Bending Spoons Buys Airtable, Snap Rips, Ads in BMW | Grace Li, Samir Kaji, John Quinn, Nikhil Reddy, Art Levy & Russell Kaplan, Brendan Carr

Let me tell you, you're watching TVPN. You're watching. You're watching TVPN. It's Tuesday, August 4th, 2026. We are live from the TVPN UltraDome, the temple of technology, the fortress of finance, the capital of capital.

It's August 4th.

You can turn off the music. Let me tell you about Ramp. Time is money. Save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place. We got to work on that intro. But we've been having fun. Making intro songs, mixing it up. We got John Quinn, the most feared lawyer, the lawyer the lawyers all fear, coming on the show live in person in the TVP and Ultradome. You know him from Quinn Emanuel. He's the founder, a legend, a legal legend. When John Quinn stands, everyone kneels. That's what they say about him.

That's right.

No, we're very excited for the show today. Lots of guests coming on. Lots of news. Big deal, the big one that's tearing up the timeline right now is that Airtable, founded in 2012 and once valued at $11.7 billion, is getting acquired by Bending Spoons, founded in 2013, at 2.7 times ARR. Once hot startup, now an unfortunate victim of the SaaS bust, says DD Doss. It raised $1.4 billion, but only to be sold for $1.285 billion enterprise value, because they had almost $1 billion in cash on the balance sheet. And so the total equity value was $2.25 billion equity value, just clearing the preference stack. So early employees, founders probably got something. Later investors probably got 1x their money back, but probably had it tied up for a few years, so not a good outcome, but there's some interesting silver linings here. Obviously, it could be good for Bending Spoons if they got a good deal and they turn it into a mammoth cash machine. Also, there's some nuance to where different pieces of the business are going, because they're sort of dividing it up. But of course, Bending Spoons is a public company now. And if you want to back Bending Spoons, head over to public.com.

Or short it.

Invest in Fritos.

We don't give financial advice. Do whatever you want.

Stocks, options, bonds, crypto, treasuries, and more with great customer service. Go long, go short. Whatever you do, just don't use too much leverage. That can be risky. Matt Levine has a great piece today.

And actually, it really sort of exemplifies the current moment where a company can sell for over $1 billion and everyone's like, wow, that's right.

Yeah, yeah, yeah.

But and in a wake-up call to the to the many maybe younger companies that are at a lower revenue run rates that are raising at much higher valuations and basically signaling to them like you've got some many, many years of compounding to do.

Yeah, it's interesting. There's a SaaSpocalypse narrative, which is like these companies are going away, software won't exist, you'll just prompt it. And I think there's a lot more nuance to it than that. But one thing that,— it does feel like the underwriting, the financial trajectory of these single-point solution SaaS products, single player, somewhat sticky, maybe not that sticky, maybe replaceable— it's not that they're going to zero. Bending Spoons wouldn't be buying it if everyone was churning and it was going to be a zero. But at the same time, you can't underwrite it—

It's growing 20%.

Yeah, yeah. But you can't underwrite it at 40x revenue, 100x revenue anymore. And it feels a little bit like what what happened with D2C e-commerce, honestly. There was a moment where e-commerce brands were venture-backable, and you could underwrite them, or they were being underwritten similarly to venture startups that had true moats, true compounding advantages. They would get the same multiple as a SpaceX or an AI company or a social media company. And that never really made sense. It was sort of a— just a weird quirk in the system for a couple of years. And then go forward a few years when there were some pullbacks, some of the IPOs went out, they didn't do that well, they traded down. And all of a sudden, it was like, OK, well, if we're doing, in my case, like food on the internet, and we're going to VC back a food company, like, has Nestlé been disrupted? Like, no. And that would— or Unilever. Is Unilever trading down like crazy because they're facing so much pressure? And that's usually what happens when there truly is disruptive innovation. Like you see this with, I saw some crazy post about how people were bearish on Starlink for a while and the company that they were competing with just went bankrupt. And you see this with, you know, social media came out and yes, like the newspaper and the internet, actually the newspaper stocks did actually trade down. That never happened in e-commerce, D2C e-commerce, any of that. And we're now in this new regime. So I think that there's actually a pretty safe path if you just build the business if you're saying, look, I am in this SaaS industry, it is going to be more competitive going forward, but I'm setting myself up to have a reasonable multiple so that at every point, if I'm trading at 3 times ARR, I'm happy because the cap table is set up for that, right?

Yeah, one thing I will say is I started using Airtable shortly after it— or it was 2015, right? So I probably started using it a few years in. But the product, the way that I used the product back then for my first business, I would 100% just vibe code a solution today because I was using very basic dashboard functionality. I wanted dashboards that we could use internally, share externally. Now it'd be very quite easy to just do all of that in Codex or your favorite agent. So I do think that business is very much under threat over the long run, but it still has sort of compounding that it can do, just given how deep it is into the Fortune 500 and the long tail of small businesses.

I'd be very interested to know what is new user, new logo growth like versus just expansion within an organization. Because if you have some company that's already sort of running on Airtable, they're growing, so they're adding seats, they're adding functionality because they're sort of bought in and they're not going to rip it out. What does that growth rate look like versus, you know, new companies actually going and signing up and saying like, yes, this is the best tool for the job because you do get a lot built for free up and like it's not like they don't have access to AI agents that can improve their systems. Like you might be doing a lot of maintenance on your Vibe-coded solution, but at the same time, a lot of people, especially in smaller organizations, were using this as like just one small dashboarding tool, one small database that is replicable. So interesting to see where it goes. And yeah, just like an interesting data point in the SaaSpocalypse, late-stage growth. Is it a zombie corn? What term are you using for it? Jared Sleeper has some more thoughts. $480 million ARR growing 20%. It's a unicorn exit. Many very satisfied customers, including Jared Sleeper, for years. Fantastic run. 'Never underestimate how much VCs love products that make, quote, everyone a builder. The wrinkle is that one-player products are high churn.' Was Airtable— I feel like the whole pitch for Airtable was that it was multiplayer, but maybe that wasn't the way that people were actually using it that often. A lot of people would just be like, 'Oh yeah, that guy on the team is using Airtable for that thing.' But I mean, certainly from very early on the pitch.

I think by one player, he means that you can— it's more like product-led, like one person at a company. Because I've only used it in a team capacity, right? But one player can sign up, start using it, invite other people to the team.

Yeah. But it's an advantage because you're bottom-up. You can just ramp into a different company so quickly. But at the same time, if you're not like going through the CFO and being like, there is a mandate that we're using the ERP system.

Selling 200 seats at once.

Exactly. Yeah, it is a little different. 3, his third point. Gross retention remains the single greatest predictive variable of terminal value for any business that doesn't have a scale effect or network effect. It dictates somewhat mathematically what folks like Bending Spoons will pay. Great to see employees get liquid, but perhaps sad that some are learning about liquidation preferences, SaaS multiples, to their disappointment. So of course, if you had mentally marked your stock at $11 billion, But then this deal happens. The investors that put in that money at $11 billion, they're going to get their money back first and then you're going to have to fight for whatever's left over, the scraps. Silicon Valley can do a better job talking to employees about what stock is worth in various scenarios, but no one is incentivized to do it. Well, I mean, you can also just look it up. There's a whole bunch of blog posts about it. You can ask ChatGPT and get a whole deep dive on, run this whole scenario and play out every possible option. But yes, I agree. No one likes to, you know, oh, we're closing a big candidate. Let's tell them about what's going to happen if Bending Spoons comes in.

Let's tell them what happens if we get our spoon bent.

Bending Spoons is emerging as the Constellation Software of the prosumer high churn, higher churn SaaS. It's a fascinating turn of events and a good thing for pref stacks everywhere. Just because there's more liquidity for systems?

Well, it's a buyer of last resort.

Yeah, whereas—

Like no founder setting out, hey, I want to be— I want to go on this generational run and then get my spoon bent. But it's still great because it allows— again, it is real liquidity. This is a win. I think bending spoons will— obviously end up, I would say, like right-sizing the company. They're not gonna continue to run it the same way that it's been run. But at least it allows everyone to get out and go on to do new things.

Yeah. In an interesting twist, speaking of getting out, Shiel Monat shares this about the Airtable acquisition. They spun out their AI business, HyperAgent, prior to the acquisition. So the company lives on and likely some cash with it probably can be recapitalized. Seems, seems like a fantastic outcome. Shed the old business and focus on AI. So I would be very interested to see who is going to be working at Hyperagent. Is this something where the founders didn't want to exit the business entirely and then just start from scratch? They wanted to take a whole bunch of the team with them and they wanted a bunch of the resources and learnings. And so they were able to package everything up. And this will be something that's very founder-led. Or was this just, OK, that is a more expensive piece of the business? Bending Spoons, it's earlier in its ramp. Maybe it's working, maybe it's not, but that's not the business that Bending Spoons is in. And so this is more like a Bending Spoons-driven spin?

Look, we've had both founders on the show. Bending Spoons is like, hey, let's take a great product and just basically right-size the team, run it efficiently, and we're not— trying to get— they're not ever in the business of trying to get growth back to 200%, 300% a year. Howie's been on the show. He talked a bit about HyperAgent. That's a product that if they execute well can grow 10x year over year, right? And so I could imagine, again, leaving some of the team to just continue running Airtable indefinitely. And then again, like you said, taking some of the talent out and saying, let's take another big swing.

Yeah, yeah, it's a fun twist on like post-acquisition. What do you do in this scenario? Like the team doesn't necessarily need to stay on. That's not the Bed Bath Beyond thesis. There's not some crazy earnout probably for most of the team members. And so you have the opportunity to go do something new. And if you're already all set up, it's like, yeah, just put everyone together before the deal happens.

