachad using the computer. It's very funny. But man, there is so much IP infringement. Love a Code, not from Lovable.

Love a Code.

Build with Vibe Code. They're like, I wonder who they're trying to SEO against. Clearly there's a whole bunch of people that are just trying to steal market share from incumbents that have name recognition. Do you remember that company that was doing vibe coding on the iPhone, and they would tap your phone and basically AirDrop you the app. We talked to them at YC Demo Day last year. And there's a number of these companies that are trying to be like the AI game store, sort of like the Meta simulator, like build a simulator and create a harness that's really good at vibe coding a game. It feels like a really valuable category if you can crack it, but you are going to be bumping up against the App Store all the time.

You just have to compete with Sam Altman, Dario Amadei.

Yeah, maybe.

Amjad.

I do wonder if there will be some sort of— I mean, Roblox would be like the bigger one, maybe. Or does it come out of Codex and Cloud Code instead? Like, how do you— I wonder what the barrier and how much distribution matters there for that audience of people that want to like vibe code a game.

Yeah, for gaming, I just think Roblox is just going to continue to be like Roblox is the Roblox of Vibe Coding.

And yet we did not build our simulators in Roblox.

Yeah, but they're not like massive multiplayer games.

And we want people to be able to click a link and use it on their phone immediately.

We have a new simulator coming, by the way. We're addicted to simulation. We do love simulators. This one we're putting a little bit more effort into. And John's already addicted.

I would say this one is just actually fun. Jeremy Giffen's simulator was more just like a different way to experience podcasting. It was educational. It was educational.

Yeah. It was for people that didn't just listen to his episode on Invest Like a Bestie.

They sat down, simulated it. Yeah. But the actual interaction pattern of guessing and whatnot, it was like, oh, this is novel. And then it was like, OK, this is a chore. And then it was like, OK, how many questions are there?

200?

There were what, like 60 questions or 40? 48.

That's a lot of questions. This new simulation, you're not going to be able to put it down.

You're not going to be able to put it down.

And it can effectively solve some industry-wide issues that are happening right now.

Potentially.

We're going for impact. We're going for fun.

But Apple's had this longstanding policy around do not— like, they want to review the software. And so you can't create an app that rewrites its software. But—

Yeah, I wonder if Apple can do anything to create more like a peer, like a mini peer-to-peer experience. Because I remember I was learning how to build iOS apps when I was an early teenager. And I was so frustrated that I had built Pong, but I couldn't just share it with my dad and say, hey, you can play this. You can do TestFlight, right? Yeah, TestFlight. But TestFlight is still like— it's certainly not designed for—

It's not like AirDrop or iMessage. —actual social peer-to-peer experience. You still have to opt into the TestFlight network, do all these jumps. And for some reason, it is weird that it kicks me out of the Apple ecosystem when I get a TestFlight.

Old Rock says, high key, Tyler could make better Siri in Replit with a $1,000 budget.

—And Replit, two massive things.

No, better Siri in Replit.

Oh, better Siri. Oh, OK. Oh, yes.

For his last $1,000.

In Replit.

He's down to his last $1,000.

That's a good challenge. Yeah, maybe that should be the new constraint on hackathons. Instead of it being like two days, it's like $20,000 is the max budget. And what can you do with that budget for the token allocation.

I mean— 20K is big budget for hackathon.

Yeah. I mean, maybe the $200 plan or something, like exhaust it. You can't use multiple or something like that.

The question is how much is Apple itself vibe coding? Because the software quality in the apps that I use—

Mark Gurman said they're using cloud all the time, right? Yeah.

No, but to me, I'm saying so far, my experience recently, I've had an issue with the most important application on my phone, which is—

You were complaining about the Photos app. That was just poor design.

The Phone app has gotten a lot of hate. Yeah, John finally came around because the Phone app is like, you're like, okay, I'm going to hit this button. I might be calling this person out of the blue, even though I just want to—

And I'm not sure what phone line I'm calling them on. I guess they're designing for a world where people only have one phone number. But I still have a lot of people— this is blowing your mind, Tyler. But back in the day, like a home phone. Back in the day, people used to have multiple phone lines, multiple phone numbers. It's true. I'm so uncool. But yes, like a work phone. Like, yeah, like a work phone. I mean, when I was at FF, I had two phones, but two phones. That's right. But a lot of people will have a home phone and a mobile phone. And so that was the thing that you saved in your contact book a long time. And the problem with the new iOS phone app is that I've been calling randomly people on their home phone if I have it saved. So I need to maybe go delete those numbers or put them in a comment field so that it always calls their iPhone, 'cause I have moved to just calling people on their mobile phones. But anyway.

Well, the good news, we have our first ever Apple employee coming on the show. Really? Oh yes, we do. We do. I'm very excited. First ever. That's going to be exciting. Apple employee on the Technology Business Programming Network.

That's not— I mean, current Apple employee. We've had ex-Apple employees, of course.

But live players.

Before we move on, let me tell you about Finn.ai, the number one AI agent for customer service. If you want AI to handle your customer support, go to Finn.ai. And let me also tell you about Console. Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. So Martin Shkreli.

We got to go over to SF with Martin Shkreli. Yeah, wait.

Oh, okay. This is part 2. We're just jumping straight into this.

We're going straight into part 2. Okay, we're going straight into part 2. Take it away. He says, have you ever had the thing that you know a lot about become the current thing? That's happening now with peptides. Holy shit. I don't know where to start. Pharma basics. Most people obsessed with peptides don't know a few things. Peptides as pharmaceuticals have been around since the 1950s. Overnight. A peptide is just a small protein. Peptides have extremely short half-lives.

1950, that's like, that's like 10 minutes before 8 PM yesterday. Yes, correct.

Peptides have extremely short half-lives, often on the order of seconds or minutes. So if you're saying you're interested in peptides, you're saying I'm interested in biopharmaceuticals, but only drugs with very weak Pharmacokinetics. Pharmacokinetics. That's a new word for me. Drugs of which peptides are a subgroup usually have a specified target. This is an electrostatic interaction, usually hydrogen bonding between the atoms of the drug and the atoms of the target, typically but far from always the receptor. If you can't tell me what the target is and how the drug is binding to it, you do not have a drug. You have delusion. Next, drugs are rigorously tested, rigorously, not only for safety reasons, just Identifying the pharmacokinetics of a substance, how it travels in the body. Pharmacokinetics. Pharmacokinetics. It's a collab. They're linking and building. Exactly. Is arguably the most important starting point for any medicine. How is it metabolized? What is its half-life? Without this basic information, you can't even begin to have a medicine. You can start pharmacokinetics in animals and scale to humans, but you also need a therapeutic hypothesis. Hypothesis. This is a thoroughly vetted biological idea considered a priori as to why this medicine just might work. You very rarely discover these after the fact. Determining target engagement requires assays. Assays. Assays. This is going to be a rough one. This is a rough one. Also, I guess I didn't think—

a priori, not a priori.

Brutal. A priori assumptions. Exposed. What assay was your drug tested in? What did it show? Direct target engagement is very important to falsify your biological hypothesis.

And you can continue, John. Preclinical studies are so manufactured and fraudulent in today's day that I wouldn't rely on them for biological hypotheses unless they are from an incredible lab, were done a priori, et cetera.

There we go.

Clinical reality is far harsher. Without a double-blind placebo-controlled study, there is often nothing to talk about. If I hear, but I know dozens of people, one more time, exclamation point, Screw the FDA and pharma. Really? Really? Coming from Martin Shkreli? He is saying maybe don't screw the FDA and pharma. When most of the SF and elsewhere crowd talks about peptides, they're not thinking— and this is going to be a hard one for me— octreotide. They're thinking some random stuff that's been thrown in animal models and is not FDA approved. Look, I'm not a softie. If there was a drug that could help me or my family, I'd find a way to get it. But I'm also not stupid and spent 20 years looking at pharmaceuticals. Drug companies like to make money. Drug companies love looking at random molecules and putting them in clinical trials. There are thousands of biopharmaceutical companies that are publicly traded. It is not hard to do a trial, a clinical trial from a university. If your drug has never been tested, there is a reason. The reason is not that you are a biopharmaceutical genius who has found something cool that everyone else missed, the FDA plays an important role. They make sure that whatever is on the label is actually in the drug. That's why prescriptions are important. If I operated one of these research chemical shops, wildly illegal I might add, I would just ship people alanine or something. No one would have any idea that it wasn't BPC-BS or whatever is popular right now. The other side of the argument. But Martin, There has to be some unapproved drug out there that's useful to take. Yes, there are plenty. That is how I made a living, says Martin Shkreli. But it is not for you, world traveler, to think about this. The things you know do not apply to pharmaceuticals. It's not that you're not smart. I'm sure you're smarter than I. It just takes practice and time to understand medicine. I believe some places will even require you to go to school before you can decide who who takes what drug. Just ask your doctor for medical advice. There's a reason you don't do surgery on yourself, fly a plane by yourself, et cetera. But Martin, I want to optimize my health.

You could fly a 747 then.

I could. That is the thing that I'll take.

That's a stab. I agree with a lot of this. We don't need studies to know that you could land the 747. I could land the 747 if I needed to. If it was asked of you.

Yeah, 100%. But Martin, I want to optimize my health. No, stop it. You're not sick. It's all nonsense. Leave medicine to physicians. You do not know what you are doing. Become a physician if you are that interested, or spend a lot of time and money on biopharma. I have zero doubt you'll change your mind. There are no healthcare professionals that I know of who give an shit about these unapproved research chemicals. But wah, there are actual people, there are actual dying people in the world. Duchenne muscular dystrophy. Pecan, Leuphora. Go fix those diseases. You'll make someone and their family a lot happier than LARPing that you know about medicine. This has to end. Interesting. Lots of debate.

Yeah, so we are going to— we have a debate. Max Marcioni from Superpower will be coming on Monday at 12 to debate Martin Shkreli, the professor himself. So we're gonna have a little debate. Superpower I believe, sells these small proteins. Yeah. And so it'll be an interesting conversation. So Monday at 12 Pacific, we can look forward to the great debate.

So there's, there's some conversation in the comments here. Michael Drugan says, when people talk about peptides, they mostly mean things like retatrutide, RETA. Which is in stage 3 clinical trials and looking extremely good, or BPC-157, which has tons of clinical and anecdotal evidence. Your critique is just self-aggrandizing fluff that falls apart when you apply it to the actual examples most people are using. So he's saying, look, most people aren't using this stuff that's like crazy far out there. They're just pulling forward things that are actively being worked on by the pharmaceutical—

Just Another Pod Guy says, great post. Would be even better if you were in a flow state. With a low dose of Retta.

Martin does not think— he doesn't like BPC-157. He says BPC-157 has no evidence, LMAO. Retatrutide is literally a biopharmaceutical from Eli Lilly.

Yeah, I mean, the concern with BPC-157 has always been that it could accelerate cancer growth. Yes, it stimulates growth. Yes. And so because it hasn't been studied well enough in humans, that is a risk that people, I think, should be aware of.

How much of this is actually because of people's AGI timelines? Is there a real overlap in San Francisco between, like, yes, it might give me cancer in 20 years, but I think we will cure cancer in 10. So if it makes me look good in the next 5—

But I think it's all about people just want some type of edge. They want to alter their states. It's somewhat basically human nature. Yeah. You can make the same argument for why you should wait though, because then in 10 years, AGI will create like a super drug that will just instantly make me jacked. Oh. Right?

So it's like you can do it either way.

But no one has a decade to wait to be— Yeah, yeah, yeah.

You want to be jacked now. Because all things equal, if the cancer risk of both scenarios is zero, you'd rather be jacked for 45 years as opposed to 40 years.

How much would we have to pay you day to not lift anything heavier than a single piece of paper. There's no amount of money you can pay me. That's right. Exactly. You can't wait 10 years.

Ridiculous. 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. And let me also tell you about Cisco, unlock critical infrastructure for the AI era. Seamless, real-time experiences, new value are unlocked every day with Cisco. Pretty iconic. Someone hire this kid. Robin Smith is sharing a poster.

Warning, warning, this man has been known to increase shareholder value with a QR code out to his LinkedIn. I love this. And his name is Lachlan von Eggman.

That's a great name. And he's looking for a job as a summer 2026 electrical engineering intern. And I cannot recommend this gentleman enough. He has the, the correct mindset to go into a summer internship optimizing.

Lachlan von Eggman, you should get a job as an electrical engineering intern. If you don't, hit us up. Yes, this is good marketing. Yeah, you can— maybe we could hire Ben. Maybe we could hire this guy to work on our little electrical engineering problem back here. We got a lot of cords.

It's just doing cable management. I think he has bigger bigger fish to fry. But whatever increases shareholder value the most, he might go for it.

If you can handle cable management in the UltraDome, you can go far. Dependable thermonuclear work ethic.

That's a good phrase. Thermonuclear work ethic. He says he's smart with a capital S, and he shares his email. So if you're looking for an electrical engineering intern this summer, give Lachlan a call or shoot him an email.

Let's head over to Eric Suford. Okay. At Mobile Dev Memo. He writes, Private Credit and the AI Value Reallocation. Starts with a quote, the essence of technology is in a lofty sense ambiguous. Such ambiguity points to the mystery of all revealing, i.e., of truth.

Martin Heidegger.

I like it. Says, a familiar intuition about the emergence of any new technological paradigm is that new methods of engaging with the world create uncertainty, which is most easily interpreted through the lens of perceived negative outcomes. As private credit markets deteriorate, it's tempting to not, not only to blame AI for that decline, but also to extrapolate any dislocation to its logical extreme. That is where rising default expectations among software companies are increasingly framed as early signs of a global systemic crisis. The most convenient analogy is the GFC.

And yesterday, Carried No Interest was giving a little bit of a doomer take around some of these software private equity deals, but he was not ringing alarm bells to the tune of the global financial crisis, correct? He was just saying that some of these deals are underwater, some of those investment professionals might be needing to join different firms to find different opportunities, sort of the bull case for special situations, right?

Yeah, yeah. Leaving the firm and be like, I didn't really work on it. I was an investor, but I didn't do much investing. Saying during 2018 to 2022. I was mostly just sitting there saying, guys, I don't think we should do this deal. I guess the beauty of private credit is that you have all these different funds that are being deployed, that have been deployed on different time horizons, that have longer time horizons in general. And so you can have basically like a rolling collapse versus like a— versus, versus like, like a run on the bank where you have like one day where everyone realizes you're rolling glass. It's kind of like the worm. Like you were doing the worm yesterday. Yeah, it's kind of like, like that.

Yeah, that does not seem great. It's like a wave. Like when you start— when you start the worm. Yeah. When you start small and it gets bigger and bigger and bigger, right?

Yeah. So you're, you're 6'8". You're starting the worm and you're really tall and you're kind of coming down.

Fall like a tree.

Falling, but then you bounce back up like this and you're steadily kind of like losing.

I hope that's not where this goes. Let's continue. This framing is incomplete, says Eric Sufer in Mobile Dev Memo. It isolates the destructive effects of AI while ignoring the mechanisms through which those effects propagate and where value ultimately accrues. In this piece, he makes the case that contemporary economic conditions bear no resemblance to those leading up to the global financial crisis of 2008. I also argue, he says, that any weakening in various categories of the software landscape as a result of AI will not only mostly remain contained there, but will likely lead to economically expansionary productivity gains and efficiencies that offset potentially disproportionately losses in private credit. That's very exciting. The private credit market— he gives a little history on the private credit market. The private credit market has grown precipitously since the 2020 era pandemic. In a note from the Federal Reserve, you're just like, there will be a rolling collapse, but also I'm excited to grow.

Like, you got to hit the air horn for growth.

The authors— in a note from the Federal Reserve, the authors remark that private credit—

it is funny that if you go back to the to our early bits. Yes. We were joking about praying for bubbles, encouraging leverage. People were listening. I'm not saying— Encouraging ads. Encouraging ads. Yeah. A lot of things came true. Obviously, we were joking. They were bits. I guess some people may have taken it a little too seriously.