Yeah, the other thing that's notable is Notion. I think just did a tender.

They just did a tender or they did a tender a while ago. I thought that—

I thought at some point, at some point within the last year, really at $11-ish billion, they're at $600 million.

No, no, no.

You're right. This was January of 2026.

Yeah, I saw this quoted as like, oh, and do you have any idea how quickly they're growing relative to Airtable? I imagine they're growing significantly faster.

So the news was that on January 26th, 2026, Notion which very much feels like, you know, easily comp to Airtable. Airtable was like modern Google Sheets. Notion was modern Google Docs. Of course, there's a ton of overlap in the capabilities and there's a lot more to it. But Notion sort of feels like a similar shaped business. But GIC Sequoia and Index purchased Notion shares in a private tender offer January 26th of this year. The total tender was $270 million, an $11 billion valuation, which was the last valuation, I believe. They say we waived a 1-year vesting cliff on options for current employees. So I think everyone got to participate. There's a whole bunch of returning investors. In 2025, Notion's growth rate accelerated on the back of AI adoption within our platform. As of last year, more than 50% of the ARR came from AI-enabled customers. The growth over the last 12 months is happening globally. They don't share an actual growth rate, but I don't know. Everyone must have been optimistic at the time because they did the deal at the full $11 billion. So anyway, Bryce Roberts is laughing at the VCs, I guess, who were able to play the markup game and were able to raise one, maybe two funds on the Airtable markups alone. I'm sure there are some, but I mean, most of the funds that were in Airtable were in a bunch of stuff. So I don't know how true that is. Maybe the really early stage people. I don't know.

Yeah, Airtable did enough rounds back to back to back that pretty much every fund is in it. And they also have a bunch of other investments that are high performing during that era. I don't think he's making fun. He's just pointing out the—

Silverlake? I don't know. Liquidity is laughing about the fact that Bending Spoons has a very soft, friendly brand. Bending Spoons branding, very simple. And it just sounds like such a simple name. It's not like Cerberus Capital. And yet, when they negotiate, they're presumably difficult to negotiate with. I don't know. I mean, in a certain situation, you can be very friendly and amicable and just sort of say like, yeah, we're a buyer at 2.7 ARR. We're not going to screw you over here. We're just going to give you— this is a fair price. Take it or leave it. And you can take it. You don't necessarily need to be this crazy werewolf of a shark.

Yeah, there's just not that many buyers for a company like Airtable.

Yeah, and it's a—

just sort of in this economy, a slow-growing, enterprise software business, right? There are not that many buyers. If Airtable had wanted to sell in 2020, I'm sure they would have gotten meaningfully more. They probably did have offers at different points. You could imagine Airtable, you know, ending up at a, you know, a Salesforce as an example, right? But now Luca is, Saying the buyer of last resort is maybe unfair.

Yeah, just like a value-based buyer.

Yeah.

I don't know.

Yeah.

I mean, I guess it's like, yeah, the first resort would be like the hyperscalers, but—

Yeah, ideally, you never sell.

Yeah.

You go public and just keep compounding. Sacabase would be a strategic, right?

And it just feels like the enterprise tooling market is pretty mature amongst the hyperscalers. Like Microsoft's not gonna replace Excel with this. Google's not gonna replace Google Sheets. Meta, Amazon, they don't really want to get into this particular thing, a whole email suite, like the Office productivity suite. And so you get to this price and it's like IPO or sell. And so they chose sell. Anyway, let me tell you about Shopify. Shopify is the commerce platform that grows with your business, lets you sell in seconds online, and in-store, on mobile, on social, on marketplaces, and now with AI agents. Snap earnings happened and Evan Spiegel was on CNBC talking to Sorkin. And it's a very interesting storyline because people are, people are obsessed with the SPACs. I mean, the job, job completed in terms of like getting more attention for the company. But it's been sort of a weight around the company because the question is like, okay, is, is everyone going to be buying $2,200 smart glasses from you anytime soon when that's not really the story? The story is actually like the revenue growth, the profit, the operating leverage that's coming back into the business. So, I mean, overall it was a good quarter for Snap, clear beat. Revenue was up 19% year over year with profitability and cash generation rising even faster. So that's that operating leverage. They're growing revenue faster than their growing costs. And so they're increasing their margins. Advertising revenue grew 9% to $1.3 billion in the quarter. But the really interesting number that was surprising to me was they're making $316 million a quarter in subscriptions and paid services, up 85%. So just subscribing and being a paid power user. It's only 3% of the user base, something like that. That's, that's now a billion dollar line of business, obviously very high margin, working really plays to the strengths of what Snap offers. And usage is up globally, but they're losing a step in Western markets. So the North America, I think, fell 7% and Europe fell 2%. And so that's not great for long-term ad monetization because you want to be on the the richer countries generally, but still revenue is accelerating. They are monetizing better. And that's probably an AI story. It's just like a boring AI story because it's like the ad recommendation system got a little bit better. Nobody really cares. People want to focus on specs, but they're $2,200 and they're bulkier than competing smart glasses.

Well, it felt like in that interview Spiegel was talking about specs. Yeah, but he wasn't wearing them. And so, yeah, does he need to wear them? I think he kind of needs to wear them all the time.

I mean, you're going to hold Apple to that. Tim Cook's got to wear Vision Pro everywhere he goes. That'd be hilarious.

No, but, but the Vision Pro is not augmented reality.

It's augmented reality. It's meant to be worn 24/7. If you're a serious person. Let's actually pull up—

You should do a full show if you love—

I could.

Apple Vision Pro so much.

Not be a problem.

Not do the full show. OK. Do the full show.

OK.

This Friday.

Book it.

We'll test Reddit with Tyler. So here's an interesting tidbit. Specs, it feels like, oh my god, they're spending so much money. It's like so crazy. I think the rough estimate is like $300 million a year, which is a lot, but it's only 5% of their cost. It's like 20% of their overall R&D budget, maybe, something like that. These are very rough numbers, but— It's not like if they just spun out SPACs or cut SPACs and they were just like, we're not doing anything there, all of a sudden the business is wildly profitable. It's not that big of a stone around the neck. It's more just like a distraction, a question. And it's also not great that on the earnings call, they asked how many preorders have you sold and he sort of danced around it and didn't really give a straight answer on that. If it was good, you'd probably be like, yeah, we sold a lot of these things. Yeah, yeah, yeah, yeah.

Yeah. And even, even Meta selling what I think now is millions of devices.

It's pretty good.

Is not exactly like shouting. They don't want people to necessarily focus on that.

Totally.

Even though it is a proof point that there is a market.

Totally, totally.

For camera glasses, basically.

Yeah.

The crazier thing with Snap is that this is a business, even after it's traded up 13% today, the business is at a $6 billion a year run rate and it's trading at under $10 billion, right?

Yeah.

So which just goes to show how even with some momentum across the business, people just do not have faith in the company.

Okay, let's play a little bit of Evan Spiegel on CNBC. I want to hear him talk.

Snap reporting Q2 results after the bell on Monday. They scored an earnings beat with $1.6 billion in revenue. It's 19% higher than just one year ago. And joining us right now for more, Snap CEO Evan Spiegel. He did not bring his glasses on the set, but we're going to talk about those glasses in just a minute because I think that's been a big part of the story. But you beat across the board, not just, by the way, on the revenue piece, but on the margin piece, which is what I think the market was actually looking for.

Yeah, well, first of all, thanks so much for having me on, Andrew. Such a Beautiful morning here in Aspen. It was a great quarter for Snap. And I think what folks are seeing is that the free cash flow in the business is really starting to inflect, which is allowing us to offset dilution, to strengthen our balance sheet, and of course, to continue investing in the future, which is so important to us.

Okay.

Let's look forward to the future.

2:30 pieces of that story. I mentioned glasses. We'll get there in a second. I think glasses—

I want to hear him talk about glasses.

About $2,200 right now. $2,100 is what they're going to come in at.

$1,995.

Okay. So the question I keep $5 under. How quickly those can come to market in a way, at a, at a way and at a price point that people buy them, you know, en masse. And how do you think about the competition coming from whatever you think Apple is ultimately going to create, whatever you think Google is working on and whatever you think Meta is going to do next?

Well, we've been working for more than 12 years to reinvent the computer and make it feel overnight. I think people are spending more than 7 hours staring at screens. Specs represent the opportunity to bring computing into the world, to make it a shared experience, and ultimately to help bring all the productivity gains we saw in desktop computing and laptop computing to the real world and to real-world jobs, which is the vast majority of jobs, about 60% of jobs. So I think this computing transformation is incredibly exciting. We're certainly the leader in the space. I think specs represent a totally new category, right? If you look at the landscape today, you have very bulky but capable headsets. And then you have very limited but lightweight glasses and specs represent the capability of some of these VR headsets. And in terms of the immersiveness and the ability to really have a full workstation experience, but with the wearability of some of these lighter weight glasses.

But do you say to yourself, Apple's going to come and do the same thing and Meta is going to go? I mean, so how do you think about that given the cost of putting this all out there and the amount of money that some of these big companies can actually throw at this?

Yeah, well, I think, you know, as we look at the history of innovation, I actually think one of the things that helps power innovation are constraints.

Right.

And one of the things that makes Snap so unique is that we've been so laser focused on specs for such a long period of time. So I think this focus, our history of innovation, you know, and our first mover advantage in the space—

it feels like it should be a different company. If you want to be laser focused, it feels like the benefit would be like, do the Elon thing, start a separate company. That is laser-focused on it. The investors in that company are laser-focused on that. The employees—

Yeah, have a distribution deal with Snap.

Yeah, just have this, like, you know, get— we're good on this video. Just get Snap to a really polished, oiled machine. You own all the equity. You have the founder control. It's your financial backstop. Your credibility to actually go and truly laser focus because it's weird to be laser focused on a thing that's not your core business. And then everyone's just constantly asking you like, we'd love for you to laser focus on the main business that's making billions of dollars every year.

Seems pretty solid.

Yeah, it's like making $6 billion. Well, what do you think, Tyler?

Yeah, I mean, earlier this year they did spin off the AI video company. Yeah. And that seems like much more related to the main like, you know, Snapchat app that would have been glasses, right?

What is the AI video company?

I think it's called Dotmo. It was like their internal generative video team.

Interesting. Yeah, I feel like having—

Yeah, it also— he was talking about enterprise use cases for the glasses. I mean, like workplace use cases. It's so hard to imagine. It's— as an independent company, you make a really great device. I can imagine that company having its own go-to-market motion saying, hey, Amazon, we want you to use our devices across your workforce and maybe they do a pilot. But trying to sell in like Snapchat glasses into the enterprise feels like just going to be a tough sell.

I also wonder, I mean, I don't know if this would actually be a good strategy for such a big company, but I would be interested to see if there's an opportunity just for another physical product product that could actually get to scale that's way cheaper. I'm thinking like some sort of like disposable camera that doesn't have a screen on it. You take the photos and then they upload to Snap in the app and it's like this special way of like disconnecting more on theme but way cheaper because it's just like a basic camera in a shell or something. Or going into like, you know, a ring or a wristband. Like there's been so many companies that have figured figured out how to get a wearable or a hardware device actually in the hands of millions of consumers. Like, maybe do that first and then grow from there, as opposed to taking like the straight shot. I mean, it's a straight shot to full immersive augmented reality, the most challenging thing, which is, which is just, it's just a really, really tall order.

Um, yeah.

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. Well, OpenAI is firing back at Apple. A little, little glazy, though. They say Apple is one of the greatest companies of all time. What a funny way to kick off a blog post firing back at a lawsuit. Apple just cooked OpenAI. Just—

well, they say, and built a reputation for obsessing over the smallest details.

True.

And then they go on to to say maybe there were some details that were missed.

Yes. So what were the key things that they said? They accused— Apple accused OpenAI of ignoring them. Turns out the lawyers emailed the wrong Asian guy because two Asian last names look similar. They claimed you discussed the allegations with OpenAI's general counsel. They admitted that that conversation never happened. Accused an ex-employee of improperly accessing files. They forgot to mention Apple employees were allegedly asking him to access those same files after he left. So there seemed to be some scenario where like the guy had left, but he had so much internal knowledge that someone at Apple was just like, hey, like, can you remind me where this thing is or something?

Yeah. When reading through the initial complaint, Apple made it seem like the employee or the former employee was just running wild through his old laptop.

Yep.

He had sent a message to someone at Apple saying something to the effect, I'm paraphrasing, but I still have access to my computer, LOL. But the reason that in the text messages that they shared, you can imagine the reason that he was there and sending that message. Why would he send the message if he was doing something that he was that he, that, that, that, that was wrong, that he would feel guilty of. Right. He was helping his former teammates.

Weird.

And they say there's 10 other people I could ask you, but you're the smartest or something like that.

Yeah.

And, and so when you have all this context, the story looks quite a lot different.

Yeah. They're sharing a lot of text messages. You can go read the blog post, but there's like full back and forth iMessages messages between Chang Liu and Apple employees, where Chang, whose last day at Apple was January 22nd, 2026, is being asked by his former colleagues to help them locate files and information to assist them with their Apple work. Note, other individual names and Apple confidential information has been redacted.

So, uh, certainly another, another wild, uh, yeah, I'm still, I'm still just very surprised that, that Apple during the middle of a, you know, a year-long talent raid, uh, would know that someone significant had quit and that would not go through the process of actually taking back their laptop and making sure that, um, you know, the separation was, um, you know, really finalized. Yeah, people have been saying stories how historically, you know, maybe call it 15 years ago, if you quit Apple, someone would show up to your house immediately and take back any prototypes that you may have had or anything, anything of the sort.

Yeah, I'm excited for prediction markets to get on this. Honestly, it was very helpful during the, the Elon Musk OpenAI case. Currently, Kalshee has a few here. Will the OpenAI Johnny Ive device have a screen? No is at 82%. When will OpenAI release Astra? There's some dates here. What else are they? Will OpenAI increase the cost of ChatGPT? No at 87%. No market on the Apple lawsuit because it's, it's still very early. There's not even like a court date yet, but we'll keep tracking it because it's an interesting story. Uh, let's move on to the next story. But first, let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web app servers, databases, and more while Railway automatically takes care of scaling, monitoring, and security. Um, so, uh, is BMW forcing owners to watch a Spider-Man: A Brand New Day promo when they start their cars up? That's the question. A lot of people We're upset about this. Jordi Hays this morning is like, this is awful. There's some nuance here. So BMW owners are debating the company's latest in-car promotion after videos spread across X showing a Spider-Man: Brand New Day animation appearing on their vehicle's infotainment screen at startup.

And the reason why this fake news triggered me was back in the day in college, I was getting a Kindle. Yeah. And on the checkout page it was like, do you want to save $10 and get the ad-supported Kindle? And as a college student, I was like, hmm, saving $10 sounds pretty nice. And then I had to live with a Kindle that would sit on my bedside table just blaring ads in my face constantly. And it wasn't like I would have actually appreciated ads. Can you just flip it upside down? True. Wow. One simple trick.

Jordi Hays' one simple trick.

One simple trick. Um, and, but it wouldn't be ads for like Spider-Man. It would be ads for random books.

Books, right?

Yeah, books.

That, uh, and the targeting wasn't good. So again, people don't like ads that—

If the target was good—

Um, anyways, what happened here, John?

Okay, so venture capitalist Shiel Monat wrote, quote, when you start a BMW, it shows you an ad for Spider-Man. Really cheapens the BMW, in my opinion. A16Z partner Josh Elman added, of all the brands I thought I might— I thought might bombard you with in-car ads on screen right when you start the engine, I had BMW pretty dang low on that list. This seems the opposite of luxury and performance. Is BMW a luxury brand? I thought they were premium.

I— yeah, I think it's a premium brand.

Premium Even Paul Graham weighed in, though, reposting Monat's video and saying, I'm never buying a BMW. He's never buying a BMW. What about an old BMW? They can't show you ads in an E39.

Wait, but what actually happened? You have to opt into this.

Community Notes added important context. According to OneNote, the Spider-Man promotion is an optional startup banner available on compatible BMWs from July 27th through August 10th, 2026. The banner does not automatically play a full-screen ad. Instead, owners must tap in to launch a themed animation featuring music and synchronized vehicle lighting. BMW has offered similar limited-time startup experiences in the past, including holiday-themed animations. Jordi Hays.

Very, very, very different. You can just see the pop-up there. Surprise! Spider-Man just dropped into your BMW. And that's an ad.

That's if you press the button. You have to actually turn it on. It's not—

Yes, yes, but I'm saying there's a banner ad.

Oh, there is a banner ad.

It's just a banner. Look at the video. Start it over. Start it over.

Let's see. Let's be the judge of this.

Look at this banner. It's a banner.

Oh, okay.

And then you click— If you click the banner, then it plays an ad. So it's a popup that you click.

Okay.

And it's like, wait, what is this popup? Why does my car have a popup?

Yeah. It should probably pop in the—

And then it plays an ad.

Into a setting somewhere, there should be an icon that's like promotions or themes.

That didn't look tucked in anywhere. That was the main— That was the main—

Because I'm fine if there's a theme and you're like, I want red lighting in my car. I want blue lighting. I want Spider-Man theme. I want Christmas theme. Go over that.

They're literally saying surprise, a Spider-Man ad. And they're not even— Like they should just say, hey, we have an ad for Spider-Man here. If you'd like one.

Okay, this is right on the line.

But this, this is, this is, this crosses my line. This crosses your line because you're like, what does Spider-Man have to do? What does, what does Spider-Man have to do with my car? And then you click it and you get an ad for Spider-Man.

But I just think of the consumer surplus. This is making BMWs cheaper. This is making, this is bringing, bringing BMWs.

Let's look at the price of inflation.

Yeah, also not necessarily the best deal.

Yeah, let's look at BMW inflation.

Yeah.

Um, anyways, yeah, really, really bad move. I wonder how much they actually made.

I think you're just judging it because it's— it's— I think you would feel completely different, completely differently, if you— if you saw a banner and it was Nitro Circus and it would just take over your car, or if it was, uh, just a full takeover of Sam Sulek at the— at the Arnold Classic. Something like that. That would be much better received. Uh, it's not, it's not a pin—

surprise, Spider-Man trademark just dropped into your BMW.

Okay, the ultimate driving machine. Yeah, um, not, not super well executed, but BMW's been going back and forth with stuff like this pretty consistently. Like, aren't they the ones that have the subscription to the heated seats? Isn't that BMW? You know what I'm talking about? So— Yeah, I remember something like that. They put the wires to heat the seats in every car and then if you want to use it, they charge you monthly. And this is just like a— it's basically just a nerd snipe or like a— I don't know, like a treasure hunt for hackers because if you're a DIY hacker and you can get into that and unlock it, jailbreak it. Uh, you just get— well, not free because you've already paid for it, but you don't have to pay the subscription.