Isn't there some humor in every joke? Isn't that the point of humor? You mean some truth. Some truth. Some truth in every— what do they say?

There is some humor in every joke.

Yeah, some truth in every joke.

As John Coogan once said, there's humor in every joke.

I'm only on my second Diet Coke. Don't talk to me until I've had my 10th energy drink. Cheers. Anyway, this is what the Fed had to say. Private credit has emerged as one of the fastest-growing segments of non-bank financial intermediaries, NBFI, over the past 15 years or so, reaching a total asset class size of $1.34 trillion in the US alone by the middle of 2024. A report from Morgan Stanley published in October of 2025 estimated the size of the private credit market at the start of 2025 at $3 trillion. A report from Vanguard said it was $1.8 trillion. So people are sort of all over the place. Private credit is an asset class that functions as a parallel banking system. Don't call it a shadow banking system. It's just merely parallel in darkness.

Look, I'm going to be concerned. Here's what I'm going to be concerned. When you have leaders at these private credit firms that say, you know, look, private credit's amazing. Yeah, but yeah, it's unfair that we're keeping all of these gains private and we need to make them public. Yeah, yeah, yeah. So some type of like federal kind of involvement. Yeah. In the private credit. Sector could make sense.

It would also be really, really bad if one of the most respected leaders of one of the biggest banks in the world was to compare the industry to— To a cockroach. Yes, like some really, like, you know, some bug. That would be a bad thing.

You don't want to be compared to a scuttling creature. No, no, no.

Maybe a soaring eagle instead. There are some soaring eagles in this in this portfolio of private credit assets. So while it predates the global financial crisis of 2008, the scale of assets under management for private credit began growing precipitously following the global financial crisis to fill a funding gap. As the regulatory standards for lending and capital requirements applied to banks, such as those introduced by Basel III, intensified in the wake of the GFC, non-banks such as private credit lenders became a key source of funding for companies, especially those operating unprofitably, as many software companies might be earlier in their life cycles. Chernenko specifically attributes an increase in non-bank lending to regulatory rigor. So the banks become more regulated after the financial crisis. And so non-bank lending that is not subject to the same regulations increases. Many, many such cases. Another note from the Federal Reserve identifies private credit as a significant vehicle through which private credit is— or identifies private equity as a significant vehicle through which private credit is deployed, because private equity acquisitions are often funded with debt through leveraged buyouts. They used to go to banks for that, for those lines of debt. Now they're going to private credit firms.

Okay, and I think we should fast forward to down further private credit and the AI bull case, because this article is very long, and you should go subscribe to Mobile Dev Memo and read it in entirely, but let's skip to the AI bull case. Okay.

So last month, Blue Owl sold $1.4 billion in private loans across 3 of its funds and announced that it would end regular quarterly liquidity payments in one of its funds, OBDC2, opting instead for payouts tied to asset sales and other liquidity events. This month, Blackstone experienced a record number of redemption requests.

And this is— they wanted to inspire their LPs to grind harder.

Yes. The company indicated it would honor all requests.

If we give you too much mailbox money. Mailbox money? I've never heard that term. Mailbox money. That was money that, you know, shows up. You have some apartment somewhere and they just send you the check. Okay.

No, I've never heard that before. Oh, wait. Do you know that, Tyler? Mailbox money?

No. Okay. Am I crazy? I don't know. Mailbox money refers to passive income that arrives regularly with minimum active effort. Okay. Commonly in the form of real estate, learn something new, oil and gas royalties or dividends.

That's why you listen to TVPN.

It also is a mixtape by Nipsey Hussle. Nice.

And that's why you know. Also this month, JPMorgan marked down the value of loans the bank holds as collateral from private credit funds related to software businesses, reducing those funds' ability to borrow. Some of this negative sentiment is attributed to recent bankruptcies from firms to which private credit facilities were exposed. First Brands and Tricolor are among the headline cases, but the substance of the current negative outlook might be traced to fears that AI will render the vertical SaaS products into which PE funds invested in the 2020 to 2022 vintage irrelevant. So this is the SaaS-pocalypse narrative, but playing out in the private markets. In a recent note, JP Morgan States, private credit has 21% exposure to software, driven by the attractiveness of SaaS recurring revenue models and the fact that 96% of software companies are privately held. Exposure rises to 40% when including broader tech and business services, the highest among extended credit markets. We see software volatility as a sector-led reset rather than a start of a macro default cycle. Similarly, last week in an analyst memo, UBS proposed that investors increasingly want to talk about AI disruption and our tail risk scenario, a rapid, severe AI disruption. This is not our—

this 21% exposure thing is important because when John Zito was talking to the Journal, or he thought he was talking off the record and being a little candid, he was saying, you know, you, you, uh, I would expect, you know, some of these loans to record only recover 30% to 40% of the overall loan value. And that just says to me, OK, if you have 21% exposure, but you can—

Get 30 cents on that. Yeah, still recover. You're down 15% overall.

Yeah, it's not—

But the goal is not to be down at all.

Yeah, but you're also getting— a lot of these companies have been getting interest. You've been getting some amount of interest back on the loans. But again, it doesn't feel like you're getting anything close to a GFC-style systemic risk yet, especially when you account for the fund kind of time horizons and the fact they can just say, sorry, you can't get any money out right now.

So here's what the Swiss bankers over at UBS had to say. Said, what is new is a clearer catalyst, rapid severe AI disruption, assuming contagion impacts not modeled in our earlier note, we anticipate US HY, LL, and PC high-yield leveraged loan and private credit defaults could rise to 3 to 6%, 8 to 10%, and 14 to 15% for private credit respectively. Blue Owl's updated redemption policy has been compared to BNP Paribas' suspension of redemptions for 3 funds exposed to US subprime MBS in August of 2007, roughly 1 year before Lehman Brothers failed. For reasons I've stated above, I think that comparison is flawed. The GFC was a systemic crisis driven by leverage, opacity, and reliance on short-term funding. Mortgage credit risk was embedded throughout the financial system and amplified by opaque securitized products like CDOs, CDO squared, and CDO cubes, cubed securities, which established tight interdependencies between institutions. Furthermore, the deep interconnectedness of the US financial system rendered the GFC inherently systemic. Home prices were tethered to a mortgage derivatives market that was amplified in value by complex instruments that couldn't easily be valued or understood. The same isn't true of the private credit market, writes Eric Sufert, which comprises relatively straightforward bilateral loans and is largely concentrated within private equity-backed companies and dedicated credit funds. While pockets of stress have emerged, the system is less dependent on short-term funding and less tightly interconnected with core financial infrastructure. While the total value of the private credit market is vastly larger than the high-yield energy debt market of the 2014 to 2016 period, my belief is that distress will remain similarly localized and not spill over into the broader economy since there are limited direct risk conduits into the core banking system or short-term funding markets. So further, valuation multiples for PE-backed software companies, and especially vertical SaaS companies, are compressing for two reasons. One, rising interest rates have increased their debt burden as a result of their floating rate loans. So a lot of these companies were sitting at a Fed funds rate of 3%. Now, it's at 6%, or I don't know what the Fed funds rate is, but interest rates have basically doubled since the, the ZIRP era. And so their debt burden has also increased. And then two, AI and AI-enabled coding tools more specifically jeopardize the defensibility of their businesses. So viewing the composite impact of these two effects as a potential threat to the broader economy ignores the productivity gains posed by AI that imperil vertical SaaS software solutions in the first place. If a company can replace a SaaS vendor with a homemade solution because new AI-enabled engineering tools reduce the time to deployment of a narrow task-specific piece of software to days from months or even years. That's exactly what we've done here. We have a number of point solutions that didn't really exist before, but previously we might have gone to some sort of, some sort of software company and said, oh, well, can you do a special version for us?

Or can you change the economics? Change the economics. Reduces basically like moats.

Yeah, so that same company is not only saving money by no longer paying for that specific point solution, but they are also likely accomplishing more across a number of other software development-related vectors, like clipping in our example. What might be critical here is the hurdle rate in terms of labor to accomplish any task. If one person can build software now with with assistance from AI-enabled development tools that previously would have required a team, then building software becomes more accessible generally throughout the economy. For small companies, there is no build versus buy tension. The base cost of SaaS solutions may have simply priced them out of the market completely since hiring a software development team is likewise unattainable. With AI-enabled tools, the cost of building a point solution is reduced dramatically. Certain companies may now benefit from those solutions where before the economics were prohibitive across either building or buying. Citadel Security notes in a rebuttal it published to a recent macro short thesis that was something big is happening or no, that was the Cetrini prediction. Job postings for software developer roles have increased with the availability of AI-enabled development tools. Critically, AI reduces not just the cost of building software, but the cost of attempting it. Expanding the set of economically viable solutions across the long tail of firms. Not just more software, but more bespoke solutions to commercial problems that should result theoretically in better firm-level outcomes that aggregate to greater levels of economic productivity.

Let's give it up for better firm-level.

Philippe Aghion, who won the Nobel Prize in Economics last year, speaks to the expansionary effects of technology in his book, The Power of Creative Destruction. Co-authored with two other authors. Automation is thus not in and of itself an enemy of employment. Automation is not an enemy of employment. By modernizing the production process, automation makes firms more competitive, which enables them to win new markets and therefore to hire more employees. This is what we call a productivity effect. This same productivity effect was at work in prior industrial revolutions. Those induced by the steam engine and then by electricity, and explains why neither of these revolutions produced the mass unemployment that some had predicted. How can we reconcile this optimistic conclusion with the more pessimistic finding mentioned earlier that automation has a negative effect on employment at the level of commuting zones? One response is to invoke the difficulty of measuring automation and robotization at the commuting zone level, as we discussed earlier. A second possible explanation is that firms that do not automate sufficiently end up downsizing their employment, outsourcing their production, or simply going under. This would reflect an eviction effect of automation on employment. Firms that invest significantly in new industrial equipment substantially lower their likelihood of going out of business over the following 10 years compared to firms that do not make such an investment. It is thus not automation or manufacturing processes that causes firms to eliminate jobs. But rather missing the critical juncture of automation and consequently finding themselves forced to reduce the scope of their activities or even exit the market. In other words, it is through the process of creative destruction that automation can lead to job losses. In that sense, the vertical SaaS companies acquired by PE firms under low interest rate conditions face headwinds not just from rising rates, but also because it's simply more costly as a result of institutional momentum for them to rebuild their workflows and internal development processes around AI than it is for a company to establish those workflows and development processes in the first place. In other words, established companies face higher costs in adapting to the AI era than do new entrants or companies that are internalizing these tools de novo. And PE-backed firms are fundamentally averse to cost increases.

All right, go subscribe to Mobile Dev Memo and Before we get to our first guest of the show, we got to check in with Jean's, the restaurant in New York. They are weighing in on private credit. They say, if you think private credit is a problem, wait until you find out about the annuity businesses they all bought.

I love that this story has gone so broad that it is now being posted to the Instagram stories of New York restaurants. But that's where we are in this cycle. Anyway, let's move on to Andy Fang, the co-founder of DoorDash, who is waiting for us in the producer waiting room. Let's bring him into the studio.

Andy, how are you doing? Hey, what's up, guys?

What's happening? Good to see you again.

Welcome back to the show. Wait, actually, we did not have you on the show.

No, I was not on your show. I'm sorry. Yeah, my first time. Stanley and Tony were both on the show.

That's right. The boys. Is it just the three? It's just the three of you guys, right? Or is there a fourth?

Yeah, it's the Three Musketeers.

Yeah, Three Musketeers. How's life these days?

How's 2026 going? 2026 is good. I mean, a lot's going on. I mean, I guess, you know, for us, I think AI is like a big focus of mine personally at the company. So yeah, I mean, recently we just announced—

Yeah, maybe start with some background on your role, how it's changed over the period of building DoorDash and growing the company. I'm sure it's been a wild ride, but I'd love to hear it from your perspective, and then we can go into the acquisition. Yeah, totally.

I mean, let's see, 13 years in the making, I guess. I mean, so I've been always involved on the tech side, so really focused on product and technology at DoorDash. I helped scale up our product engineering teams for, I don't know, like 8, 9 years, first 8, 9 years of the company. More after that, we brought on more senior leadership to kind of manage the growing organization. And I've been more focused on how do we help DoorDash find the next big bets to help us keep growing. I think whether it's our international businesses, our DashPass subscription program, more recently with AI stuff. And so I think really just like sinking my teeth into various technology and business problems. And figuring out how we can, like, kind of continue to grow the business and find those next seeds.

Love it. So talk about how the news today ties into your focus.

Absolutely. So yeah, I was just briefly mentioning, so AI is like something that we're really trying to figure out at DoorDash, how we can incorporate it into our business. And so earlier today, we just announced that we acquired a company called Metis. They're an applied AI research lab. They went through YC and we met with them, I want to say like 6 or 7 months ago. And I think we were thinking about how to—

Was that intentional? Be honest. 6, 7?

6, 7 is viral, man. 6, 7?

I don't know. I was like 6, 7.

It really just pops into your brain. It's such a weird segment of numbers.

Oh my gosh, I'm such an old head.

No, me too. I was reading through some Wall Street Journal article and I was I kept saying the two numbers that shall not be named together.

And Jordy was like, are you doing this intentionally? September. September.

September. September. You meet them. Absolutely.

So yeah, so then September is when we first started talking to them. And we were trying to figure out how to productionize AI use cases at DoorDash. And so really, I mean, we kind of spread in terms of talking to a bunch of different startups. There's a lot of startups like trying to figure out how to deploy AI for like the enterprise use case. And we just like decided to be really open-minded about it and partner with a bunch of them. But Metas really stuck out to us. I mean, we were working with them since September. There are some really cool use cases around e-commerce and physical intelligence that I can talk about a bit that we're working with them on that we're super excited to accelerate with them joining the company. So Yeah, I mean, they're like a bunch of cracked 22-year-old AI pill people and they're like super high hustle, very knowledgeable about what's going on in the space. And we just want to inject a lot of the energy into how we're trying to innovate here.

Yeah, maybe take me through some of the history of AI at DoorDash because I imagine you were doing recommendation systems very early. Matching. Yeah, all sorts of stuff. And then when the LLM revolution happens, like the obvious bolt-on chat box went into a lot of apps, like did that happen? And then everyone starts using coding agents and IDEs. And so what have been the key turning points just internally at DoorDash in AI adoption?

Yeah, I mean, so I think To your point, I think there's two themes in terms of how we're trying to think about it. I think one is like employee enablement or like productivity. I think obviously you guys are very well aware of like since November, basically, I think there's been an explosion, especially on the coding side in terms of people being much more productive. I would say, I think at DoorDash, our evolution in terms of AI on like the product-facing side has been a longer journey in the sense that I think we've been trying to figure out how this all fits into our marketplace and how this would benefit our customers. Like, to your point, I think a lot of people, ourselves included, we're just trying to bolt on some sort of chatbot just to see how it would play in. And I think, one, when we did that a couple years ago, the LLMs were not nearly as smart. But two, I think we just need to be more thoughtful, like, how do we integrate this into the experience? And I think agentic commerce is something that a lot of people have been talking about and trying to figure out. For us, I would say, like, Agentic Commerce is basically figuring out how we can leverage this technology to make it actually easier for restaurants to run their business and for customers to find what they want on DoorDash. I think one of the big problems that we've had on DoorDash is, and what people tell me all the time when I talk to them, is like, look, Andy, you guys got a ton of selection. I have no idea what to order. It's like there's so much, so many different restaurants and all these different cuisines. I'm having decision paralysis right now. And so I think that's a very interesting problem where we feel like AI can be beneficial in helping you find what you want and give you confidence in that as a customer.

So I can't— One of the greatest luxuries or first world problems in life is decision paralysis on what to eat. It comes for everyone.

Yeah. Have you been seeing folks like vibe code sort of like plugins or add-ons or features that then you can potentially use to adopt or roll into the actual roadmap. I swear I saw someone post something about this where they vibe coded basically a Chrome plugin that would filter DoorDash for vegetarian options. And it was like a pretty mature, functionality and they just went through like a series of prompts and got to like something that they were very satisfied with. But that feels like they're sort of doing the R&D on their end. And then once it's working, you can actually bring that back. I'm wondering about this like primordial explosion of, you know, people building custom software and then the pipeline to actually bring that into the business as new features.

That's interesting because I think I haven't heard of that one specifically, to be honest, but I think there has been a lot of interest. It's like, hey, can you figure out a way to enable whatever my agent or whatever software I'm building to integrate into the DoorDash ecosystem? I've gotten some pings from founders and be like, yo, when's the DoorDash CLI coming out? I'm like, hold on, we need to figure out what that means. I think I would say, and we also did something with OpenAI in terms of being an early partner with their apps ecosystem and stuff. We're trying to figure out how we want to play in this space, to be honest.

So just walk me actually through that OpenAI partnership because like, wouldn't that be the CLI? Like if I have Codex, that's a CLI. And then if you have a partnership with OpenAI, can't Codex CLI talk to DoorDash then? Or is there some break in the chain that needs to be resolved?

I think it's more like in ChatGPT. It's like not, it's not really accessible through Codex, I don't think.

So I have to tell Codex to go use ChatGPT, open a web browser, Click on the ChatGPT box. It is weird. Like, these unhobblings are so intuitive, and then you realize that, okay, there's actually some sort of business reason or whatnot.

Yeah. Yeah. I mean, I think a lot of people are just trying to throw stuff at the wall and see what sticks. So I think one thing that we've been finding really getting a lot of traction, especially with what we're working with Metason, is I think there's some actual agentic workflows, like customer-facing workflows that we're working on that. Just us playing around with it the past couple of weeks, We legitimately think this is going to be something that's going to be beneficial to customers and help you find what you want and help you order. Okay.

So how do you think about even positioning these products? Because obviously broad public opinion on AI is not great. Is this the kind of thing— I'm assuming you're just trying to deliver a great experience and not slap like DoorDash AI, the thing you've been waiting for.

Yeah, exactly. I think we want to introduce it in a way that's actually beneficial to conversion or you finding what you want and being satisfied with it. And so that's why we've taken our time in some ways in terms of rolling something out that we want to launch more broadly. But I think some of the challenges we've had historically is these agents are not very smart at finding you what you want. This was a while ago, right? Obviously things have gotten a lot smarter. The question is, how do we leverage the intelligence that a lot of harness building, context engineering type stuff. But it's also like, how do you build a great experience around it? There's obviously LLMs, they're super smart. Sometimes they are slower with their reasoning, but we're just trying to find that balance. And I think to your point, it's like, how do we create an interesting experience that doesn't just feel like we're slapping something AI on? Yeah.

Are you going to do anything with with voice at all? I can imagine somebody in the DoorDash app saying, hey, I want Chinese food. I want to avoid speed bumps.

This is where it tells you that they integrated with Siri a decade ago.

Yeah, yeah. We do have some voice modality stuff, but I think I would say that's a little bit more just a modality in terms of how you place your order. And I guess if it's more natural language with the technology, that's a natural—

Yeah, I'm saying, yeah. Natural language, you're kind of searching over all the offerings in your area, and then it's like potentially putting together like a cart for you, or an order, you can check it out. A couple more things. How does Metas tie into the physical AI autonomy? You mentioned something about that earlier.

Yeah, so I mean, I think DoorDash wants to play at the frontier of physical intelligence. I think that's something we've talked about for a while. Obviously our—

You guys have your cute little robots.

Exactly, the Dot delivery robots. So my co-founder Stanley, who has been working on that full-time, like, we're super excited to deploy those more broadly. But, you know, we have, you know, a large courier network. We have a large support network where we're collecting billions of data points in the physical world every day. And it's like, how can we leverage that to actually push forward what's possible with robotics and physical intelligence? I think, you know, there's some interesting conversations we've had with various startups and labs on how we collaborate to push forward the frontier together. And I think Metis has some great experience doing that type of stuff in the expert data space. I mean, you know, Aryan, the CEO, is like a former Merck Corp. So I think there's some really interesting stuff there that we're interested to explore. And yeah, we're excited to accelerate some of that together with them.

Awesome. Matt from Tempo. It's coming on later today. Tempo is launching today. I know you guys were an early design partner. Yes. Talk about kind of that decision to get involved.

It's funny you bring that up because Matt literally texted me before this. Hey, we're going to be on the show together. I was like, sick. So yeah, we're excited. I mean, I'll let Matt talk more about that, but I think some of actually what Matt's working on is going to be synergistic with the agent-to-commerce stuff we're working on potentially. But essentially, like, We are very excited to just partner with companies like Tempo to see how stablecoins can be disruptive and innovative in our space. More to come in terms of what exactly our pilot use case is going to be. But yeah, we've really enjoyed working with them. And Matt's been great. We've known Matt for a while. Awesome.

I want to ask one more about Agentik Commerce. I'm trying to think about, the product that I might actually want. I understand like when most people say agenda commerce, it's like open up ChatGPT and fire off a query, say order me a burrito and it goes through DoorDash and you handle the rails. But I'm thinking more about like the next version of that and maybe it's something like, okay, I want lunch every day, but I need you to check my calendar because the show ends sometimes at 2, sometimes at 2:30. That information's on my calendar, but also sometimes I'm leaving to go to the airport and I'm flying, so I'm not available. And I could potentially, like, encode all these preferences. I have favorite things at different restaurants. I'd like some variety but some consistency. Maybe you should fire off a text message to me and let me know that I can opt out of today's, you know, order. And I feel like that's the system that someone would vibe code if you gave them the CLI, but that should probably just live within DoorDash. So how do you think about that sort of operating at a higher level, like encoding all of your preferences and then putting your DoorDash experience basically on autopilot? Totally.

I mean, I think one thing we say to ourselves is if someone was building DoorDash today, it would not look like DoorDash. And when we think about agents versus just like LLMs, agents in an ideal world, they're not just telling you stuff, they're actually doing stuff for you. And I think I would say from a consumer-facing use case perspective, in my opinion, I think besides the chatbots, I think OpenClaw is kind of the closest thing we've seen to something that is like where agent to commerce is heading towards. And so we're very inspired by that type of, I don't know, use case is the right word, but that model and that paradigm of agents doing stuff on behalf of you. And I think Your example, like, kind of hit it on the head. It's like, I don't want to just like you tell me stuff, like do stuff for me. And I think we have a ton of information on your preferences. We have a ton of information, real-time information on what's going on logistics-wise in your neighborhood and your city, like real-time inventory information and restaurants and retailers. And so like we can create some magical experiences there. And again, for us, it's not just about slapping this technology on just for the sake of it, but it's like actually stuff that's like restaurants would really appreciate because we can help them run their business better. Like a concrete example on the restaurant side, just— and I'll get back to your use case in a second— but on the restaurant side, it's like, I don't know how to price my pad thai in my neighborhood, you know, like, or, um, like, what deals or promotions should I be running right now, you know, because like I just want to figure out how to grow my sales, or, um, like, help me figure out how to, you know, the lulls between lunch and dinner, help me figure out how to like you know, get more demand there. And then on the customer side, I think to your example, it's like, you know, like, I don't want to think about like ordering my lunch to the office every day or like my weekly groceries.

The grocery use case, which is a crazy thing to say because like you basically invented the I'm hungry button on your phone and it's literally like 2 clicks for me when I use DoorDash.

And I'm like, we still—

I'm in the past. I don't want the horse anymore. Give me the car. Yesterday I didn't have lunch yesterday. We actually messed up somehow.

We at the office Forgot to order.

Yeah, we forgot to order, actually. No, I did have a salad, but it was like hours later, and I figured it out.

How rare is it to be 13 years into a business that has been as successful as DoorDash to still have 3 founders in the business? Yeah, it feels like— That's crazy. I was trying to come up with an example— Narrative violation here. —of a business at this scale. Yeah, that's interesting.

I did not expect that, but I don't take that for granted at all. I mean, in fact, Tony and I— You guys secretly hate each other.

Right? But there's like some weird contract. There's some weird contract where if any of you leave, the whole company shuts down. So you're locked to each other, handcuffed on the Titanic together.

No, I'm messing with you. I think like, I'm very grateful for the dynamic and the relationship we have. I mean, I think Tony is like, you know, one of one CEO. And I think like we have a really good dynamic together and we put a lot of trust in each other to like, just do the right thing.

Have you guys always had the same roles, basically, and sort of known, OK, I'm not going to step on this person's toes in this particular region, because then there's mutual trust, as opposed to a lot of founding teams, they get in trouble when it's like there's 3 people that want to be CEO, and they're fighting over it constantly? I mean, that's usually what happens. No, it's true.

I mean, I would say we delineated our responsibilities pretty early on, and I think Tony's role has stayed consistent. I mean, CEO is CEO. But I think Stanley and I, our roles have evolved around both how to stay on the forefront of innovation, but also just making sure like we're thinking about like what are the most important problems to solve. But no, I think we have a really good relationship. I think it's high trust. We, I mean, we're getting dinner tonight. We try to catch up every so often. But yeah, I mean—

DoorDashing dinner tonight?

No, we're gonna go out. I don't know where we're going out, but we'll pick a spot, but— Fun. Poison it. Yeah, it's a special relationship. I did not expect that, but yeah, it's right there.

No, it's cool. Last question. After the very silly Citrini piece, did you guys— have you thought about doing a hackathon and getting your best engineers to be like, hey, just vibe code DoorDash? Make the agent native DoorDash. Because I think it would actually be really interesting to just actually try to do it.

And see, you know, it's funny you bring that up because there's, uh, that Citrine report like blew up out of nowhere. And I don't know, I mean, honestly, we didn't react too much to it. I mean, I tweeted something in response to it just in terms of my own thoughts there, but I would say like in general, like two words: network effects. Yeah, I mean, I would say in general like we just try to focus on growing our business and focusing on customers, but I don't know, like we try to get people to vibe code a bunch of different use cases, but not try to vibe code DoorDash.

Yeah, there's more to it, more to it. A Bits and Atoms company. Well, awesome.

Great to meet you, Andy. Next time, bring all three.

Yeah, yeah, yeah. We'd love to have all three of you. Yeah, we'll do a roundtable. We got chairs and mics.

We'll do the whole history. Awesome. Well, yeah, congrats. Give our best to the Meta's team. Team. Congrats on the partnering. That was very cool.

We'll talk to you soon. Cheers. Have a good one. Speaking of agent to commerce, let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents. And let me also tell you about Voice Mode. You want Voice Mode? Why not sign up for ElevenLabs? Build intelligent, real-time conversational agents. Reimagine human technology interaction with ElevenLabs. And without further ado, we will bring in Matt Jason from Multiply. How you doing, Matt? What's going on?

Hey guys, nice to see you.

We're talking about my favorite topic. What are we talking about? Introduce yourself and the company.

Right on. So my name is Matt Jason. I am the founder and CEO of Multiply, and we are the first ever hybrid AI media agency for media companies. We're making ads.

You're speaking my language.

Yeah, we run some ads. Okay. Do you guys like ads?

We love ads. Do ads make you guys some money? Yes, the entire business is ads and we—

We refuse to have any revenue that isn't ad revenue.

And just to be clear, we would run ads even if we weren't being paid and we have run ads that we didn't get paid for because we have a flaw of the game.

Early on we had this bit where we would just be talking, and then we'd be changing the subject. And John would go, before we talk about XYZ, I need to tell you about Gulfstream.

And it would sound like a podcast ad. But I didn't say that we were sponsored by them. I just said, I got to tell you about this company because it's really cool, and I got to run it out.

And we would put in the description, like, this show wouldn't be possible without—

We weren't being paid. It's just like, it wouldn't be possible without them. Because they're awesome. I like it. I love this company. Anyway, break it down a little bit lower. Who are your customers? How are they actually using Multiply? What pieces of the funnel are you touching in the B2B advertising space?

Definitely. All right, so first, just the big news of today is we've raised— we came out of stealth today. We raised $9.5 million. Oh, yes. Thank you. Thank you.

Congratulations.

At Source Capital, Instacart co-founder Max Mullen, Josh Woodford leads Gemini at Google, and lots, lots more great people. And the reason we're all huddled in this advertising space, this is an old world in advertising. Google has built a massive business, LinkedIn and Twitter, but massive advertising businesses. But ads feel the same way they did 5, 10, 15 years ago despite the crazy rise in AI. Yeah. And We work for B2B companies. We work for— so I came from Brex before this. We work for high-growth B2B companies who have these crazy goals in their head. They have to go 5x their sales pipeline. And that's a lot of, I want to say, your sponsors here too. And it's pretty impossible of a task today. You've got to go figure out across all your channels, how can you go grow faster? And advertising is just stuck in the old world. So what we found a way to do is say, hey, the big problem in advertising is stuff goes stale right away. You launch a new ad, your audience is tuned out in 2, 3 days, a week tops, and your message is stale. So what we've done, we've said, great, let's go combine AI agents with the best humans in the world at media buying. And we built this combination, AI plus human media agency. And what we do together is we build self-learning apps. So really what that means is you put out initial campaigns. We're, we're connected into your sales call recordings. So we see why your customers actually say to your sales team, hey, we love what you do. Here's why we want to buy from you versus competitor. We read your customer emails. We get the same thing every time there's a new quote or pain point. We ingest this and we can actually go create tens or hundreds of new ads dynamically learning based upon why people love just what you do. Yeah. So that's our world. We're already partnering with companies like Vanta, like Listen Labs and Superhuman powering their B2B motion. Thank you. And the growth is insane. And for a lot of these companies, they're hitting 300% to 500% or more growth on these ad channels just from changing from static old world ads to new world self-running ads.

Basically, there's all this knowledge that has been trapped in sales and you're basically just like bringing it over and helping.

And it's different from just going to an off-the-shelf LLM provider and saying like, come up with some taglines for my business because you're ingesting proprietary data that happened on a phone call between the top salesperson and an interesting client that surfaced something interesting. And then you can actually pipeline that through to the creative. That make sense? That's it.

That's it. Spot on. It's grabbing this proprietary data for your business. Yeah. Like, yeah, we see why people love, like, what you do, not like some generic company might like it. Yeah. Like, specifically your product. Yeah. What you have. Yeah. And then we help you grow based on it. Yeah. Which we love. Like, one of our mission is to help companies get discovered.

And a lot of that stuff isn't— it's just not available on the open web. Like, that's probably true for consumer. People do post about their clothes that they purchase and the food that they eat, but a lot of people aren't actually sharing how a particular piece of B2B software is making their business more efficient just organically. So even if you have—

I'm in the dark right now.

Even if you have an LLM that can scrape the web and is very good at searching, it's not going to have access to that data.

So that's why data is really important. Help me out with a kind of thesis I've had for a while now, which is historically ads had very limited targeting. You'd be, let's say, running a TV ad. On a specific channel, like very limited targeting. You're creating creative that's gonna be seen by tons and tons of people, and you can't really control who those people are. It's just like the most general ad. And then with social, you could create an ad that was like effectively for 100,000 people, and you run it, and it gets stale quickly, and you have to create more creative. And it feels like we're on a trajectory towards at some point, I know that people that hate commerce and ads and the economic engine of the internet won't like this, but it feels like we'll get to a point where like every ad is basically created in real time for a specific individual. Do you see that? Do you see us kind of headed in that direction? And part of that is just like the right message at the right time for the right, like one specific person. And there's obviously kind of like layers to that stack. Maybe you need to multiply. That's also working with like a, you know, a Meta platform, etc., but headed towards like targeting that's so extreme.

It's like one ad for one— or potentially a Vibe.co where you see brands, B2B startups, AI companies advertise on streaming TV, pick channels, target audiences, measure sales just like on Meta. You could deploy for your customers. You could potentially be a big partner to Vibe.co. Our partner.