Uh, yeah, Techno Chief says, wait, I thought we loved ads here. I do love ads, that's true, but I think buying— unless, unless BMW— I think, I think if they gave the option at purchase saying, do you want to buy the ad-supported manual transmission BMW M2 Or do you want the ad-free premium version? And there was a bit of a price difference.

Yeah.

Then it just feels, it feels wrong. They didn't do that.

Sort of like a Ferrari tailor-made program. Oh, like what stitching did you get? Oh, did you go ad-free or ad-supported? Like, oh, you went, you went manual.

Cool.

Yeah, I went manual but ad-supported.

Yeah.

Manual ad-supported. Yeah, I got a manual SP3. Ads support. I mean, the ads in an SB3 driver are probably very hard to reach, so pretty valuable. I would love ads there. But it's got to be ads for the most elite products. It's got to be you're getting your SB3 and it shows you an ad for AmanGiri or something. Spider-Man's doing well though, so the ads are working. So, you know, maybe the BMW owners showed up in droves because the new Spider-Man movie scores Hollywood's second biggest debut ever. That's actually huge. The Sony film grossed $932 million through Sunday, topped only by 2019's Avengers: Endgame. That is massive. Did you realize that this movie was making that much money?

I mean, if you had asked me you an hour ago, is there a new Spider-Man movie coming out? I would say haven't heard of anything.

Yeah, I haven't seen that much energy about it. I've been aware of it, but I have not— I've not seen like a massive takeover. But also, I'm not driving it out of support of BMW. So maybe that's the problem. Spider-Man spun a massive weekend for movie theaters, superhero fatigue and giving a major boost to Hollywood's summer block box office. Sony Pictures' Spider-Man: A Brand New Day opened to an estimated—

we got some good ideas here in the chat.

Yeah, what we got?

So ad-supported cars that get you free self-driving where they— the, the, the ads take over the whole entertainment system in the car. So you're sitting there relaxing, you know, the car is driving itself., but then it's just blaring high volume every—

I'll take it a step further.

Every 5 minutes, one 30-second ad.

What if there was a whole company dedicated to building self-driving cars that was funded by an advertising company? Like a company that just prints billions, hundreds of billions of dollars in advertising. Like their DNA is advertising. And then they go into the self-driving car market.

Why has no one done this before?

They would have a huge advantage in terms —of targeting.

Yeah, why has no one done this before? Yeah.

They could even have a video platform where they're making a lot of money running video ads, and they could take that video platform— Oh, almost like a YouTube. Like a YouTube. Yeah, exactly. So you could have a huge ad network, YouTube ads, and then you could have a self-driving car company, put them all together. That's synergy right there.

And maybe you could spin the company out at some point and raise some venture capital too. Spin the company out.

They secretly still advertising now. I don't think Waymo will do crazy ads.

I think, I don't know, maybe a major streaming executive just texted. A major streaming executive just texted me and says, okay, going to get Tubi to build a car right now.

You got to build a car. You got to do it. What else is going on here? Okay, let me tell you about CrowdStrike real quick. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. The owners of MySpace say they're planning to relaunch the once dominant social media platform. And I feel like this has happened multiple times. I'm having déjà vu here, but we'll dig into it. In a recent documentary called MySpace about the company, we got another app-loving cars, Axel AI, Axel.

Axle. No, it's, it's, it's a self— it's a self-driving car, but you have to play mobile games to keep it going.

That's a feature. I mean, you get a Tesla, you can play Palutopia there.

Replace the steering wheel. Yeah, why not Clash of Clans? Big iPad and you're just playing mobile games.

Tesla's like 90% of the way there.

If you stop leveling up, the car just pulls over.

You gotta see— you gotta keep playing. Okay, we're getting somewhere. This is good. Uh, so in the recent documentary MySpace, Chris and Tim van der Hoek The brothers who acquired MySpace in 2011 after its decline said they intend to relaunch the platform with a new vision. There's no launch date that's been announced, but they say they want to build a social network that feels different from today's algorithm-driven apps. They don't believe in preference falsification. They think when people say they don't like algorithmic feeds, they actually don't like them. We'll see how it pans out. So MySpace launched in 2003 and quickly became one of the world's largest social networking sites. It became famous for customizable profiles, music discovery, and its iconic top friends feature before Facebook eventually overtook it, uh, in, in users and became the dominant social platform. At its peak though, MySpace attracted roughly 150 million monthly visitors worldwide. That's not necessarily people that had profiles, but just visitors you'd land on a MySpace site. So, huge for the day but not huge by modern standards. So, the platform's ownership history mirrors its dramatic rise and fall. In 2005, Rupert Murdoch's News Corp acquired MySpace's parent company for $580 million. It was like a— I'm pretty sure it was a fantastic outcome at the time. Now, by social media standards, it's like, "What are you doing?" Just 6 years later, in 2011, News Corp sold MySpace for roughly $35 million though. So completely, you know, got racked to Specific Media led by brothers Chris and Tim Vanderhook. Justin Timberlake invested in the company and became a prominent creative advisor during the acquisition. In 2016, Time Inc. acquired Specific Media's parent company in a deal valued at about $87 million. 3 years later, Meredith Corporation acquired Time Inc., briefly inheriting MySpace before selling its stake back to the Vanderhook Brothers that same year. All told, MySpace went from a $580 million trophy acquisition at the height of the social media boom to a roughly $35 million distressed sale just 6 years later. Today, MySpace.com is still online, primarily serving as a music and entertainment website while its parent— while its owners prepare what they hope will become a second act. Are you bullish? What do you think they should do? What's the advice for the owners of MySpace.com? Nostalgia. You got to call Bending Spoons, right? Is it making any money?

That's the issue for Bending Spoons.

You get one Italian teenager to run this whole thing.

They're like, "If you were making $500 million, we would love to buy you for $1.2 billion." But since you're making less than—

I don't know if they're making any money. They must be making some money just as a website with some ads on it or something. Who knows? Anyway, let me tell you about Cisco. Critical infrastructure for the AI era unlocks seamless real-time experiences and new value with Cisco. Our next guest is here already, I believe. We have Grace Lee from Intelligents, the co-founder raising a seed round from Index Ventures. Brought you in a little bit early. What's going on? How you doing, Grace? Good to see you. Great to meet you. Thank you so much for taking the time to come chat with us. Please introduce yourself and the company.

Well, I'm Grace. I'm the CEO of Intelligens. We just launched yesterday. We've been the team behind Design Arena. Actually, the company started because we were making games. We were trying to make a game engine. We noticed that the models could make playable games, but the games were not fun at all. Like, nobody wanted to play them. And we thought, you know, building harness probably doesn't seemed like the right move. What if we could turn it both into a product for people to build games, but also a product for the models to figure out what it is that humans like to play, like what are playable games that are fun. And then we started that. We just— one of our friends linked it on Reddit. He didn't even tell us about it. And it got like 3,000 users overnight. We're like, well, that's more than the game engine has ever gotten. And since scaled that to, oh man, 5.6 million people across 192 countries. So every country in the world except for North Korea. We've been told that there are sanctions there, so we can't go in there yet. And then on the Frontier Labs side, we pretty much work with all the major players to help improve their models and hard-to-verify domains. Amazing.

Can you walk me through how Design Arena works, the various incentives that happen to actually get people to review designs for models and just the whole— it feels like a a product that would have sort of like a cold start problem. How did you solve that? How do you scale it? How does it function like in equilibrium?

Oh yeah, like really good question. So we, we took a look at the optimal form factor for these cases where users are trying to do something, but you don't really know what the optimal end state is. So designers, right? What do they show their clients? They actually, they get the client prompt and then they show the client multiple multiple versions of maybe what they might want to see. That's exactly what the user interface looks like for our users. So they don't actually need an incentive because it's a design tool for them that does what they want because they show them multiple versions, just like in the real world. Like, how does a singer figure out if a song resonates? It's not that they cook in a room, really, and like study. They just release multiple songs, right? They see what hits work. How does a game developer figure out what games are playable? They release multiple games. And they see what works. So for the user, they actually don't need an incentive to be on the platform. They want to design something or make a game or make a website. And the form factor perfectly fits that. The, the models, the intelligence doesn't exactly know what the user wants, but then they'll get multiple versions. And through feedback, you can get closer to the end state. And that just happens to be in the perfect shape that the intelligence layer needs to improve more to what it is that humans want in the first place, the preference signal, the, the user behavior, et cetera.

A little bit of like a— I don't know— philosophical question, but how do you think about optimizing design for the user and what the user thinks looks good versus what is good for the business? Because I've been— like, I've just been bombarded with products and websites that feel like they're horribly designed. Designed, but I understand that they are economically valuable. John Gruber was on the show yesterday. He wrote a blog post recently about the Temu app being a mess. And if you go to Temu, it's like so many pop-ups and spinning wheels. It is the epitome of terrible design in the, you know, aesthetic sense. But I'm sure every design decision that they've made has been economically technically justified. And I don't think that they like never thought to make it clean, right? And so how are you thinking about those two tensions like actually instantiating themselves in the work that people do when they go to an AI model and ask for a design?

Yes, very good question. So it is our job to figure out what better means for this person. If you are on the platform, to make a beautiful portfolio, then you are going to give us feedback that hints us in that direction. Through a couple of interactions, we can figure out, okay, this person is actually just in it for the visual preference, right? But if you're on the platform to build a highly performing website, like a Temu website, for example, you're probably optimizing for dollars converted at the last screen. And so the end signal that you're tracking is different. The way that we like to think about it is we're almost doing the work of a PM, right? What a PM does they sit in the product and they took a look at the analytics of what people are doing and they have an end goal in mind. The PM's job is to figure out what is the right KPI to be optimizing for and then to bring that to the table of engineers and to the resource allocation and should try to get it in that direction. So it is our job to figure out for this particular user, what does better mean for them? And then measure those bright signals to then label correctly and then provide that as feedback to improve the models.

How do you— Jordi, please. You had a follow-up? I have a follow-up there. Just following up on Design Arena, how are you balancing the various downstream business opportunities that come from having a product, a platform like Design Arena? Because I could imagine you can sort of become a model router in many ways. You can do forward deployment. Deployed work and be an expert and partner to companies that are picking tools. Like, there's services as a software, there's going down the financial side. Like, there's so many opportunities. It must be hard to— like, you're sort of stuck in the idea maze longer than usual, even after you have traction and fundraising and revenue. There's more opportunity. What have you looked at? What's not interesting? What has stuck out as the easiest path?

Oh my, okay. That is the best question, by the way, to be asking in this space. Like, there are so many downstream opportunities that come from figuring out what it is that, that people want. What we've found to be the most interesting actually is an intelligence marketplace. Okay. Right. Um, there, there is some, this is a little bit kooky, but I do believe that there will one day be as many suppliers of intelligence as there are currently suppliers of information. This feels like the early days of the internet, right? Like where anybody can make their own intelligence. I don't care if that's a foundation lab with a main API or something that's post-trained or something that's totally off the shelf with a skill, but there's like a million different suppliers of information. And if you can be the person that best matches the supplier of information with somebody, a supplier of intelligence, I should say, somebody who's asking for intelligence the same way that PageRank for Google search was the critical bottleneck to get the internet, um, into everybody's hands. That is the position that we actually want to be in. So whatever the work it is that we do, we want it to fit into this flywheel that we have, and we want to make sure that we're not just good at value creation, like not just good at improving the models, but also good at value cap— uh, value capture. Once the models get better at design, it's not like our work becomes less useful. It actually means our product can reach more people people because they're better at design. So we care about doing things in the opportunities that are coming our way, which we're lucky to have a ton of inbound from, from yesterday, that it feeds into this overall flywheel that if the models get better, we also get better.

Yeah.

Uh, almost $60 million of revenue. I think, I think a lot of people were quite surprised, not because the company isn't like significant and at the center of this, you know, explosion of creative intelligence just because it's— you guys had kind of flown under the radar. So first, I want to hit the gong for $60 million-ish of revenue. There you go. Great hit. Thank you. And a very— and a very like— I like the ratio of the fundraise to the revenue. You know, normally when somebody comes on with this much revenue, they're like, raising a billion. I'm sure that's up next for you. But I wanted to ask, like, what do you think creative superintelligence looks like? Because with, like, images, I think images are going to be solved, right? At least, like, photoreal images are pretty much here. You can imagine with a couple more turns, like, at least making images with AI will be solved. And it's getting a lot easier to prompt and harnesses will get better at editing and modifying and getting better outputs and all these things. But like photoreal images will be solved. What does like, what does creative intelligence look like in 1 year or 2 years down the road?

Creative superintelligence to us is whatever you can think will exist. In fact, things that you can't even you have the scope to think of can exist. So, one thing that the distance between an idea in your head and it being in the real world right now, it's very far. Let's say you have an idea for a new pen, what a new pen might look like. There's a very long distance that you have to travel. You have to know how to design the pen. You have to know how to process CAD. You have to find a manufacturer. You have to build. There's a very long— there's a very long— and because of that, We have to localize on a couple of pen form factors that everybody uses. Like, it just, it becomes standardized. But what if creative superintelligence makes it such that you can sustain infinite varieties, kind of like how you can have infinite versions of the same information being represented in different formats? That's the way that we view creative superintelligence. Which is any idea that you have, the instant that you have it, it is visible in the real world. And there's no latency and friction between your ideas and them being born in the real.

That sounds different than the models coming up with the ideas, which is interesting. I don't disagree. I think you're right. We were debating this yesterday about will a model be able to come up with the idea before you have the idea. You even think of it? What is the timeline for that? That's the next thing.

Yeah, we were talking about this because it feels like mathematicians have been having somewhat of an existential crisis for a while now, but especially over the weekend. And I was just thinking, I was joking to John yesterday, just saying like, you know, they're coming to us, right? They're coming for us next. Like, we're like, like, I'm, I'm, I'm, you know, I consider myself a creative person. Yeah. And right now when I look at AI-generated outputs, the ones that really resonate are ones where there was like a fundamental creative idea. Yeah. That then the model did a good job of instantiating it, but it's not like the original idea was that complicated. The example we always use is like Harry Potter Balenciaga, right? Like, It's not a complicated idea, but to me, to me what will be interesting, and I'm surprised no one has like really built some type of like effectively like loop around this, which is like combining two random ideas and then creating the visual sort of output. And just doing that over and over and over because it will end up mimicking like the full pipeline of human creativity, which is like original idea, which is combining two things. Harry Potter, Balenciaga, combining these two things and then just creating the output. And then once you get that, then you're just running on this loop where already it feels like there's no real new ideas in the world. Like for 100 years we've just been as humans like recombining different things that already exist. And it's sort of sad that we can't come up with very many new ideas, but then eventually the machines will have just created every possible variation of every idea there could ever be. And then a human will have an idea and then they'll be like, oh, well, actually the machine figured this out like 2 years ago and here it is on a website.

I mean, I think something that's pretty incredible in, you know, is people don't know what they want until they see it. Right. And oftentimes it's not even the new idea that is like the— it's showing it to a dozen people and then 2 dozen people and then 100, then 1,000. And seeing what sticks. Like, again, this is kind of the creative process, right? Like, you, you will draft up 100 versions of the same script and then see what are the gem moments that actually resonate with people. And the resonating with people, like, that's the verifier part. Like, that's where humanity comes in, right? Like, if you're trying to give something to people, you need a way to measure whether or not it actually, it actually resonates with them.

That said, company Simile, who just raised a big round. Might make humans unnecessary to even judge what will resonate with you.

I think we're safe.

Every AI lab I talk about is working on this thing called superintelligence. I haven't met a single AI researcher working on room temp intelligence. And so, like, they're not even trying to displace me. It's ridiculous.

Actually, there's probably a Neo lab called normal intelligence.

Probably, probably. Last question. Gaming arena. We got design arena. You have gaming DNA. What about models building games? The games go up against each other. Which one did you play longer? Which one did you have more fun? This feels like the next thing. Is this you? Is this another company? This is us. That's you.

This is us. So we've actually been working on this for a while. Scoop. Here we go. Game dev is incredibly hard to verify, and that's why it's so much fun, right? Because you don't just need a really good storyline, you need a way to verify if something is fun. Yeah, like that's impossible to to do, you know, without people on the other side. Exactly. Yeah. And so for a while now, we've been, we've been helping companies try to get better at, at this domain. And it doesn't just involve like the strategy and then being able to save game state. You've also gotta have beautiful graphics. You have good multi-processing. It's a very hard frontier. Yeah. Yeah.

That'll be a lot of fun. Yeah. Very, very interesting to think about like, oh, I'm gonna go and review models on gaming arena, see you in 400 hours because I need to decide which game is better, this 100-hour game or that 100-hour game. Yes. That's kind of what's going to happen.

Yeah. Yeah. You literally just need to put it, like the way to think about it is you need to put it on the equivalent of the internet and you need to see if it gets picked up by people.

Yeah. Yeah. Like the algorithms, like the Steam store, the Instagram algorithm. Like these are the final bosses of the arenas perhaps. In many ways. Exactly. Very interesting to think about. Well, congratulations on the round. Congratulations on all the progress.

Thank you so much for your time. Yeah, amazing progress. And thank you for— thank you. I think you're doing a public service, right? If you can get all these AI companies to make better outputs, the world will become a more beautiful place. Yeah. And I think it's important.

Well, have a great rest of your day.

Great rest of your day, guys.

Great to meet you.

We'll talk to you later. Goodbye. Let me tell you about Console. Console builds AI agents that automate 70% of IT, HR, and finance client support, giving employees instant resolution for access requests and password resets. Our next guest is Samir Khaji from Alloc8. He's the co-founder and CEO. Samir, how are you doing?

What's going on? Good, good, guys.

How are you guys doing? We're doing fantastically.

Uh, fun day. Uh, do you drive a BMW? Have you been getting bombarded with Spider-Man ads?

I, I wish. No, I don't drive a BMW.

You're lucky. You're lucky. You're lucky.

That's the hardest part, is burning up the time. They're all ad-supported now.

They're all ad-supported. Yeah, apparently. Apparently, it's not even opt-out. Does Allocate take a couple weeks off in August as a company?

Oh, no. No, there's no taking any time off. Unfortunately, a lot of our clients are taking time off.

That's what I'm saying. That's what I'm saying. If you wanted to plan— I think it can be good, at least for smaller companies, to say, hey, we're all going to take vacation. In this general window. Sort of collusion. Yeah. Yeah.

Well, I will tell you, though, I did go to Cafe Venetia the other day, so I don't know if that counts as a vacation, but it was, it was nice to kind of get away. But yeah, no, it's, it's crazy busy right now. As you guys know, so much is happening in the market. So many funds are raising, so many companies are raising. And having been in this market for 27 years, this is like the craziest I've seen it.

Yeah. So does that scare you?