So I look, I love, I love that vision. We're seeing companies are demanding the same thing. We're running these personalized ads where we see, hey, a similar set of customers decided to buy food for this reason. Actually, let me step back. When we're seeing financial services companies with a certain sector buy for a certain reason, we're already running different ads to them than manufacturing customers with certain revenue ranges. So it's already— we're actually ingesting CRM data where we can do a version of what you're talking about and saying, hey, don't advertise to these folks. And for these folks, give them this specific messaging based on why they buy from you. So we're going there. We're having companies ask us all the time for more personalized. So we're building more for one-to-one as well. I think the question and the thing that I think is so interesting about the space, even if you can go do one-to-one advertising for everyone, how do you know it's the right message when it's predictive? And really, one of the things that we focus on is not saying that we know the right answer. A lot of traditional media agencies have like a playbook. They'll say, hey, we know the best way to break through. And we fundamentally disagree with that. We think the best approach is always-on obsessive testing. So when we talk to growth marketers, like the way we frame a lot of our AI, we frame it as we built the world's most insatiable AI agents. And just like a good growth marketer, there's no such thing as enough pipeline. And the reason I say that here is We're running hundreds of experiments every week for a lot of companies. It's thousands over time. And when you see all of that, you're actually learning by doing. You can't just say one-to-one, here's the right message. It might be one-to-one, let's try 10, 20, 30 things until we crack. This is what really is speaking the language of that person or that customer that speaks to them. And you want to speak their language, even though it's your value you're bringing to life.

All right, here's a challenge. Sell me a pen in the next 24 hours.

Sell me— actually, sell me a— it would be the B2B equivalent. Like, I want an ad for a co-packer that can make me custom pens by the billions if I want. TBPN-branded pens. This could be our next merch drop.

Thank you so much. Great to meet you, Matt.

Congratulations on the raise. Very exciting company. We'll talk to you soon, Matt. Let me tell you about Sentry. Sentry shows developers what's broken and helps them fix it fast. That's why 150,000 organizations use it to keep their apps working. And let me also tell you about Cognition. They're the makers of Devin the AI Software Engineer. Crush your backlog with your personal AI engineering team. And we have Cameron here in the studio. Are you a doctor? Should I be calling you doctor?

You don't need to call me doctor, but I am a licensed clinical psychologist.

Wow. Congratulations. Give a little bit of your background, explain the company, and then obviously I want to dive into the current thing, which is the peptide debate.

Of course, yeah. So I'm Dr. Cameron Sepah. I'm a licensed clinical psychologist. I served as an assistant clinical professor of psychiatry at the UCSF School of Medicine for over a decade.

That's a good school, right?

It's the number 3 medical school in the country. And since 2012, I've become a serial healthcare entrepreneur. So I was on the founding team of Omada Health, which is a unicorn that went public last year that creates online weight loss programs. It's helped 1 million people lose 10 million pounds and cut their risk of diabetes and heart disease in half.

That's a great KPI. Yeah.

The thing I'm much more proud of than the financial outcome. Spent a year at Trinity Ventures on the VC side of the business, and now I'm currently the founder and CEO of Maximus, where we are pioneering performance medicine.

And what's the core focus of Maximus? Obviously, we're going to talk about peptides, but you've been in— the business since before the peptide boom, correct? For sure, yeah.

And peptides is really almost like a meaningless term. It just really means drugs. It's a particular chain of amino acids that are getting hyped right now. But performance medicine spans hormones, peptides, and other small molecules.

Yeah, it does. You'll appreciate that we were at a conference in 2024, and it was like we had just started doing the show, and we were asked to give a talk. Kind of last second, so we were like, oh, what should we talk about? And it was meant to be somewhat of a provocative subject. And we were making the case that basically founders should be on a performance drug stack, right? And that venture capital firms and their platform teams should be helping them optimize.

They have their go-to-market team. They have their brand team. They'll help with hiring. They'll help you optimize your performance stack. How important is performance these days outside of bodybuilding competitions?

Yeah, I think a lot of, in fact, the trend that we're seeing, this huge consumer wave, whether in peptides or sort of performance enhancement, really trickles down from professional athletes and bodybuilders. One kind of hack I describe is you can look at the WADA list, the World Anti-Doping Association, and that gives you a pretty good indication.

That was our point, is like, there's no banned substance substances. I mean, there are like actual legal— yes, yes, drugs, but there's not like— there's no governing body in business that says you cannot take these because they actually— they work too well. Yeah, exactly right.

And I have a private practice on the side. I work with the top Silicon Valley CEOs and VCs. I can tell you without naming names, a lot of the folks that you interview are taking essentially doctor-prescribed but performance-enhancing substances in order to be more effective at their jobs with less stress.

And what does that mean effective? Because like you just being in shape and, you know, being able to walk around and not be too tired during the day, that's valuable. But then there's also focus if you're working on a spreadsheet or programming, like there's so many different like levers to pull on performance broadly. How do you narrow it down? Is it case by case?

Yeah, so the way I describe it, I think there's sort of 5 foundational health behaviors. So diet, exercise, sleep, sleep, stress management, and intimacy are the 5 things that are critical for healthspan and lifespan. And let me make a little bit of an analogy that I think really captures the field of performance medicine. So Steve Jobs used to love to tell a story about how there's an article in Scientific American that would track how effective various animals of the world are at locomotion. So who can travel a kilometer with the least amount of calories, right? Now, the condor, a bird, was the most efficient animal. And human beings, right? Human beings, despite being, you know, the crown jewel of creation, were like a third of the way down the list. So it was pretty unimpressive showing, but someone had the bright idea and they said, if you give a human being a bicycle, they skyrocket to the top of the list. And so Steve would make that analogy that the personal computer is the bicycle for the mind, right? And really revolutionized the world with that idea. I'd argue now we're seeing the precipice of two new revolutions. In that AI is the motorcycle for the mind. It's not only as smart as a PhD, but it has all of the PhDs and can work 24/7. And similarly, we're seeing a revolution in biology in this shift towards performance medicine. So if you take the 5 foundational health behaviors that I described and you want to enhance your ability to do these things because they're critical to your health, you probably have a supplement stack like you were saying, right? So let's just walk down those 5 things. So if you're trying to enhance your exercise and the main goal of exercise to build muscle mass, you're probably going to take the most studied ergogenic supplement, which is creatine. Now, creatine, if you just sit on your couch all day, is not going to do much for you except make you bloated. But if you combine it with strength training, it will increase slightly the amount of your lean mass. Now, what happens if you introduce testosterone? Now, testosterone— there's a great study that was done.

Don't get John started.

Don't tempt him. There was a great study that was done that gave guys testosterone and told them not to exercise. They gave another group, they told them to exercise without testosterone. And then obviously the combination. Now the combination obviously grew the most amount of muscle in that testosterone and strength training are obviously synergistic. But the fascinating thing is the guys who took testosterone and sat on their butt all day grew far more muscles than the guys who were lifting weight 5 days a week without testosterone. Right? It is literally a cheat code. And so it literally mogs creatine, if you will, in that it is far more performance-enhancing, which is why it's bad by a lot.

That's your best guess. That's ridiculous.

And so really, if you want to grow muscle in the most efficient way possible, testosterone is clearly the best way of doing it. Now, what about diet?

Wait, quickly on creatine, I have a question. So I heard a theory that creatine has benefits outside of muscle building. And the basic effect was that it helps with hydration. It's good to be hydrated. It's helping your cells retain water, which is why some people complain about creatine. Oh, it's just water weight, but water in your muscles makes your muscles look bigger. So people are happy with that, but that creatine could potentially be beneficial outside of a bodybuilding context.

Yeah, so there's some recent studies that show that it can help with sleep deprivation. Okay. Oh yeah. So there's some brain absorption effects.

Don't you have to take a ton of it for that though?

I think it's a little overhyped.

I tried doing that on sleep. I'm like, I don't know, boss. I don't know if I can have the 50th gummy today to get—

Well, that's the irony, right? People are megadosing creatine in order to eke out some very slight performance-enhancing benefits when they could be taking testosterone and having very clear performance-enhancing benefits, right? So the other health behavior, diet, obviously critical for minimizing fat, body fat. So the most common supplement that you can take to try to regulate your appetite is fiber. Right? Almost everyone takes fiber in some form or another, whether through whole foods or a supplement. You might lose 1%, maybe 2% of your body fat if you're consuming a fiber supplement. Okay. But now with these prescription GLP-1s, semaglutide, tirzepatide, there— its clinical studies are showing 14.7% weight loss, 22% weight loss. So it absolutely trumps anything that you can take in terms of a supplement. And it's obviously literally bending the obesity curve. In America. That's remarkable. If you look at sleep, the most common over-the-counter supplement for sleep is melatonin, right? Melatonin mildly works. It can particularly help if you have jet lag. But what if you take a growth hormone peptide? So when you— That's really aggressive. It is, but think about this. Your growth hormone peaks at puberty, and it actually declines throughout your lifespan. And so—

What if you stub your toe?

Have you considered Tren? Well, Tren is an anabolic steroid that I would not recommend for most people.

Or a stub toe.

There's a lot of interesting—

Jon's like, "All right, you forced my hand, I'll take it." Yeah, for sure.

But the fascinating thing is because growth hormone declines across the lifespan, you can take a growth hormone peptide and just restore yourself back to youthful levels.

And that's what a lot of the scientific consensus is about. Not the super physiological levels, not getting your testosterone 10,000 times bigger than what it would be naturally, but if you are truly at a deficit, bringing you back in line. That's sort of the medical consensus, correct?

Exactly. So there are growth hormone peptides now, sermorelin and tesamorelin, that enhance sleep. They enhance recovery, allow people to sleep back till they're towards that 7 to 9 hours of sleep. That's critical for physiological restoration. If you think about stress management, the other kind of fourth health behavior, the most common supplement that people take is magnesium. It helps you kind of be calm and kind of— to not be too stressed. It works mildly as well, but now you can take oxytocin. So oxytocin, you probably know, is sort of the love connection hormone, right? We've actually developed a topical formulation of it. So you can actually apply it. I'll show you what it looks like. This is a little container. You can apply it intranasally or on thin skin, such as the scrotum or the perineum. It absorbs and it reduces cortisol, the stress hormone, and it helps people feel more connected far more than a magnesium supplement.

This is what the kick streamers need if they get frame mogged. Oh, that's true. The cortisol spike. Yeah. An EpiPen.

Yeah. It will prevent actually a cortisol spike.

Yeah, yeah, yeah.

So interesting. If you're getting frame mogged.

Yes. There you go.

Yeah, you don't need to stab it.

You can just apply it, I would say. And then finally, intimacy, I think, is the most underrated health behavior. Zinc is the most common supplement that people take in terms of enhancing sort of sexual function. But the ED drugs, tadalafil, Viagra, vardenafil—

Yeah, there's been a big boom with Hims and Hers and Roman.

Exactly. So people are taking that because it not only enhances sexual function in terms of erectile strength, ability to last longer, enhancing the enjoyment of sex, but it actually increases blood flow to all of your body and into your muscles, to your brain. People take it as a pre-workout. There's interesting associational research showing that it reduces the incidence of Alzheimer's and dementia. There needs to be further studies that needs to be done to substantiate that. But that's the future that I see, is that almost everyone right now has sort of that supplement stack. But I predict, this is my kind of bold prediction, is that the top founders and VCs in the next 5 years will be taking at least 3 of those 5 things that I mentioned. So testosterone to build muscle, tirzepatide to drop fat, tesamorelin to increase growth hormone and recovery, tadalafil to increase intimacy and blood flow, and oxytocin to reduce their stress.

Okay, so those 5, so you have the 5 supplements and then the 5 drugs that need prescriptions, right? Is your prediction that the FDA will approve off, non-prescription versions of those exact chemicals, or that everyone will sort of elevate to working with a doctor that can actually write prescriptions?

Yeah, it's a great question. Prescription drugs that I mentioned are already FDA approved. Yes. But a physician has the right to prescribe any of these off-label. Yes. Right. So tadalafil is a great example. It's FDA approved for erectile dysfunction. Sure, sure. But we actually prescribe it as a blood flow enhancer. So some people literally take it as a prescription pre-workout for non-sexual reasons, or they are taking it as an anti-aging sort of drug. Sure. Because that's sort of the vision of performance medicine is it's not just treating a deficit or a medical problem, but it's enhancing your health, your aesthetics, or your performance. Sure.

And so between those two, you would expect many more doctor's visits, many more consultations with medical professionals to get access to that tier of prescription-only medicines. Yeah.

Well, I feel like that fits into this— put differently, you're positioning this as much more proactive approach approach to medicine, not just trying to survive when something bad happens or experiencing some sort of disease, but like proactively trying to take yourself from the baseline. Yeah, to the best you can possibly—

Yeah, you absolutely nailed it. The problem is right now in America, we don't have a healthcare system, we have a sick care system, right? And it's run by an oligopoly of insurance companies that are not in the business of providing healthcare, they're in the business of denying healthcare. Very sadly and tragically, when the UH G.

Yeah, there's some issues where various health tech companies will be talking to VCs, and they'll say, oh, we're going to be able to get access to insurance dollars. And the issue with that is that insurance companies don't care if something will make you healthy in 10 years, because you're most likely going to be on a new insurance plan. Absolutely. They're just wasting money, right? The average American stays at their job, like, you know, low single-digit years, I think. And so that just means, like, they're like, that's great. You're going to be healthier later, but I'm not going to pay for the benefit of some other insurance carrier.

Yeah, that was what we learned at Omada, because we sold through employers, that if the employer didn't retain their employees for more than 3 years, there's no point of selling into it, because they're like, it's not my problem. Someone else is going to have to pay for it. So really, the antidote to that is performance medicine. In that it bypasses the insurance system altogether. And within, basically, we're a cash-pay private practice that operates in all 50 states. And our belief is that the only person that should be making a decision is you and your doctor, and that every drug, of course, has risks, but it's up to you and your doctor to decide whether the cost-benefit is worth it for you in making that decision, not the insurance companies who are really in the business of minimizing the amount of care. I'll give you a really great example of this. If anyone measures their testosterone levels, unless you're below the 2nd and a half percentile, the insurance company will deem you're not sick enough, essentially, to be warranted to reimburse you.

You have to be in the bottom 2.5%. 2.5 percentile.

So, the question is naturally, what happens to the other 97.5% of Americans? Obviously, the lower half of that would probably feel better in terms of their energy, their motivation, their drive, their sexual function, their ability to build muscle, mass. But even folks who are, let's say, low-normal would probably benefit from a higher level of testosterone in terms of the performance enhancement that we're talking about. Yeah, right. Insurance will never pay for that. And so there's been a proliferation of the rise of sort of private practices like Maximus in order to really meet the needs, because I think there's this huge consumer trend, which includes peptides and hormones, of people who want to enhance their health, enhance their performance, and enhance the quality of their life. Yeah.

Uh, what are you How are you advising clients and just people broadly around the risks associated with peptides? There's a debate raging right now. Is BPC-157— Just like efficacy, but also the potential risk because we don't necessarily have— for some of them we have plenty of data, even if we don't have full-on studies.

But how are you kind of guiding Yeah, really the question is which peptides and which drugs, because we can't sort of— 'Cause some are fully approved. Yeah, exactly. So the GLPs are really, these are peptides as well, oxytocin that I mentioned, those are also peptides that have literally been FDA approved, some of them for decades, have a very substantial body of literature on them. And so, you know, comparing that to, let's say a BPC, which is not FDA approved, is a very different ballgame. Responsible clinics, I would say, like ours, are only prescribing FDA-approved drugs that have a very substantial body of human literature, in addition to a lot of clinical experience, right? So some of the, like, we're actually the largest prescriber, for instance, in clomiphene in the United States. We publish studies with Ns in the thousands on the safety and efficacy of these things. And so with the body of literature, then I think you can evaluate each drug on their own merits and decide, okay, is this something that's worth prescribing? Prescribing as a clinic. And then obviously each person between them and their doctor make a decision of whether that cost benefit is worth it for them. Now, I think individuals can always make a choice in taking something that's a little bit more experimental. That's up to them. But really, I really recommend that people are getting it through a doctor and it's fulfilled by an FDA-inspected compounding pharmacy.