Because obviously it's good. It's obviously great for your business, which we should let you introduce, but maybe introduce Allocate first and then I want to get into—

Yeah. Why don't I give it? I'll give you the quick sort of 30-second sort of pitch on Allocate. So having been in the venture market, working with private funds, kind of realized a few things. Private funds are now more numerous. We went from 3,000 asset managers to 30,000 within 15 years. Wow. The amount of capital that's being raised by these funds is enormous too, right? Like, you had Andreessen raise $15 billion, you had Thrive raise $10 billion. So these companies are staying private so much longer. So the private markets grew from a trillion to $17 trillion. And when you look at that, where is that money going to come from? Well, historically, the funds were capitalized by pension funds, endowments, foundations, and then it became clear that they needed to find new ways to capitalize themselves. And one way was family offices and high net worth individuals. The problem is it's kind of a headache to operationalize those small checks. And for the people investing, they can't write $10 million tickets into, you know, name your top fund. And so they go through the wealth advisor. So we create basically the bridge between the wealth advisory world and private funds. And for the wealth advisors, they have one place where they can discover access, execute, and report on private assets. So, there's $10 trillion of capital being right now managed by these independent wealth advisors. If you guess what percentage is in alternatives? Less than 5%. 3%. That was great. 3%. Good guess. Well, endowments, foundations, pensions, 20% to 50%.

Yeah. 20% to 50%. Yeah. Yeah, exactly. The Yale model. Yeah, we got to get everyone on the Yale model.

Yeah, well, you know, you got to be a little bit careful on that for sure. But, you know, look, I mean, the reality is like you can't ignore the private markets. You had Anthropic, OpenAI worth almost a trillion. Yeah, SpaceX goes public for $1.77 trillion. Like, it's crazy.

Yeah, yeah, yeah, yeah. And, and it would be— it just feels like from just like not even the financial numbers, but just the idea of just being in the American economy and being bought into the next iteration. Like there's been this discussion of going, going public earlier. It's very onerous, etc., etc. But there is another world which is just broader exposure through funds. And that's a lot of what's happening here.

100% agree. Yeah. Okay. Once a week, maybe twice a week. Sometimes 3 times a week or 4 times a week or even 5. There's a company that I've never heard of that comes on and raises half a billion dollars. And, uh, you know, they're maybe like 1 or 2 years old. That I get a little bit wary because I sort of like, my career came online, let's say in 2018. Yeah. I'd been paying attention to startups prior to that, but like really was paying attention in 2018. And we went through this sort of crazy period and then we saw a pretty massive correction. I got to sort of be on that roller coaster. And so now, while we're back in boom times, I'm just much more wary than I was before because I was young and maybe overly optimistic, right? And so today, I'm just— when you look at the public markets, there's a lot of— a lot of valuations seem relatively reasonable. In the private markets today, there's, you know, we saw Airtable. If Airtable was a, was a, was just like, I know a, I know a SaaS company that has a tiny, tiny, tiny fraction of Airtable's current revenue that, that is doing a new round well beyond Airtable's valuation. And it's just because they're squarely in serving other AI companies when ultimately they're just as much of a software company as Airtable was. And so you see this massive disconnect between exits and private valuations. And so I have a lot of alarm bells going off, but I'm curious how you see it and LPs on the platform are seeing it.

Yeah, one of the most dangerous things to say in the investing world is this time's different. Because rarely is it completely different. So, you know, similar to you, I've had some scar tissues built up. I actually, you know, started my career in '99. I was at Silicon Valley Bank, I was at SVB, and I was lending into the dot-com. And back then, you know, companies were going public based on one metric, which is eyeballs, right? Like, how many eyeballs, how many people are visiting your site? Well, today it's AR, which, you know, as we know, can be a little creative in terms in terms of how people are actually saying AR. What's actually true right now? So I think we have to disassociate, you know, AI as like this technology innovation, which I think is going to be bigger than anything we've ever seen, bigger than the internet, bigger than mobile and cloud, and maybe the biggest innovation since the railroads. The challenge, though, is when you have a new technology of this size and scale, especially moving this fast, the supply and demand for those companies changes. So you have so much capital being raised. People are incented. Like if you're a VC, you're incented to invest in the next hot AI company, which has grown from, let's say, $1 to $10 million to $100 million in revenues, knowing that full well that if you catch the tiger by the tail and you get the Anthropics or you get the OpenAI, your career is minted. And so a lot of capital goes into these companies. And every single time that I've been through one of these, the companies that do win are bigger than ever. By a massive degree of magnitude. The issue is so many expensive mistakes. You mentioned Airtable. That was like 2021. I think their peak valuation was $11.7 billion. They raised almost $1.5 billion and they sold to Bending Spoons for $1.25 billion. Right. That means common shareholders get nothing. And the people that came in at the end, they may get a return on capital, but that's a horrible—

I thought common shareholders were getting something because the actual— the total price value is $2.7 something billion. Yeah, $2.2 billion. There's roughly $800 million. And there was a— but again, it's like a billion tiny fraction of what they would have probably marked it themselves.

Well, if that's the case, that's great. But that's not the case for a lot of those 2021 companies that raised at sometimes— I was seeing 50 to 100x multiples for these companies. Now the multiples aren't that crazy this time around and these companies have grown. The bigger challenge is like which one of these companies is actually durable? Like, I mean, Is a big frontier model going to subsume you? What happens with open weight and open source models? Still too early to tell. So I think you're right in that you should be somewhat nervous. But at the same time, venture has always been a parallel industry, right? 6% of total companies that actually return over 60%. That's historical. I think it's even more amplified right now. But I'm, you know, in the mindset right now that I'm been through these periods, and people always have short memories. Like, 2021 was what, 5 years ago? And it's almost like we forgot 2022 and '23. Things do not go right when you just plug so much capital with such speed. And so, I expect the same thing to happen. I don't know when this changes and when gravity comes back in the private markets, 6 months, 12 months, 2 years, maybe it's even 5 years. Maybe we're all wrong. But I would I would actually be willing to bet that within the next few years, we're going to see some pullback. And then, we'll see what they say when the tide goes out, you see who's wearing the swim trunks. And I think we'll see that next year or the year after.

Oh, okay. Yeah, that's a better way to put it. Who's wearing the swim trunks. Well said. Well said.

I want to know how are new venture firms changing in their construction? What— you mentioned the boom from 3,000 managers to 30,000, but what are you noticing over the last few years that might look different in the way a new fund gets built? Maybe it's more interns or more or less banker crossover types or more operator, more founders. Are there any trends that you've noticed that you're seeing? Okay, that's clearly working at certain funds. More people are following this even in the way they're structuring or the, or the pacing or anything, anything trend-wise around new fund formation. Yeah.

And new fund formation has slowed down, you know, you know, for sure over the last few years. Right. It was at a fever pitch in 2019, '20, and '21. In fact, everyone felt like it was cool to be a VC without understanding that it takes forever to know if you're any good to get paid and all those things. Fundraising is really tough. Yeah. And you know, a lot of it, what came down to when you have so many funds in the market, everyone's searching for differentiation. Mm-hmm. And so they create these things, which is like, I'm gonna create a talent team. I'm gonna, going to create, you know, um, a, AI sort of way to vet things. And I think these are all around the edges, but any venture firm at the end of the day, all that matters is, you know, sourcing, picking, and winning. You got to do 2 of those 3 things really well. If you're a really small firm, it's sourcing and winning candidly. I have no clue if you're a good picker, and there's so many false positives. So I think people sometimes lose the plot a little bit and say, let me create something really unique that I can tell my LPs versus focusing on, am I going to see the greatest companies or the right entrepreneurs in the thesis? And once I see them, am I going to win given the competitive arena? And I think that it's really tough to be an emerging fund. I mean, raising capital, I think it's a great time to start if you have like a reason to win and a reason to exist. Yeah. Now, big funds are different. Like, let's— they're playing a different game. Yeah. Um, you know, somebody raising a $10 billion fund is just trying to get into the very best companies and plug a billion dollars into that company.

Has— how often do you see someone coming across and pitching sort of a new firm that feels like an old firm in the sense of we're going to go and compete and win $20 million Series As and Bs or something around there as opposed to the early fund manager, $50 million solo GP. They're going to do $100K check into this and like little add-on that feels more more winnable, to your point about the importance of winning? The Mount Everest of winning and picking is probably you're going to go lead that major round without the signal from a tier 1. Yeah.

I mean, look, if you're going to do a $20 million Series A, which is not going to be like a big Frontier Labs type of company or one of the big mega rounds, you have to— I mean, to raise them in a fund, you probably want to raise $350 to $500 million, right? And who is going to give you the money? It's people whereby you have a track record somewhere else. So we have seen a few, and this is all public now, but Chemistry is an example of one where you had 3 people, one from Index, one from Andreessen, and one from Bessemer, right? They come together. They are clear Series A will lead, and they kind of sit in the middle between the seed funds and the big mega shops who are doing different types of Series A or different types of Series Bs. I see very few of those now because the bar for those is pretty high. Yeah. And if we look at 10 funds, maybe one falls into that bucket. Yeah.

Yeah. It does seem like it's maybe an underrated white space because there's so many funds that can— they're so big at this point, even though they might be able to win, they're also not that upset if they they lose that round because they're like, yeah, okay, we'll just come in with $100 million later and get our position. Like, it seems fine anyway.

What, uh, how have you— how did you— looking back, like, what's the takeaway from the Solo GP, I'll call it meme? Because it's interesting because it was something that became like very desirable and was— and it was something that I think LPs were, were interested in. It's something that certainly people that aspire to be venture investors. I think a lot of people have this idea of like, oh, it'd be amazing to be like a solo founder or solo GP, right? You get all the glory, all the economics or whatever. But it's interesting because it just feels like a solo GP, when it works, it's amazing, right? I think of a lot of, you know, Gills of the world as, you know, the, the top of the mountain there. But, uh, there's so many cases where for LPs it's like, well, it'd be great if you had like a solo GP would be cool, but what if you had 3 or 4 partners that were all fantastic and it just increases your luck surface area? And then for a founder, it's like, yeah, you want to basically pick one partner at a firm to partner with, but it's awesome if there's a few other partners at the firm that can maybe make an introduction or open up their or whatever. So it feels like something that in practice, like, is— it was like a cool idea and there was this moment around it, but now if you want to win, why would you— why would you even like brand yourself in that way?

I mean, look, I think it can still work, and I'd rather back a solo GP than somebody that, you know, decides they want to start a venture firm and decides that to raise capital I just need to bring on a partner and they've never worked with that partner partner because the problem with partnerships too is there's a lot of politics. Are we aligned? How does decision-making work? If a deal is not consensus within the partnership, does it get done? And those can actually create— Bessemer, for example, has this great anti-portfolio. If you go to their website, all the stuff they miss, and I guarantee some of those misses were one person had a ton of conviction, but they couldn't get it through the partnership. And so solo GPs at least can move really quickly. They don't have to go through a committee. Even Elad Gil, who you mentioned, he has a team actually, right? So he's solo GP from an investing standpoint. Wow. Stolen valor.

Stolen valor. He has an amazing team, but I can't name any.

And also the solo GP branding was very much like foisted upon him. He was not like, I will never hire anyone. It was more just like something that was to talk about.

There is a guy like Oren Zev, right? So Oren, I don't know if you've— Oren has been doing it, and he will say, I'm a solo GP, I have no admin, I have no other partners, I have no one. It's just me. And that's what you're going to get. And like, he's done, you know, candidly, very well for himself. You've seen everything's public on him. And so I look at it and say the bar is higher. And a lot of solo GPs realize it's pretty damn lonely, right? So you're basically doing the raising, you're working with companies, you're trying to do everything, you're running the firm. And then you're like 4 years into it and you're like, wait a second, um, you know, my fund is trading at a 1.1x because I went crazy in 2021. No one's giving me money. I'm, I'm pulling a salary of the $20 million fund that I could probably get in a quarter working somewhere else. And you're like, why bother? What's the purpose of doing this?

Yeah. Have you seen anyone, uh, try and tap the traditional venture capital backers, the LP base, for something that looks like more competitive with a Thrive Eternal? Like, I'm going to go buy a baseball team, and I want to tap Yale for that, or something like that, as an example.

I mean, not that much. I mean, Thrive is pretty unique in everything they do. I think they're really progressive, and Josh and team have done a fantastic job. I mean, GC is also— General Catalyst. They've done some pretty interesting things. I mean, they've gone into hospital systems, right? So they've done things like a PE and credit. Yeah. And, you know, that is more like private technology finance and going into these adjacent areas to create competitive advantages. Because when you're raising $10 billion, guess what? There's about 15 firms that you're going to compete with. So what is the differentiation outside of your partners?

Yeah. Do you think the situational awareness story changes RIA?

Consideration? I don't think so. I mean, look, I mean, you know, it was a very unique thing. I mean, the guy had— I mean, the guy is incredibly smart, and, you know, he still is up, you know, overall from, you know, the day of. Um, but when you drop that much and use that much leverage, like, it's gonna be a big blowup. RAs themselves, like, you know, are we talking about RAs from the asset manager side or RAs from the wealth manager side? It was just a—

there was just a conversation a while ago about, oh, VCs, like, they often have really great deep understanding, sometimes board seats in these public companies. They take the company public, then the stock trades down, but they know that there's a really great thesis. Maybe they should be holding on behalf of their LPs instead of distributing. The RIA sort of sets you up for that. I'm less talking about RIAs as like going long crypto or going long secondaries. I'm talking about playing in the public markets. Someone in the tech community just got their hands sort of burned Does that change like the appetite from LPs to allow venture capitalists to play in the public markets at all?

Yeah, I don't think most LPs want their VCs to play in the public. Obviously Sequoia, everyone knows, you know, has a long-haul only. Then there's crossover funds that do it, but it's just a different skill set, honestly. And like, you know, if we're an LP, right, or somebody else is an LP, you're at the end of the day paying the GP to You know, invest in private companies and then exit those private companies at some portion where we can make the decision on what to do with the public stock. Yeah.

You were at SVB in 1999. Any good stories of venture capitalists using leverage?

No, but what people— I'm surprised.

I like— we hear John and I were talking about this yesterday because, like, you know, the venture capitalists, or at least there was —some that were kind of doing a little grave dancing on Leopold last week. And we were talking yesterday after the show, being like, yeah, it's funny that VC funds can't blow up. You know, it's just like a slow death. If they die, yeah. If they die. But I'm wondering if there was ever a period in Silicon Valley history that Silicon Valley was like, you know, we should be using leverage. I want to re-leverage. And then some sort of set of events that led to that not being, uh, very common.

Now, now, fortunately not too much, right? So not in terms of true leverage. Now, there are people that have used nav lines, but you haven't to be an RA. Most VC firms are exempt reporting advisors, so they're not able to do those things. What we did see, and this is less in '99, but 2008, right before kind of the GFC started, there were some funds that did warehousing. And what they would do is they would go to a bank like SVB and say, We're raising a $200 million fund. We are going to start making investments before our first close. Can you lend us the money to make those first few investments? And guess what? There was a couple that did that more on the fund-to-fund side than, you know, the, you know, direct side. And guess what? The market changed. They couldn't raise capital and they had this huge warehouse sort of facility that they had to pay off with no capital.

Interesting. That is a crazy scenario. Yeah. So it has happened. Technically. Thank you for the history lesson. This is exactly what we were looking for.

Perfect guest for our earlier debate. Very cool. Well, it's super fun. Really, really fun hanging. Great to finally meet. And let's— we're trying to get a pulse on what's happening. We think what's happening in your world ends up sort of echoing later on with a lot of the coverage that we do.

Yeah. So we'd love to do it again.

You guys do a great job. Big fan of the show. Great guests. Actually, my board member was on your show recently, Carter Rean. Oh yeah. Yeah. Carter from M13. Great guy. And whenever you want me back on, I'm happy to do it.

That'd be amazing. Thank you so much.

Great to hang out.

We'll talk to you soon. Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Up next. Next we have John Quinn, the most feared lawyer. Welcome back to the show. Great to see you.

So we, we have a video that we'd like to play.

You want to play this?

I want to play, I want to play it for John. I'm gonna play it for John.

We made a heavy metal song about you, the most feared lawyer.

About time. About time.

Can we get it up on the big screen?

Play, play the, play the last 45 seconds.

This is what John was spending like probably an hour last night working on burning the midnight oil. Burning the midnight oil for this. Here we go.

You can turn around and we got the full, the full AI rendering version of it.

Let's get it louder, please.

Here we go. But I think this really captures the spirit of what you bring to the courtroom, what you bring to the legal profession. Just, uh, total fear and chaos. Yeah, the firm, of course.

Thank you very much.

Yes, we'll send it. You can, you can circulate it with the team.

So we're going to talk about agentic AI. Let's do it. Law.

Yeah, I want to talk about—

I have some news this morning. Give us the news. Do you mind if I do? If I reveal something? Please. The general press doesn't have you.

Absolutely.

So a case that our firm has. Yes. And that I personally worked on. Okay. Within the last 2 hours, I got the opinion from the Ninth Circuit Court of Appeals. Okay. The case is Amazon versus Perplexity. Whoa. And it's a very interesting case right at the intersection of agentic AI and e-commerce. Yeah. So we represent Perplexity. Yeah. Amazon sued Perplexity because Perplexity has an agent, the Comet, that users can deploy to go shopping and doing whatever. Yeah. Amazon didn't like agent Comet. Going to its website and buying stuff. I think because it doesn't have eyeballs, can't see ads, can't be, you know, can't be upsold. Can't be manipulated. So Amazon sued Perplexity under the Federal Computer Fraud and Abuse Act, which you can bring a civil claim under that for damages. It's also a criminal statute. What it requires, the elements are that you access a computer without permission that, you know, you take information, you get information, and you cause damage. Amazon sued Perplexity saying, "Your agent is being deployed. We don't want that to happen. We've told you we don't want your agent on our website shopping and whatnot." Very unfortunately, and they moved for a preliminary injunction to shut it down immediately. Very unfortunately, the district court up in San Francisco entered a preliminary injunction a few months ago. We lost at the trial court level, but there's nothing a trial court can do wrong that a court of appeals can't fix. Okay. And just 2 hours ago, we got the opinion from the Ninth Circuit, and the Ninth Circuit understood that Perplexity is not accessing Amazon's computer.

The user. The user. The user.

Yes. The user downloads the Comet, deploys it, and tends it to go shopping. Yep. So justice prevailed. But that's— I mean, it's an interesting question at the intersection of agentic AI. There you go. Very interesting. Thank you.

All right. Yeah, very, very interesting.

How foundational is this? Because it— because Perplexity is just one company that is trying to do this type of thing. I think there's a lot of consumers that want the ability to deploy AI, whether it's self-hosted open source. There's not even a company behind it. They want to be able to do these types of things. At the same time, there's going to be corporate interests that don't want this.