Yeah, what kind of— have you heard any horror stories around these kind of like manufacture, compounders, just people getting—

Those aren't compounders. Those are just— yeah, there's a lot of people who are just literally buying black market illegal drugs directly from manufacturers or from resellers in the United States. And yeah, I mean, there's literally labs that are— the problem is you can't ensure what you're getting is pure. It's even the compound, like Peptide Sciences, which just shut down. People were testing it. They're buying a non-FDA-approved GLP that's soon to be on the market. And literally it didn't— Retta? Yeah, it didn't contain any of Retta in the vial. Or what's a common thing is they might contain it, but the dose might be off by 50%. And so what you're seeing is people having side effects because Retta is a glucagon agonist. It has effects on the heart, increases heart rate, reduces heart rate variability. If you're not able to dose it properly, properly, you're more likely to have side effects as a result. So it's a wild west. Any responsible clinician would say, don't buy things on the black market. Now obviously, individuals can make their own libertarian choices about what they do, but the difference is that there are good alternatives on the market. For instance, tirzepatide is generally efficacious for weight loss. There's no reason for most people to be taking Retta and also to be dealing with the fact that—

We have one that's FDA approved.

We have one that's FDA approved. We have two actually. You know that it's pure, you know that it's, relatively safe because there's a wide body of literature and it doesn't have the cardio effects.

So do you think Hollywood is gonna be somewhat of a model for Silicon Valley in the sense that like actors have historically been like, how did this actor have this insane transformation? Chicken, broccoli, a lot of water. And it's like, I don't know about that. But basically it's like bodybuilders lead the charge, Hollywood, kind of picks up on that, and then, but is Hollywood kind of somewhat of the model? Yeah, I think so.

And I think increasingly Silicon Valley investors and founders are gonna increasingly become the model. The problem with Hollywood is until maybe recently, a lot of people literally have to lie and say they weren't on TRT, despite the fact I have friends who are personal trainers to these guys who prep them for the movies. They're clearly on performance-enhancing drugs.

Nobody bills like, "Just say no walk." It is irresponsible for like, Zac Efron to be like, yeah, I'm on TRT, because a bunch of young people are going to see that. Totally. Totally. I want to look like that.

Exactly. But this is a really great point that you mentioned about sort of young people and TRT. And this is actually one of the whole founding genesis of Maximus, right? So testosterone is a really interesting compound. The gold standard is injectable testosterone. So if you inject exogenous testosterone from outside of your body, your body realizes it's getting enough. And so your testicles will shut down, shrink. And you become infertile. So nobody under the age of 50, or nobody who wants to still have children, should really be injecting testosterone. Now, you can take something called HCG, which is a replacement for LH, and it can help you get off of it and restore your fertility. But it's not perfect, and it's— there's some risks involved. However, there are alternatives nowadays. So we actually prescribe oral and topical testosterone. So you don't need to inject it. And it's native testosterone. It's the same testosterone that your body makes. The problem with injectable testosterone is we made sort of an interesting devil's bargain, is we took testosterone and we added a molecule to it called an ester. So you probably have heard of testosterone cypionate. The benefit of the ester is that because it's annoying to inject every single day, you only need to inject it once a week. So it becomes more convenient. The downside is it redlines your testosterone for 24/7 for a whole week. And so it shuts down and suppresses your endogenous production. So you basically become dependent on it. And if you come off of it, you're gonna go through withdrawal symptoms. The benefit of oral testosterone is that it peaks within 2 to 4 hours, lasts about 6 to 8, and topical testosterone peaks at about 2.5 hours, lasts about 12 to 6. And so there's shorter half-life, shorter duration, and they're less suppressive. The interesting thing that we found in our research is that if you take oral testosterone, it's actually not very suppressive for most people. And if you combine it with enclomiphene, which is a selective estrogen receptor modulator, it basically which blocks estrogen in a particular part of the brain, the hypothalamus and the pituitary, it prevents that suppression from happening. And so you can kind of have your cake and eat it too for the first time in that—

You want like a free lunch.

A little, there's no such thing as completely a biological free lunch, but you can maintain your natural testosterone production and then supplement it with oral testosterone and enclomiphene. And then the really revolutionary thing that we found is topical testosterone, which has been used for over half a century, You apply it to your scrotum. It can increase your testosterone by several fold. It's normally suppressive, but when you add enclomiphene to it, you can maintain your natural testosterone production and enhance it to high normal levels. In fact, we have two patents on the combination of oral and topical testosterone plus enclomiphene. And so actually, men as young as 18 can take these without suppressing their fertility markers. So it's really a revolution and it allows us to really fulfill the vision of performance medicine because there are safer alternatives. The injectable testosterone is still great for the guys who are over 50 and they don't mind being on it for the rest of their lives, but they're now essentially safer alternatives for younger guys or guys who want to maintain their fertility. And we're actively doing this research, publishing it, patenting these protocols, and disseminating them so we can democratize performance medicine.

How do you think about the personalization medicine. We talked to Paul Coeningham on Monday. He was the fellow who used AI to process some DNA data around his dog who was suffering from cancer and was able to work with a lab to synthesize a custom mRNA vaccine that was targeting the specific cancer that his dog had, sort of looking at the delta between the healthy dog's DNA and the cancerous DNA in the same animal and then was able to create sort of a one-of-one medicine for that dog. I could imagine that coming to, well, there's these peptides, they work broadly on populations, but in the future, is there a world where you're getting a tirzepatide that's designed specifically for you? And would there actually be a benefit of that, or is that where we're going?

Absolutely, it's already here. In fact, with tirzepatide, we have microdoses.

Okay. So this is like— That's modulating the amount of product, but not the actual molecular construction at all.

In this particular case, it's not. But the benefit of doing a microdose, for instance, it's for people who are not overweight, right? So BMI under 25. They may not even want to lose weight, but they might want to take it to reduce their inflammaging, is sort of the term that's used. It reduces impulsivity and can just help reduce some of the food noise. And maybe there's some folks who are not necessarily overweight, but they want to maintain a six-pack without having to constantly fight this willpower battle for the rest of their life. That's obviously a little bit more aesthetic and performance-enhancing, but that custom dose, we can— people can kind of find the right dose that fits for them. So yes, there's gonna be both personalized medicines in terms of the, the, the type of medicine, but also I think the dosing is really critical because it allows flexibility. So inclomiphene is a really good example. When inclomiphene first was introduced to the market, literally the only dosage dosage that was available, 25 milligrams. Yeah, yeah. And it's like him, that's what they studied. You get 25 milligrams of Viagra no matter who you are, right? We offer 8 different dosages all the way from 3.125 milligrams to 25 milligrams. Sure. And the way we titrate it is actually through lab testing. So we have an at-home blood test. Okay. Um, you don't have to go to a Quest. You can just stick it on your shoulder, you press a button, and you can take out about half a pinky full of blood. And we can not measure 100 markers like at Quest, but we can measure about a dozen of them.

Including testosterone.

You take it to a lab that— Exactly. Mails next day air to a lab. So you can take your baseline levels, you know where you stand. Yep. And then we can put you on, let's say, topical testosterone plus enclomiphene. You'll probably notice your total testosterone goes over 1,000, your free testosterone goes over 200, and then we can titrate you to the just the right amount by retesting you after 30 days. Yeah. And usually after 1 or 2 iterations, we can get you to make, make sure that your lab numbers are ideal, but also that your feeling and functioning your best. So this is really the promise of personalized medicine that you're talking about, that will be personalized to every single person.

How much do you care about wearable data when working with clients?

I would say in my private practice, I look a little bit more at that. At Maximus, I honestly think most wearables are a little bit overhyped.

Yeah, I noticed. I don't think you're wearing one. Yeah. You have a nice watch. Thank you.

It has no battery or technology. My anti-tech watch. Yeah, yeah. Yeah, the interesting thing as a psychologist, I think it actually causes anxiety in a lot of people, right? A lot of these sleep wearables, because they look at it and you're like, oh, my Whoop score is not 100. Now I'm going to be sleep deprived.

I have to take Tren. Exactly. Right?

To John, his entire life is just signaling like, take Tren.

I got a 98 on Eight Sleep, and I immediately Googled like D-ball plus Anavar plus Tren. Plus massive needle this big. There you go. That's what I want. Your sleep score's gonna go to about 20. I know. Do not do any of this.

But yeah, I mean, look, it can be helpful. We use tracking really when we want to see if an intervention is making a difference. So a lot of our clients actually, when they're using oxytocin, because we've published a study showing that it enhances sleep quality, extends sleep by about 25 minutes, sleep duration on average, they'll look at their Oura or their Whoop scores, and then they'll notice improvements in, let's say, REM or deep sleep. And this is kind of, I think, beyond sort of randomized control trials, the top of the evidence-based hierarchy is running really a randomized control trial of one. And so you can look at your own data, and then if you see, if you do sort of an A/B/A test where you don't do the intervention, you add oxytocin, and then you come off of it, and during that period of time, that B portion when you were taking the intervention, you're clearly sleeping better, your sleep duration's better, your deep sleep and REM are better, it gives you a pretty positive personal indication that this treatment is working for you. For you.

You didn't mention across any of those 5 categories, like the ADHD meds that I feel like have been popular in Silicon Valley for decades across Ritalin, Adderall, Vyvanse, and modafinil. There's a few others that people take occasionally. Some of them are prescribed. What is your overall thesis on that? That one feels— it's always felt like a Faustian bargain, but what's your take?

Exactly. And I think it is. And I'm— especially as a former psychiatry professor, I'm incredibly conservative about things that are psychoactive. And specifically work on the neurotransmitters.

Sure, it doesn't come from a biomarker. It's not like I get a blood test and it says, "Oh yeah, you're low in." No, there's no great blood marker for dopamine, for instance.

Sure, okay. And the diagnosis of ADHD is problematic. Most people aren't doing a full neuropsychological examination to get a true diagnosis. It's a little overdiagnosed these days.

One of the questions in the survey was like, did you lose things as a kid? And I was like, who has never lost anything as a child? Everyone loses their sweatshirt when they go to school in the morning and then it gets hot. You take the sweatshirt off, you lose it. This is not like—

I never did. You never did?

Okay, well then. Fair enough. And there's a lot of folks who may have, there's kind of a notion that people outgrow ADHD in adulthood. Really what may be happening is they're learning to behaviorally compensate through routines and patterns. But anyway, obviously if you have true clinically diagnosed, properly diagnosed ADHD, prescription stimulants can be like Ritalin, Adderall, the various amphetamines can be life-changing in that despite their, despite the risks and trade-offs, obviously, if it lets you function and keep a job, that cost-benefit, it's gonna be beneficial. From a performance enhancement perspective, I've been very public in companies like Cerebral, we're doing the wrong thing by shilling Adderall on TikTok. We won't prescribe it because it's addictive. And we don't prescribe things that are addictive because I think to your point, it is a Faustian bargain. And there's significant trade-offs in doing so. So I didn't mention focus as part of those 5. I do think there are things—

Well, it feels like your philosophy is that focus comes from the other areas.

Yeah.

If you're balanced everywhere else.

Well, in 2019, I popularized dopamine fasting as a behavioral way of improving focus. The main thing that's actually messing with people's focus for the majority of people is not ADHD, but digital distraction, right? So everyone's distracted by social media, Gaming, gambling, shopping.

I do a dopamine fast every single day, like at least 6 hours.

Yeah, that's fantastic.

Usually from like 11 PM to 5 AM.

So yeah, I mean, you know, in China actually, literally from 11 PM, I think, to like 6 or 7 AM.

You just can't access it, right?

You cannot. Computer game manufacturers are legally required to take the servers off.

Yeah, they're not allowed to take the servers off.

To do it. And so there's something there, right? That's like institutionalized dopamine fasting. Yeah. Yeah, it's interesting.

How often do you do full blood panels? And by 2035, how often do you think the average kind of American will be doing a blood panel? Yeah, I mean, assuming right now the average person does it if they're sick and they are having an issue, that's starting to change. But what's your protocol now?

Yeah, it's a great question. Unfortunately, a lot of people don't even have a primary care physician. I mean, I think the standard recommendation and one that I I will continue to espouse is everyone should get an annual physical every year and have a relationship with a doctor, ideally, a good primary care physician. Unfortunately, because of the health insurance situation that we have, it's tied to employers. A lot of people are freelancers. They don't have health insurance, or if they do, it's like a minimal amount through the open market. They don't have that. So that's problem number 1. Problem number 2 is the blood test that you get through your primary care is very sparse. So I'll tell you a funny story. Being at Omada and literally being in the diabetes industry, I never had my blood sugar level checked. And when I brought it up with my doctor, they're like, you're young and healthy. You don't need to check your blood sugar. And I was like, my dad has type 2 diabetes. I want to check and just make sure I don't develop it. But that's the mentality. Similarly, it's not routine to measure your testosterone levels or get a hormone panel unless you are incredibly low energy and symptomatic and your doctor would be concerned. They don't think it's necessary. And so the unfortunate thing is going through the traditional healthcare system because they really want to minimize the cost associated with the panel, you're not going to get a good comprehensive blood test done, right? So through Maximus, we have a comprehensive test. It costs $199.99. You can get 110 biomarkers measured, true unique biomarkers. There's a couple of companies out there that repeat tests every 6 months and they're not unique biomarkers. So that will give you an assessment of, you know, standard CBC, CMP, your lipids, your hormones, your thyroid. All the things that you need to know. And obviously now with AI, you can, you know, throw that into an AI and identify areas that are problematic. And obviously if you have a clinic or a good PCP that can work with you, it can help optimize it. Like a good example of this is almost everyone is vitamin D deficient 'cause we just don't get enough sunlight. It's the easiest thing to fix. You can take a multivitamin like the building blocks that we produce, and we do before and after lab testing so that after a couple months, you can get into an optimal range and it's going to enhance your mood, your sleep, 100 different biological functions.

We got to pop the top on this UltraDome, let the sun in, get the vitamin D going.

We actually want to eventually get to natural light.

It would be so good. We got to make it happen. Pleasure. Anyway, thank you so much for stopping by. Have a great rest of your day and we will talk to you soon. And I will tell everyone about Vanta, automate compliance and security. Vanta is the leading AI trust management platform. And I will tell you also about Graphite, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. And without further ado, we will kick off our lightning round, the Lambda Lightning Round. We will go to the cloud. Hopefully we'll see the Lambda Lightning Round begins now with Chris from AdQuick, one of our favorite people. TBPN royalty is the chat likes to say. Let's bring Chris into the TVPN Ultra Dome. Chris, how are you doing? Great to see you. Is that a TVPN green suit? Beautiful jacket. That is a fantastic jacket. I like the liner too. Anyway, I picked it up in London at Charles Turret. No way. Very cool.

Couldn't miss out on the opportunity to get dripped out with my fellow technology brothers.

I love it. I love it. Reintroduce the company and sort of explain the shape of the business these days and then take us into the news.

For anyone that hasn't heard, 1,000 AdQuick reads.

I think we did 250 last year or something like that, probably more. Yeah.

So AdQuick's building the infrastructure for real-world advertising, so out of home. So when you think about it over the last 30 years, every dollar of every ad spend has gone to doing one thing, reaching human beings with a message. And that's primarily been done through digital channels, cookies, pixels, real-time bidding, programmatic. It's all digital. But what they did forget is that people are outside. People are constantly getting exposed to an advertising medium. And this one hasn't been optimized. It hasn't been architected with technology. And so, you know, the big players like Google, Amazon, The Trade Desk, none of them have cracked the nut that is out-of-home advertising. And, you know, we're building the pipes and the plumbing for it.

Yeah. So there's a lot of interaction with the physical world. At a certain point, someone has to physically print the billboard, put it up, like tape it down, tie it down, do whatever. And you are able to abstract that and then serve a company with a thing that looks like a more traditional digital platform, right?

Yeah, absolutely. And most importantly is proving that the medium works. So we now live in an age where, you know, you don't get investment in advertising unless you could prove that it works. And so we pioneered the measurement and the analytics. So if you get exposed to an out-of-home ad and then to subsequently do something you buy online in an app, go into a store, we can attribute that to the exposure and then model it out with all the rest of your channels.

So it's out-of-home advertising made easy and measurable. That's what I'm getting from this. That's right. But take us through the more recent news.

What's going on with the business? Yeah, so there's 3 big players in the space in the United States. It's a $10 billion market. 2 billion roughly is sold by Outfront Media. And you'll see their billboards and their transit in New York, Los Angeles, and all the major metros.

And to be clear, they own the physical real estate? They own the physical, like the structure that the billboard goes on?

That's part of their business? Or just the rights?

Or the rights? The rights and the structures. Got it. Okay. Yeah.

And so this is an organization that's like 2,000 employees. 1,000 sellers, all regionally distributed. And they've been selling this medium very much like, you know, real estate agents have prior to like your, your Open Door, your Zillow, so on and so forth. And for the first time ever, they're going to have world-class technology to support their sales efforts across these 1,000 sellers. And then they'll be also using our audiences to find the best fit for their inventory against an audience, and then also measure the success for their clients.

Amazing. So yeah, walk me through the measurement problem because on my drive to work every day, I drive through Hollywood, I see so many billboard ads for movies. And whenever I buy tickets, I wind up going on Fandango and there's a huge disconnect there. I don't know that I've gotten a survey But with the right data, you could probably see that I was in this area, or I purchased something on a credit card panel in this area, and I might have been exposed to that billboard. I could imagine how you could piece that together, that I went to this particular movie because I saw this billboard. I've been seeing billboards for the Project Hail Mary, and we are actually going. And the billboards should get some of that attribution. But other than hearing me say it right now, I don't know how you would puzzle that together. So what is the process these days?

Sure. So your weather app doesn't make money by telling you it's going to rain. OK. We buy up all the highest quality mobile traffic data. Sure. We pipe 7 billion pings into our platform every single day. Yeah. So we know where you are. Yeah. What time you're at, what direction you're heading at, what speed you're heading at, and all these things. Sure. And then we We use that to basically identify when the mobile device is inside of a geofence that we have geofenced all of the inventory in the platform. Got it. And then we count that as an exposure. Sure. We then map the exposure to a pixel on the website and an SDK in an app. Sure. Or we also, in the case of brick and mortar, will actually geofence the store location. Yeah. And so we can kind of connect the dots that way.

Yeah, that makes a ton of sense.

Breakdown kind of trends in the SF out-of-home market. Like how is kind of all the AI funding boom?

Yeah, does a billboard up in SF cost like 10 times as much as anywhere else? Like what's the ratio at least?

Yeah, I mean, it's after, you know, we had that 2023, we had the software recession. 2024, we started to see things come pick back up. And in the last 12 months, it's gone gangbusters for the AI companies. To answer your question, yeah, the prices are some of the most expensive in the market. But some of our customers are some of the leading AI companies, Fin being one of them. ElevenLabs, we've done campaigns for. And so yeah, you're starting to see AI companies use this channel to kind of break through the clutter of digital.

What's the most underrated out-of-home form factor in your view?

If you're B2B, I really like account-based marketing or account-based— so basically you take all of your HQs for your target accounts, you put them in the map, and then you buy up all the inventory surrounding them.

That's like bus stops, basically just flood the zone in a 2-block radius kind of thing.

Yeah, exactly. And I don't know if you recall, but say maybe 5 or 6 years ago, the founders of Brex took over all of San Francisco, most of San Francisco being their target market. And I think they spent like half a million bucks and customer acquisition.

Yeah, I think people haven't fully processed that there seems to be increasing returns as you spend more on out-of-home, which is that if you spend $50,000 on a campaign somewhere around LA, you're gonna have some effect. But if you spend dramatically more than that, like an order of magnitude, it seems like you really break through in a different way. It's funny, Sam Blonde, who did that original campaign. He's back with his new— a new company and taking the same approach, you know, heavy, heavy, heavy out-of-home spend. Yeah.

So some of our teammates in San Francisco were saying that Monaco's up and down the 101. I think they bought like 60, 60 or 70 billboards. So they're making their presence felt.

Yeah. He told me about that before we had him on the show and I was like, That is such a bold way to introduce your company. But it really does break through. I've seen photos of it. And if you're there, you'll see.

What about on the supply side? How heavy are the restrictions on bringing on new out-of-home supply? I'm assuming, obviously— You're thinking for your house? City by city. Yeah, I mean, I'd love to put a massive billboard on my house. No, I just think in general—

Hey honey, I sold the front yard.

I sold the roof.

Hey, our mortgage is going to be half the price now. That's amazing. Yeah.

So in terms of limitations, in 1965, the Highway Beautification Act was passed. So that kind of restricts how many—

Required every highway to have beautiful billboards.

Yeah. So basically there's only a certain subset that you could put near highways. But you're starting to see all these other formats kind of come about. There's folks putting ads on trucks. There's folks wrapping Ubers. They're putting ads inside of Ubers. They're pulling planes in the sky, which is kind of old school. Skywriting? And then drone shows are the new big thing. And expect to see a lot more of those in the future.

I know. I'm surprised that hasn't been utilized. The only other thing is, can you— Can you put a blimp on AdQuick? Like, I've given this advice to enough companies now, and they haven't done it. I feel like— But just like flying a blimp, like a Goodyear blimp over SF for—

Goodyear's just cornered the market. I think that they wouldn't— I don't think Goodyear would sell their blimp for $1 billion. I think it's just their greatest capital asset they have, because you could monetize that thing so well flying around San Francisco.

You know, we do have 2,000 media owners on the platform, and I'm pretty sure that might fit the bill.

That'd be amazing. Love it. Amazing.

Well, congratulations on the partnership. And thank you again for being our— you had to have been the second advertiser right after Ramp to back us. And we'll never forget your support.

We appreciate you, Chris. Thanks so much.

Congrats to the whole team.

We'll talk to you guys. Great to see you. Goodbye. Let me 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. Well, Railway automatically takes care of scaling, monitoring, and security. And let me also 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.

I'm expecting our next guest to be duct taped to his chair.

Oh, oh, is he? Is he? He was duct taped. Oh, he's free. He got free. I'm alive. Okay, okay. Introduce yourself, introduce the company, and then explain the duct tape, uh, stunt? I love it.

Co-founder, CEO of Hanover Park. Think about us as financial infrastructure to power investment firms, private equity, venture funds, the most unsexy industry on the planet. We're now making sexy. Thank you. Thank you for yourself. Yeah. Yeah. So I was duct taped to a chair and I announced a blooper that I tagged Jordyn, you guys, and I was like, okay, how can we make the most boring industry in the world sexy? And that was obviously replacing human duct tape. And so, yeah, pumped to do it.

We launched our Investor Series data today. OK, so the metaphor is there's a lot of human duct tape inside these funds piecing different puzzle pieces together to manage the financial assets and LP relationships. And you make all of that go away, correct?

100%. And also, like, I don't know, I've been saying this for 2 years, like, B2B SaaS is dead. We want to sell outcomes, not tools. And so tools are things that are a bunch of duct tape around with humans and software. And we're like, we're going to replace the human heavy services business out there too. Sure, sure.

Sure.

What is the key moment for an allocator using the platform that really— Clicks. —wows them or clicks?

When you want to make a $100 million follow-on decision and you have to ping your CFO and say, hey, can you pull together the data that we haven't had in real time for the past 2 months? Can you figure this out? AI agents solve the problem by getting you real-time data and then help you make those decisions. And so we're pulling in data from granola notes, we're pulling in data from emails and board decks, we're pulling in data from all the boring accounting data on those positions. And so people are like, oh my God, I can finally make a decision in real time without waiting a bunch of days.

Okay, so there might be a whole bunch of decks from all the different investments and all the different companies that have various ARR figures and growth rates, but they're sort of buried. Handover can go in there, get that data, put it in some sort of centralized database, and then visualize that or maybe even help with marks? Is that something you do, or is there still a human in the loop for that?

Yeah, even more importantly, we think there's a world where you don't even need to log into Hanover Park. We're going to expose all the data via MCP server. And so we launched that where people are accessing it via cloud, cowork, and via GPT, et cetera. And so that's been a key piece, I would say, for like GPs, whereas like CFOs are going to use us for things like capital calls, distributions, financial reporting, according to LPs, et cetera.

Got it, got it. So interesting. So if somebody's on the enterprise OpenAI plan or something, do they have to integrate this themselves or is that just an off-the-shelf plugin? And then if they want to actually get this into a deck, what's the current stack to get that information? They would be exporting it, transforming it over in their LLM tool of choice. And then taking it to the LPs.

It's actually even worse. You would have to email some human called your fund admin and say, hey, can you pull that data that's locked in some random tool you don't give me access to? Yeah. You then get a spreadsheet, you then put that into another tool, you then put that into Claude or GPT, etc. And so a lot of the value prop here is you're going to pay a fund admin anyway to do this boring financial reporting. LPs demand this from a third party. And so instead of having that data locked away, we can at least give you access to it, and then you can put it into cloud and other things.

Got it. How big is the market? This is a market that, like, you know, compared to, like, SMB—

There will never be too much venture capital, Jordan.

I know, I know, I know. And we're trying to get more people to be brave enough and enter this market. There's $130 billion raised by equity. OK. Yeah, so I'm assuming with— venture funds, private equity, all this stuff, you have to kind of ignore like individual SPVs and entities. But I'm assuming you can kind of clock pretty precisely like what the— how big the market is, and then you got to go and try to get as much of it as you can.

Yeah. So there's $100 trillion of global assets and there's a $20 billion public market.

I know, but isn't that like somewhat of a—

Like, are you going to get PIMCO at some point? Right? Like, it's—

yeah, it's like hedge funds, right? You have like private equity funds, real estate, private equity, venture, etc. Like, SS&C is a public company today. They have like $5 billion of revenue. CITCO has $5 billion of revenue, right? So there's a bunch of like these massive dinosaur companies that exist, and it's really like venture and private equity is just the start.

Okay. Okay. And so within— I mean, venture also scales from $15 million fund to $15 billion fund, like I imagine that there's some trade-offs where the bigger you go, the bigger the dollars, but also the bigger integration, the more requests, versus somebody who's just starting a new fund. They don't have any software, so it's super easy to get off the ground. Have you been picking one or the other as like a beachhead market, or have you just been doing like whoever you know? Yeah, you'd be surprised.

The bigger it is, the easier it is to differentiate, actually, and so a lot of these bigger funds, funds. They've been like running around with players like S&C, which I call like the evil empire, which is basically like you're using them and all of your data is kind of trapped. Whereas like an early-stage venture fund that's just getting off the ground, things are pretty clean and pretty easy. And so we've actually focused on like mid-market enterprise. So think like $250 million plus in assets, venture and private equity to start at least.

Do you do any work on the— is it all equity or are you starting to do stuff in in private credit or private or like public debt or any of the other asset classes, or are you just purely in alternatives? Purely in alternatives today.

I would say like 6 to 12 months from now, obviously focus on expanding into things like private credit. Yeah, obviously big opportunity being here in New York. I've been saying this for a long time. Favorite sponsor, Ramp. It's like Stripe for payments, Ramp for expenses, Hanover Park for investments. And so that's kind of the—

Have you had any customers just ask you to just handle of just all the actual investment decision-making. They're like, I'll just put money in a box and then you deploy it. Let me know how it goes. Talk to my LPs too. Not yet. I don't want to talk to them anymore.

Don't talk to founders or LPs. Not yet.

I think it's been funny. It's like a lot of LPs that we talk to are like, I keep, you know, my venture funds are investing in AI, but their back and middle office is run by the anti-AI. And so This idea of like, we're investing in it, we might as well move to it as well has been pretty exciting for people. But I think long term, it's like unlocking alpha for the investment firm based on all the unsexy financial data that we have could be interesting. Yeah, we got a lot of work to do to get there.

Unpack a little bit more of like the dream unsexy proprietary data set. Does that look like PitchBook or Crunchbase data? There's a lot of different data sources that venture capitalists go to. Throughout diligence, is do you want to be a conduit to those sources or do you imagine that the LLM tools that venture capitalists are using will interface with those separately and just do the joins in whatever work session is happening?

Yeah, it's actually not really Crunchbase PitchBook. It's really like the unsexy financial cash flow data. So like, give me the cash flows from every LP through history and how that flows into our next cap call or our next distribution or our next, you know, financial statements. So it's actually that data, which is the data that's mostly trapped in these fund admins. I think Crunchbase and PitchBook access will get commoditized. Those will go into the underlying models. And I don't think we want to play that game.

Yeah. Talk about the overall health of the venture capital market. I feel like you're in a unique position. There's been these like sort of warring narratives. We joked earlier, like $110 billion raised, but that was OpenAI. There's a lot of— there's like sort of a crunch in K-shaped recovery, maybe in venture capital, there's a variety of mega funds that are scaling up. Like, what are you seeing when someone just asks you from the outside, maybe they're, you know, public markets investor and they say like, how is venture capital doing these days?

We're actually seeing a lot of haves and have-nots. It's a very stark bifurcation because when you start a fund, you basically sign a lawyer and a fund admin. And so we see from inside perception of how your raise is going. And so literally it's— there's people that are raising in 3, 6, 9, 10 weeks for a first fund, which is crazy fast versus 6 to 12 months. And then there are people who are never raising. And so I think it's the people that are spinning out of a lot of the major funds that are targeting, call it, $100 to $200 million fund ones are having an actually an easier time from some of the emerging managers that are focused on like $25 million fund ones. And so it's a pretty big bifurcation. There.

How are you processing this new holding company roll-up firm? It looks like a private equity firm, but it also sort of looks like a startup. We've had a couple of these founders on where—

Yeah, those probably aren't clients because they just get a bunch of money.

Yeah, would they be, or could they be in the future?

How do you think about that?

It's actually funny, we've had a lot of these folks reach out because fund admin and like accounting services more broadly, right? Think about a is a subset of that because we're doing financial accounting reporting for investment firms. Yeah, that is a ripe industry for roll-up. And so people have been like, oh my God, should this be a fund admin roll-up as an idea in terms of vertical buyout? What we've realized is like if you just toss agents, the word agents on top of any accounting data, it doesn't really do anything. Having like the core ERP for the fund and the system of record that we had to build from day zero is actually like way more important than just saying like AI solves all our problems, which I don't think it does quite yet.

Yeah. Have you had to do any fine-tuning on your own data or your system, or are the agents sort of able to find their way around your system pretty quickly? Did you need to build your own MCP servers or your own CLIs for these things? We were just talking about this with DoorDash. There's demand for a DoorDash CLI now, and it's funny, but it makes a lot of sense. In the context of agentic engineering?

Yeah, we've actually had to build our own little AI fund accountant, I call it, which is doing simple things like AI cash reconciliation. When you get an email from someone to make a decision, how does a human do that? It's actually client-specific. Different clients have different preferences. And so almost how do you build that like AI fund accountant? So we're using a bunch of the models off the shelf to do that as well. And then we're doing some fine-tuning on top.

Yeah. Does that look like a, like a.md file, like a skill, or is this more of like a full harness that's actually, you know, like a whole bunch of code that you've written? Like, how do you see that system growing as you develop more core capabilities? Yeah, I think so.

There's a bunch of different ways that we sliced it, but like for something like migration, which is how do I get the most complex financial data in the entire world from one system to another, that's a lot about building harnesses and building skills in and around the core underlying model provider to say this is the weird, messy financial data from pick your massive investment firm. Here's how we need to think about this. And then for things like doc processing, a little bit differently. So depends on, depends on the workflow.

Yeah. Tell me a little bit about the progress of the company. Like where are you based? How big is the firm? How are you growing? Where do you see this going over the next couple of years?

Yep. So 2 years old, raised $31 million. Obviously we're in today, we're announcing a $27 million Series A led by Emergence. We went— oh, oh, oh, there we go.

We did it. We're just getting started though. We're just getting started. Oh, there we go. We're just getting started.

What are you talking about?

I love it. I love it.

Um, we went 1 to 15 billion in assets in the past 12 months. We 15x'd. We've had a lot of excitement from folks. I actually— I was joking with the team. I said, guys, this is a top 0.1% problem, having unlimited demand and people excited. Now let's go make them raving fans, which is our goal for '26. And so not cursor, but we're focused on, you know, keeping our heads down and executing. That's great. That's great.

Well, thank you so much for taking the time. I love it, dude.

It's awesome. Awesome to see your, your progress. Yeah. I mean, I remember, I mean, years. Yeah. Just the absolute thread grinder from the 2021 era. Just ran it up.

Yeah. And you're just— and so I'm sure you Angel invested, right? I did.

I did miss this one.

Yeah, what happened, guys? Sahil Bloom got in. You didn't get in.

This is the nature of the show. This is why we're podcasting, because we see people like you all the time, and we just miss everything. So back to the show. I don't. Yeah, Jordy makes a lot of good investments. I just don't. I just don't do it. Anyway, thank you so much for taking the time to come talk to us. Great to see you, Chris.

We'll talk to you soon.

Congrats. Thanks. 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 Let me also tell you about Fantom Cash. Fund your wallet without exchanges or middlemen and spend with the Fantom card. And our next guest is in the Restream waiting room. Georgios, what's going on? How are you doing?

Good afternoon. Thank you for having me.

Good afternoon. Good to see you. Great to see you. Please kick us off with an introduction on yourself and the company. So I'm Georgios.

I drive the engineering team at the Tempo Project. Tempo is a payments-first blockchain that we started developing in August Today we just launched the Temple mainnet live in production. Alongside it, we also launched the machine payments protocol, which is our answer to how agents should pay. Yeah, if you think about it, agents are great at searching the web. We use agents all the time in our development process by AMP, Cloud Code, Codex, whatever it is for developing. But we find it very hard to make them pay. So we developed a solution for it, and I'm excited to talk about both.

So you have MCP. A lot of people were saying, like, why doesn't MCP have a payment spec already baked into it? There's this whole debate about HTTP maybe having a payment spec in it about 20 years ago or something. MPP solves that. But is MPP like a subset or a superset of MCP, or is it a completely different spec? Is there any relationship between the two technologies? They're complementary.

You should think of MCP as a way to talk to an API that has been designed to be baked into the agent harnesses. And MPP is a way that you can payment gate an MCP. So imagine that you have your service, you have written an MCP for it, but you actually want to make that a paid service. Maybe, for example, on TBPn, you guys do great prep for all of the people that you work with. Hey, well, what if you just added a paid API to that? Sometimes.

Yeah. Awesome. So, yeah, I think anybody that hasn't been living under a data center has heard of the potential for agents to use stablecoins and blockchain technology generally for various services. You guys are working with a bunch of really cool companies. Like, what kind of use cases are you trying to guide your design partners towards enabling?

I think the most exciting use case that they have to do around commerce, basically there's a lot of very interesting work around, hey, I'm on ChatGPT, my ChatGPT is browsing around, but hey, it wants to buy things for me. I think that's one of the things that is unlocking. The other thing to think about is that your ChatGPT is again browsing the web, but it hits a paywall. You wanted to be able to, instead of saying, hey, I can access this information, you wanted to instead be able to say, actually, pay 30 cents and you can actually access this wonderful article that would have been paywalled, which might improve your research queries by, say, 30% or something.

So in that case, you'd be going to, like, so is there a plan to go to publishers, people that do have paywalled content and try to get them to enable MPP? Is that the idea?

Totally. I think that would be awesome. And the other thing that I think is just happening already is that, again, envision that people these days don't have an attention span and people these days also don't want to click buttons around. They just want to tell their, again, their AI app to do things for them. I think what we've been seeing ourselves as developers is instead of me going to say Cloudflare, Vercel, and saying, hey, log in, add billing information, get API key, pull it back to my terminal, or go to ElevenLabs and say, hey, do me this text-to-speech thing, or any kind of like API-gated service that exists today. Instead, I can just say, hey, agent, here's $5, go and figure it out for me. And actually, he doesn't just do it in one service, it does it on many services. You can imagine a waterfall of queries You can tell it, hey, go search the web, pull all this data, translate it with this API, then voice to speech it, text to speech it with some other API, and then upload it onto a website and email me that link of the website, all one shot.

Yeah, I was going to ask more on the technical side. There's been a lot of protocols created, blockchains over the last decade. You've seen pretty much all of them I'm assuming this is your Mona Lisa, you know, taking all the different learnings from the different blockchains and trying to build the perfect product experience this time around. What kind of like technical decisions did you and the team make for this kind of use case?

Yeah, so the Tempo blockchain is itself optimized for payments, which means that it's really fast. It has very high throughput and very low latency. Our block times are less than a second, which means that you hit and before the spinner even finishes spinning, you have a confirmation back, which is what any normal person would expect from a payments experience. But in addition to these, the innovation that we've done on the MPP side is a feature called sessions, where normally you would need to, let's say you're talking to I don't know, let's say again, the OpenAI API. And if you were to do payments, if you were to pay for every query that you're doing, you would need to wait for the blockchain to confirm your transaction. You would need to pay a fee anytime you do your transaction. And that these days, maybe you can do 10,000 transactions per second or something. But really for the agentic web, and it does seem that everyone will be making a lot more API calls than they were doing before, We need to scale up to hundreds of thousands, millions, if not billions of payments per second. Just to put it— the analogy would be that instead of thinking of financial transactions, we should be thinking about API calls per second. So we developed this feature called Sessions, which lets you really go at API scale by bypassing the blockchain and thinking about it like opening a tab where you say, hey, I'm opening my tab. And then every time on your notebook on the side, you write what are the updates. And then only at the end when you want to settle, do you pay, which amortizes the cost of a very expensive otherwise operation. If you want to do 1,000 or 1 million API requests, just 2, the opening and the closing.

Yeah, makes sense. How are you dealing with the just general perception that There's always a risk that AIs hallucinate. And so maybe, you know, you don't want to have an— you don't want to give an agent the ability to go spend an unlimited amount of money. Maybe you want a transaction to be reversible. There's this joke about like giving OpenClaw your credit card. That seems irresponsible. But in some ways, I would maybe rather give credit card to an agent than a crypto wallet that if it sends it to the wrong address, it's gone forever. It's not really reversible. How are you thinking about handling those long-tail errors that can be catastrophic?

That's an awesome question because I absolutely have in the past lost money using systems that others had built where you just give your agent money and then the agent goes and spends it without realizing how much access it has. Yeah, we developed a feature called access keys, which basically is your standard card permissions. It's a bit more programmable, but you should think of it as limits, spend limits. Sure. And you can program that. So whenever you open a Tempo session, you can just authorize the agent and say you can only spend up to $10, or you can tell it, hey, you can only spend up to $10 an hour. Or $1 a second. So it lets you have any kind of granularity on the assets that you're accessing, the functions that you can call or the API calls you can make, and also on the frequency that these budgets can be applied. Yeah.

How do you think about getting the long tail of publishers onto some sort of agent payment rail?

Because hopefully they're already on Stripe, otherwise they've than living under a data center. Yes. I'm just thinking about the actual— That's like why when I saw the launch and heard about the project previously, I was like, OK, Stripe already has a long tail of merchants. If they can easily enable MPP, then you're able to bootstrap a network.

Yeah. I'm just thinking if you have a blog and you put AdWords on it, you got a stream of payments from Google, whenever you would show a page, they'd show an ad, they'd auction that ad, they would send you maybe 50% of what they made on that ad, and that would be paid monthly. And they were able to onboard sort of the long tail. If you— it feels like the, the opportunity here is like this many-to-many relationship, but I'm wondering about like, is that onboarding going to happen on Tempo directly? Is Is that going to be decentralized or would we expect the big labs like OpenAI, Anthropic to say, okay, now we have opt-in for anyone who wants to get paid when one of our LLMs, one of our agents hits your website and we will use Tempo as the payment rail. So there's a few questions here.

So I think what you just said is true and will happen. It's already happening. Anthropic or OpenAI or both already have a service where you plug in only to OpenAI and then you can just have a billing relationship with all of the integrated partners that they have. Sure. So that will happen for sure. On the Tempo side, the machine payments protocol, MPP, which is what we launched this morning, it is co-built with Stripe. We co-authored it and it's payments method agnostic, so every merchant that is on Stripe. And if you go to the Stripe docs, MPP is already there and it's already an option you can enable. Every merchant could turn that on and their API and their website could become MPP-enabled, which would cover a part of that tail, if not that part that you just said.

Sure. So is this a bet on model commoditization and local compute? Because if there's consolidation and the vast majority of LLM traffic is coming from like a single lab or 2 labs or 3 labs. Yeah, winner takes most. Like the Wall Street Journal does get scraped by OpenAI, but they have a deal between OpenAI, the corporation, and News Corp, the corporation. And they sat down in a boardroom and they said, hey, this is the value, like send me the check and pay, you know, via invoice. And it's a very different relationship. But for these long tail if the long tail with Google can pay a ton of small bloggers who set up their own websites on whatever, they can go to, you know, their own blog. But is this maybe a solution to like if models commoditize and you're seeing LM traffic come from all different sources, then you need your agent to go and make the payment?

I think for the publisher use case, I think that totally could be the case. The thing that we're the most, excited about the machine payments world and developers. Right now, there's just a tailwind right now where with AI, more and more people are building things. Everybody wants to make paid APIs or everybody wants to monetize their knowledge base, their API. And these are just not things that— it just means that the tail is just going to be so long that it's not something that I expect will be realistically captured by the labs in that way. Just because any random dev can spin up an API that could be very, very, very valuable. And I think that the lowest friction way to access and monetize that API will be via MPP on Tempo.

Yeah, I think that makes a ton of sense. I've run into this issue where I've gone to a particular LLM and said, I want you to take this video, convert it to an MP3, then transcribe it, then, you know, transform that. And it can do— and it's like, oh, I need a tool for that. Like, I was like, why don't you download the file? And you're like, no, you're literally telling me like open up QuickTime and turn it into an MP3. I was like, no, no, I want you to do that. And if you Google it, like, you know, MP4 or M4A to transcript, like there's a tool out there that does it. It's, I don't know if it has an API, it probably should. There probably is an API out there for a lot of these things.

It might have malware in it. Yeah, yeah, yeah.

A lot of them are like websites that have a million ads and stuff. It's way better if it just came back to me and was like, hey, that's gonna be 15 cents and I've already approved that. Or hey, it's gonna be $50 to do what you want because you asked me to go generate a whole bunch of AI video and crazy stuff and, you know, that was a big request. Is this in budget? And I can be like, yeah, absolutely. Absolutely. And this is right now supported.

There's the MPP Services Registry where if you literally go tell your agent, hey, build me an image, convert this to whatever you want, there are APIs already integrated which let you do all of these things without ever leaving your agent, without getting into this install your own API tool thing. So totally.

Yeah. Explain if I'm a developer and I'm using something like Codex and I want to effectively enable the agent to buy different services with MPP, what is the process for— what is the recommended process for actually giving it funds, everything on that side.

So this is the use case that we've been optimizing for. It is the exact use case because we're developers and we actually dogfood this experience every day. The process is extremely simple. You tell it install the Tempo skill. The Tempo skill is available on Tempo website. You just paste the link in your agent. And then once you tell it to install the Tempo skill, the Tempo skill bootstraps the whole process and says, Install the Tempo command line interface from your codex and then onboard the user. The user's browser opens, the user Face IDs, and then— or Touch ID if you're on your MacBook— and then the user just gets onboarded via Apple Pay if they have Apple Pay, if they have an Apple Card, or via another crypto if they have stablecoins. Or we right now we provide some referral codes to make onboarding easier for people.

Very cool.

And then you're in, then you're logged in and you can go crazy.

And then you could ask the agent like, add more funds, you know, I need to add more funds. Absolutely. Yeah, absolutely. And so if you hit it, if you hit some type of paywalled experience or paywalled API, it'll just tell you I need access to more funds and then you can just enable it. That's very cool. Absolutely. Amazing.

Well, thank you so much for taking the time to come chat with us. Yeah, very good. Congratulations on the launch. We'll talk to you soon. Daniel, have a good day. Have a good one. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital. At the New York Stock Exchange. And let me also tell you about Gusto, the unified platform for payroll benefits and HR built to evolve with modern small and medium-sized businesses. And without further ado, we have Matt Wong from Paradigm in the TV Greenhouse room. How are you doing, Matt?

What's going on? What's up, guys? How's it going?

Great to finally have you on the show.

Yeah, fantastic. Longtime listener, big fan. Appreciate that. Where should we start, Jordy? Should we do a little biographical stuff? Can we go back in time a little bit, get to how we got here?

Let's touch on Tempo, since we've been talking about Tempo. And then there's a bunch of other stuff we can talk about. But glad to see you're drinking a virtual Diet Coke with John.

Yeah, I mean, I guess my question for you on Tempo is, it feels like a big project. It feels like it launched very fast. Is this faster than expected? Like, what were the keys to actually getting to this point as quickly as you did?

I think it's perhaps one of the fastest project start to mainnet launch in crypto. And it certainly was top of mind for us to move really fast. I think the genesis of the project was around noticing the problems that existing infrastructure had serving all these stablecoin use cases. And we didn't want to wait around for many years.

But certainly, you would have seen the opportunity before August of last year, right? Like, it's not like you weren't thinking about AI or stablecoins or anything. So, so was that like somewhat of an intentional strategy of like, hey, we need to build a really strong thesis of, of how this market is going to evolve and we shouldn't just like rush in, launch something and then go through kind of wandering in the wilderness to then get to the point where you could launch something to mainnet.

You know, you're ascribing a lot of foresight Candidly, I mean, obviously we've been paying attention and investing and building in the crypto space for a long time. Candidly, it was kind of two sides coming together. On the Paradigm side, we were noticing the slowness of existing infrastructure. And then on the Stripe side, they've been doing a bunch of great work in stablecoins and were themselves experiencing a lot of the infrastructure challenges that existing chains had. So we had a conversation in April about maybe we should actually get together to build something. And then a week later, the entity was formed.

Are there any other big distribution nodes that need to fall into place besides having the network up and running and then the partnership with Stripe? Because it feels like that solves everything, but I'm curious to know what are the other next steps in the rollout broadly?

Yeah, I think there's probably two layers to think about, right? There's Tempo, the chain, which all the crypto infrastructure and exchange infrastructure needs to support. And we have a bunch of great infrastructure partners there. Then I think there's just the broader trend of stablecoins. Like, if you think about payments, I like to think about it as like 1,000 or 10,000 different markets. It's not a single market. What country does it originate from? Where does it go? The size of payment, whether it's a business, consumer, et cetera. And so some of those are much better served by existing infrastructure, and some of it's much worse served. And so stablecoins are actually particularly good at certain segments of that kind of big matrix. And that's where it's being adopted first— cross-border remittances, payments. And so part of the goal around Tempo is to kind of promote stablecoins more broadly, not just offer another chain.

How predictable has stablecoin adoption been to date? Like, if you go back 5 years ago, were you pretty confident that you'd see it more in emerging markets and then it would go, you know, eventually to business use cases for, you know, businesses that were operating are operating internationally, like, and then obviously trading use case, which has been dominant since the beginning.

Yeah, stablecoins have been around for 10 years, mostly dominated by trading. And I think the emerging market use case is definitely expected. I mean, the monetary rails and monetary units are much worse there. I will say we were probably surprised the last 18 months by how much it's taken off. I think even the most bullish people inside of crypto, maybe some predicted it, but all of us have been surprised by how much stablecoins are being used outside of crypto. And that's really the bullishness that drove us to get excited about launching something like Tempra.

And is the big driver— I mean, I've heard about the like Scale AI use case. I think SpaceX has mentioned a use case for stablecoins. There's a lot of like cross-border deals. Deal is listed heavily. Is that the current narrative, or are there other little offshoots popping up around adoption? There are probably 3 big buckets.

The first is the cross-border. And so that's where you have Deal, which serves so many different countries. And Shopify is thinking about similar things. And then you have the instant nature of stablecoins. So I mean, when you think about it, it's kind of crazy that the financial system shuts off during nights and weekends. Like, the internet is on 24/7. The financial system is off more than it's on. And so I think stablecoin potential is just to bring that into the 21st century, where it's on all the time.

Yeah, and the US dollar has insane network effects, obvious product market fit, and then you're taking a product that works, that's pretty great, and then it has some of its issues. And then you just wrap it in something that unlocks all this additional airtime.

And there is a trend in the '70s and '80s called eurodollars, which is sort of the offline tradfi equivalent of this, of dollar deposits outside the US. And that also started illegal and frowned upon, pretty small, and has now grown to north of $10 trillion in offshore deposits, and fully legitimized by the US system. And I think stablecoins are probably on that track as well. But back to your question, the instant payouts, instant payments, I think is underrated too. So it's not a cost advantage. It's more getting the money immediately and being able to spend it immediately. And if you think about a lot of these labor marketplaces, that's something that the supply side of those labor marketplaces really value.

How do you think the shape of the kind of issuer market evolves? You think it's, you know, right now it seems to be like a winner-take-most market, but certainly a number of companies are kind of promoting this idea of like there should be millions of different stablecoins. What's your view?

So I think there's probably— there's the how many stablecoins will there be that generally have some usage? And what will the distribution of volume and deposits look like? And I think the distribution is going to be very top-heavy just because network effects around a single unit of account and how liquidity works in DeFi and exchanges. On the other hand, I think a lot of enterprises that we're talking to own the interface that people are going to be using these stablecoins through. And you can think of them in that context maybe a little more like stored balances or gift cards, where Starbucks has a massive gift card program. And it's essentially bank-like, if you think about it in some ways. And so I think a lot of those types of financial services that are kind of hiding inside these consumer businesses will end up considering launching their own stablecoin.

Yeah, that makes sense.

How are you feeling about the regulatory environment? It felt like 2025 was a great year for just getting clarity around where crypto was going from a regulatory perspective. Has that trend continued in your mind? Are you happy with where things are? Are you optimistic for new regulation or less regulation or any changes? Just what's your read on the regulatory environment right now?

I'd say we're in the top 1% of universes that we could have hoped for. So things are going well. That's amazing. The Genius Act passed. That's been huge for stablecoin adoption. Obviously, stablecoins were growing even before that. But a lot of the enterprises we talked to, they weren't really comfortable participating in stablecoins until the regulatory clarity. The next stage, hopefully we get market structure, the clarity bill passed as well. There's some kind of last details to work through between the banking industry and crypto, but I'm feeling optimistic that we can hopefully sort through that.

Are you a believer in stablecoins to the degree that you'd put your own fund capital in stablecoins?

Well, yeah. Tempo would not be the best place to hold stablecoins if we're not comfortable with hot footing.

Yeah. Is there actually a future where stablecoins wind up showing up in venture capital?

Well, stablecoins, I think, will be— as a cash instrument, I think for sure. Yeah. Yeah, then you don't have to— I mean, you can do instant capital calls when you sign the term sheet at 9:00 PM at night. You don't have to wait till the next day. That's actually pretty bullish.

Take us back through some of the previous investment experience you've had. There was someone in the chat asking about your early Fund 1 investments, very heavy on Bitcoin and Ethereum. Can you tell that story? What that was like from a risk perspective, what actually played out there?

Yeah, when we launched Paradigm in 2018, I mean, a lot of our thesis was we knew that crypto would expand massively. And then there was this open question, should you invest in the core protocols like Bitcoin, Ethereum, or invest in startups? And obviously, in hindsight, the answer was both. Like, you could own Coinbase and Bitcoin. But I think it's a lot like the AI landscape now of this question of, do you own the model companies, or do you own the application layer on top? And so in 2018, Bitcoin and Ethereum were like OpenAI and Anthropic. And those are the things to own. And we were thinking much more bottoms up about what we wanted to own for 5 to 10 years, rather than like, is this too much in one investment, or it happens to be a publicly traded thing. Those are all details.

That's such an interesting comp because I feel like you can almost think of L2s and comp those to cheaper, smaller open-source models today where L2s have been great in some ways, but also kind of threaten the revenue stream of the underlying protocols at points.

Take me back to the ByteDance investment. Was there a similar thesis around should you own ByteDance or Facebook or something? And in reality, the bet was to own both. What went into your initial investment? Take me through the thought process for that one.

I wish I was that thoughtful back then and could claim some kind of genius. But I was just very lucky to meet the entrepreneur, Ying. On a trip. And, you know, he's just one of those people you meet that jumps off the page. And I'm sure you guys have had this experience. And I was going to give him money even if he was building a hot dog stand.

Sure, sure. What was it that stuck out? Just the conversation, the level of detail, the technical side, the EQ, IQ? Like, how were you evaluating him at that time?

Well, it was a weird conversation because He was— I don't really speak good Chinese. I can kind of stumble through ordering dim sum.

I don't speak good Chinese either.

And he didn't really speak good English. So it was through a translator. But he's one of those entrepreneurs that seems very thoughtful, academic, highly technical, but then in the revealed nature of his actions is just extremely hyper-aggressive. And I think a lot of Some Silicon Valley entrepreneurs look like that too, in a great way. Yeah, yeah.

No, I'm thinking of a number of CEOs we've talked to, founders we've talked to. I think one of the biggest ones is Patrick at Stripe. I was going to say, yeah, exactly, where they have one demeanor, one level of— they can communicate about the technical side of the business very thoughtfully, high EQ, but the level of ambition is just unlimited. And that just seems like such a rare combination. But when you find it, you got to go all in.

How should people think about Paradigm today?

That's a good question.

So we've spent a lot of time in the crypto space. And increasingly, as crypto goes mainstream, we're finding that distinction to be a little less meaningful. And there's a lot of interesting things happening in the world.

Yeah, like it was entirely right to be entirely focused on crypto in 2018. And then as the technology diffuses, you realize like, hey, there's a lot of other businesses. And then you also have to follow your own personal interests. And I feel like if you're logging off and then thinking about biotech or AI, At a certain point, I think the best investors follow their obsession.

Yeah, and don't get me wrong, I think crypto is still one of the most interesting frontiers. It's finally winning. It's getting mainstream. It's getting integrated everywhere. But AI and many other areas are also super interesting. And I think a paradigm today is a lot of the team we've assembled— I mean, you just spoke to our CTO. He found out what a balance sheet was somewhat recently, and is not a typical GP or investor at a fund, but everyone at the fund knows how to code and is in some kind of CLI tool all day. The team really embodies this hacker ethos that lives at the frontier, and we want to be spending time anywhere that is interesting at the frontier.

Is this the best time in history to be that hacker archetype?

I think so. It's like the joke that there's a billion dollars stuck in your laptop. You just need to press the right keys to extract it out. And I think that's never been more true with the leverage of intelligence kind of on tap. Yeah, no, I completely agree.

What a time to be alive. Well, thank you so much for taking the time to come chat with us.

Yeah, great to finally have you on. Have a great rest of your day. Yeah, excited to, uh, watch the rollout.

Yeah, we'll talk to you soon. Goodbye. See you guys. 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 CrowdStrike. Your business is AI, their business is securing it. CrowdStrike secures AI and stops breaches. Uh, Jordy, what else is in the timeline. Trevor Milton is out raising money for a new company. Trevor Milton, of course, was convicted and then pardoned for defrauding investors in his truck company Nikola. Now he's raising funds for a new jet that he claims will transform flying.

And there is a quote in the Journal here that we can pull out. He said from the Journal, now Milton has joined an exclusive group of post-pardon businesspeople seemingly anointed by Trump's favor. I hope they've got a group chat for the post-pardon mafia. I walk into— Trevor says, I walk into meetings now and I'll get high-fives from the most wealthy people in the world. He said they're like, welcome to the club. You can withstand the fire. We can trust you now.

Wow. That is a wild quote. Feldman has wasted little time embarking on a second act. He's now CEO of Cyberjet.

and he's unveiling plans for a new small jet that he says will have the highest speed and range and largest lavatory in the light jet category. Okay. Yeah, it's important. It's important.

Yeah. Yes. We want to be able to take a full shower.

Investor documents reviewed by the Journal said the goal is for the plane to be the first light jet to focus on AI flight. To get to market, Cyberjet plans to create an entirely new avionics system, likening the effort to Tesla's approach designing its infotainment and vehicle control system. The company also needs to secure more funding and suppliers and overcome regulatory hurdles. Yes, it's interesting. I mean, the concept of the jet is cool. This was an existing business that he acquired with an investor group, and then they're raising more money after that. The kind of interesting thing is I feel like Trevor made a name for himself for being a great showman, but not exactly the most competent and sort of like engineering-led CEO, right? Like everyone stuck in their mind, you have the video of like the truck rolling down the hill. And I don't want to buy a jet from that kind of archetype of CEO, right? You kind of want to buy a jet from—

Like a hacker, like a hackathon guy who just like whips it up.

A vibe coder. A vibe coder. No, no, no, no. You know, you want to buy a jet from a company— Someone with decades of experience, ideally. Yeah, a company with ideally with decades and decades of experience, you know, millions of flight hours because reliability is more important than anything when you're thousands of feet in the air.

Well, we'll make Tyler demo it first.

Yeah, I mean, I think— Step aboard. Yeah, and then the AI component is interesting. I mean, I think there's a number of light jets and even I think The Pilatus PC-12. I think the PC-12 has like an— a lot of the planes have the auto-land functionality. Yeah, they still require pilots. Yeah, I'm sure Big Pilot will figure out a way to make sure that like legally.

Well, you just get a normal, a normal jet with no autopilot and then you put a Tesla Optimus in the front seat.

Yeah, we got a pilot. Drive the plane. Take the wheel.

Take the wheel. Kelsey has a market here. Tesla Optimus released this year. 25.2% chance right now. Obviously it's declining a little bit as the year goes on because every day that it's not released, there's a lower chance. It was up at, where was it? It was up at like 50, above 50% last April, but it's since declined. And the market resolves to yes, if Tesla Optimus or another humanoid robot is on sale to the general public before December 31st. That feels tough because I have to imagine that what is on sale to the general public, like this, the demand's going to be so high, you would think that you would go with business consumers or someone that's at least, hey, why don't you come in and let's talk about your use case and let's see how we'll be pricing this.

Like, because it's more— Let's talk about the security use case. Quick. Oh yeah. Wait. Go for it.

Oh, I was just going to tell everyone about Restream. One live stream, 30+ destinations. If you want to multistream, go to restream.com.

Sorry. We got to talk about the security use case. Tech companies are now using robot dogs costing up to $300,000 apiece to guard AI data centers. Remember I was saying I just want to put a humanoid in my backyard, let them patrol around. I think it'll be a good deterrent. It's working for dogs. Let's pull up this video. Yeah. All right. That's freaky.

It can open a door.

This is the last thing you want to see when you're trying to— They're sort of Mixing footage here, right?

Because you wouldn't have a robot dog walking on two feet in— yeah, that's actual business attack mode. Like, yeah, it seems like it. Yeah, I guess that's just the news that the robot dogs are being deployed guarding some of those data centers. I don't know, seems, seems relevant.

You're not impressed because you think you could take it. Yeah, totally. You kind of know you can take it.

Yeah, yeah, dropkick that thing, no problem.

I still got it. Taylor Lorenz highlighting from this Vanity Fair profile, which we'll get into tomorrow, but in the profile, Dario Amadei never actually gave me an interview, stiffing me after months of planning. So I created a version of Dario Amadei using his own machine. I fed Claude several published interviews, including everything Amadei said at Davos, plus the contents of his two books of essays, and told, which I'm sure Claude already has access to. Yeah, he told Claude to simulate the interview using variations on real quotes and to make it like a scene from Raymond Chandler's The Big Sleep. It took Claude less than 3 minutes. You probably couldn't tell the difference, and maybe there isn't one.

Okay, there was, uh, there was like some hot mic moment. Is that real? I, I, if everything's— if everything—

I still have to read through it.

Yeah, yeah, we, yeah, we got to properly read through this. I'm sure there's some interesting stuff. It does seem like there's a few interviews. I saw some quotes about like Trent saying that he doesn't wear sunscreen anymore, which I've been advocating for, you know, getting on the Trent and D-Ball Anavar because AI will cure all the downsides potentially. It's sort of a twist on that. But at this point, now that I've seen the screenshot, I'm kind of wondering like, okay, like what in here is real? Is it all going to be revealed at the end? It's a very long piece. But we'll give it a read and see what's going on there.

And last but not least, we can briefly cover introspection gate. Yeah, yeah. Mark Hendryson shared something from Marcus Aurelius' Meditations. "Stop talking about what the good man is like and just be one." And Rune says, "An entire book where the guy is introspecting." I think that's the joke, right?

Yeah. Mark Hendryson's like—

Yeah, I think— Seeing through, I think we saw, we got what Mark was saying, even though a lot of people have— he's leaned into the chaos and seems to just be having a good time. But he was basically saying, just joking about it, stop ruminating. Stop ruminating.

I mean, it's a transposition of the "you can just do things" idea. It's like, actually go and do the thing. Jeremy Giffen has another post about this, where it's like people are rehearsing for the fight that never comes. And that is like a different formulation of like the don't— it's like, don't spend your entire life just thinking about what you're meant to do with your life. Like, go do something, get some real experiences. There's probably room for introspection, and then go back and forth and watch the game tape, monitor your KPIs, but also just think about what brings you joy and what what you can do in the world that's beneficial. What's your ikigai? What are you good at? What does the market want from you? What can you make money doing? That type of stuff. Like, these are reasonable, but should that be 99% of your time spent? Probably not, you know. Never ruminate.

Never ruminate. Always spect, never ruminate. Always spect.

Okay.

Spect max.

Well, let me tell you about Plaid. Applied 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.

Bill Gurley says, I fear that AI has decimated the traditional email inbox as we know it. Too many personalized sloppy mails slip through the spam filter. Hope someone builds a better mousetrap. This one is cooked in its current form. Great use of the word cooked. Interesting. Nikita was saying February 11th prediction in less than 90 days, all channels that we we thought were safe from spam and automation will be so flooded that they will no longer be usable in any functional sense. iMessage, phone calls, Gmail. Email inbox certainly feels that way. Nikita says spam laws will need to be rewritten soon. Machines should be prohibited from communicating with humans unprompted. So this will actually make your job just actually just hitting like the button because like, no, a machine didn't send that, a human did. The human hit the button and you're gonna have to press press a button like, you know, 10,000 times a day if you're BDR sending sales emails. So anyways, it has been an honor to do the show. We have one last gong today.

Kendra Barnett was just served her first OpenAI ad impression.

It's amazing news for The Wall Street Journal, the benchmark in business coverage since 1889.

I can't imagine a better ad to get served. We love The Wall Street Journal here and we love ads. So Congratulations to Kendra for being served that first OpenAI ad impression.

We hope you all have the best afternoon, evening of your entire life, and we look forward to being back with you tomorrow. We'll see you tomorrow.

Goodbye. Flashbang.