Of course. I mean, and we were very early stages in trying to understand how traditional legal principles, sure, you know, apply to agents and agentic AI. Yeah. I mean, take this subject, it's something very basic. Contracts. Yeah. I mean, it's well known, established that, you know, you can establish a contract between electronic interaction of agents. Doesn't have to be a human being. You know, you can have a binding contract. Okay. But things can go wrong. Yeah, of course. I mean, the agent may not follow instructions or optimize for the wrong thing. For example, you tell the agent, "Guau, I need some cloud computing time. You know, go out and buy some compute for me at the cheapest possible price." And the agent does that and goes and enters into a long-term contract —to get a cheaper price. —with a huge cancellation penalty. Sure, sure. I mean, that's a problem. Nightmare, yeah. So, and just the sheer number of contracts entered into by agents. Agents can enter into thousands of contracts before somebody's realized there's been a mistake. You know, something's gone wrong. It is a disaster. So, I mean, what do you do? I mean, if you find out your agent has gone out and entered into a contract that's not what you're interested in, not what you intended, intended. I mean, one thing we know is you can't say it's the AI's fault. Okay. It's not my fault. I mean, there's even a law in California, I think it was enacted this year, Civil Code Section 1714, that says you can't blame the agent. You know, it's, it's your agent, you're responsible.

What's the human precedent? Because big companies, I'm sure, for a long time have had issues with an employee entering into a contract, you know, signing. Maybe they have—

I mean, maybe that's very interesting, and it's kind of the same concepts I think they're going to apply in the agent context. That is, did the employee have authority? Yeah. And even if they didn't have authority, did they have apparent authority? Yeah. If you deployed the employee, they had a job title, they had the card, seemed like they had the ability, the authority to enter into that contract, you're probably going to be bound by that. Mm-hmm. Similarly, with an agent, I mean, the presumption is the agent's out there in the marketplace buying things. Yeah. The presumption is going to be that the contract they entered into is binding and you're stuck with it. But there is a concept of mistake that can come to the rescue. If the agent goes out there and buys, because of a parsing error or something in the software, buys 10,000 of something, which historically you've only been buying 100 of, that may put the counterparty on notice that, wait, something's gone wrong here. And if the counterparty is on notice that this is a mistake, the counterparty, you may be able to get out of the contract. So that's just kind of an example. So these traditional legal concepts about authority, apparent authority, I think are going to come in play in the agentic AI as well. Yeah. Are there any—

what precedents are people drawing on to understand where electronic contracts go? Are there any learnings from, like, smart contracts and what the whole discussion around cryptocurrency, or is it more like what's been built up over terms of service? Because yeah, this is like an almost entirely new paradigm, and I don't know exactly.

I mean, I call it the problem of new wine in old bottles. Okay. And whenever there's a new technology, yeah, we see this. We have traditional concepts about contract. Yeah. How do you form a contract? Yeah. What is agency. Yeah. And then we have to adopt it to this new era where you have electronic agents that are interacting with each other. Yeah.

What do you think the equilibrium is? Do you think that there's going to be a flurry of lawsuits and settlements and then sort of a new, uh, new establishment of like standards, norms, default contracts? Because Silicon Valley has been through a whole turn of contracts where there used to be equity financings, then there were a lot of convertible notes, then there was eventually the SAFE contract, which is a much lighter version of a financing that moves a little bit faster, but can still hold some of the weight. And so, like, is there a— do you think there's going to be demand for some sort of standard contract that agents can sort of execute that builds on the shoulders of all the contract law, but still allows for Yeah, I think we are going to see contracts like that.

Okay. For sure. Yeah. I mean, but still, we're always going to have the potential, you know, for the rogue agent. Yeah. That does something that does— exceeds the permissions. Yeah. It's all the problem is what's built into the software. Yeah. So it's important to build in safeguards like caps, spending limits and the like. And for important and novel transactions, getting a human in the loop that the transaction can actually close, money can't change hands until some human being has actually looked at it. So I think that's very important. And then also people need to audit the history of transactions. As I said, you can have thousands of these contracts being entered into and performed before somebody realizes anything's been going wrong. You need a system where you're actually auditing them and catching those errors.

Do you know where DocuSign came from? From? Was that a legal innovation? Because we think of it as this software company that says a lot of employees, they've built this program for, uh, e-signatures. They have a lot of employees.

But what would we do without it?

But was there a moment where, uh, e-signature was illegal, was not binding, and then it flipped? Or did they sort of adapt to the conventions of what counts as a binding legal signature and sort of slot in and then establish themselves.

Yeah, I don't know the specific history on that. I mean, the guy who founded that company, I'm blanking on it. Yeah. It was actually a client of ours. Oh, no way. Yeah. Okay. So I'm blanking on that. So I don't know what the history of that is actually.

Yeah. I'm just wondering if there's like learnings from like other times when we've had to take something that happens in the real world and adapt it to the internet and thinking through through like the other foundational cases, maybe net neutrality, these different debates, because you have to imagine that consumers will want AI to go shopping for them. And that's going to happen, but they're gonna want it to be safe. And then companies are gonna want— They don't want surprises.

They don't wanna realize they got 10,000 toothbrushes instead of the 10 that they wanted.

Sometimes that happens just on Instacart. You order one bag You order a banana and you get one truckload of bananas. It happens.

I think there's also going to be issues in tort law. Sure. If you deploy an agent that causes harm or causes damage, how does tort law apply to that? Concepts of negligence, products liability. I mean, it's an unsettled question in the law whether software is a product. Yeah. You know, like we talk about products liability for defective features in cars and the like. So there's a whole body of law surrounding that.

Yeah.

Should that apply to, you know, agents that go out and do something that causes damage? I mean, we're seeing cases being filed where, you know, the allegation is that the agent persuaded somebody to engage in self-harm. Sure. The model. Yeah. You know. Yeah. And so is that a defective product?

Yeah. That's the question.

So concepts of negligence, products liability and the like. Apply. Yeah. And then people, you know, like in this case Amazon tried to keep the perplexity out from its site saying, you know, we don't— you don't have permission. Yeah. We're not permitting that. But I mean, we've all read about instances now where agents have gotten loose. Yep. Entered sites. Yeah. So what are the legal ramifications of that? There's a concept called trespass to chattel. Like, you know, trespass on land. A chattel is a thing. So there's also something called, and this is traditional, a traditional concept in the law, of trespass on a chattel. Chattel, whether it's trespass on a server, computer, and the like. Hasn't gotten very far, but there's some cases where those types of claims are being raised as well.

Okay. Yeah, how have you been processing all those cybersecurity incidents? It feels like these, they haven't, I haven't seen lawsuits filed. It feels like these will be discussed to settlements already. The CEO of Hugging Face was sort of outlining what a non-legal recourse could look like. Where do you think this all goes? Do you think there needs to be some precedent-setting legal case to create the template for how these things get worked out if they happen more and more?

I mean, the way, the way this, you know, we live in a common law country, of course. Yeah. So We have legislation, and maybe there'll be a need for legislation at some point, but until the legislators act, and it's hard to get legislators act, act in this country, the courts will be deciding these on a one-off basis, establishing precedents. So courts still will look to what other courts have done. Sure. I mean, the issue that's been teed up in dozens of cases and that everybody's familiar with is that copyright infringement, yeah, train a model on copyrighted material. Yeah, I mean, there are dozens of cases that are teeing that up. We have some indications from a couple of cases so far suggesting that that's not infringement, but it's what's called fair use. Yeah, I mean, ultimately, all these new issues that are being teed up, we're gonna have answers to them, but it'll take time. It'll just take— it depends on the persuasiveness of the judge's opinions and the precedents. And some things will end up in court of appeals and maybe even the US Supreme Court.

To make new law. How is the legal industry grappling with questions of liability around AI? If I have a law firm, my lawyer uses AI, are there, is there any nuance there or has the legal industry had to grapple with what liability, what the services delivered might look like in a world where there's AI agents that are going off and doing discovery or all sorts sorts of different things.

Well, lawyers are going to be responsible for what the AI does. Yeah, I think there's no question about that. You've heard about these cases where lawyers have filed briefs in court that have hallucinations. Yep. Cite to cases that don't exist, laws that don't exist, or cite to cases that exist but don't stand for that proposition. Courts don't have—

What do judges think about that? They're not very happy with it. It's like when I start reading something and I realize that it's just been just a prompt effectively?

No, I mean, the lawyers have been sanctioned for doing that. It's something that I think every law firm has to be vigilant about. Yeah, because, I mean, you can't get the models. I mean, you can try to tell Claude, you know, don't make anything up. Yeah, I only want double check. Yeah, make sure the case really exists. And it'll kind of be apologetic and say, you know, I'm just an LLM, I can't guarantee that. Seriously.

Yeah. It's funny. How are, how are the lawyers that you talk to at your firms and just other friends in the industry processing just the progress of AI? We had the breakthroughs from one of OpenAI's new models on Saturday that kind of rocked the math world. But I'm curious how lawyers are just processing the model's, you know, sort of broad set of abilities today?

Well, look, lawyers are, among other things, wordsmiths. Yeah, I mean, so we have targets on our back. It's amazing how much, you know, how much quicker we can get to a finished product that we'd actually want to use, serve on the other side, or file with a court. And, you know, I personally think most of our profession is whistling in the dark, that they don't realize what's going to come. There's a bit of an innovator's dilemma here because big law firms are doing really well.

Yeah. From all these AI—

why change? What's, what's the issue?

So, well, also, I'm assuming a lot of the most high-profile cases are AI-related. So at least in the moment, it's creating this sort of surge in demand.

Yes, there's a lot of cases related to AI. These novel legal issues like this Perplexity case we never would have had 18 months ago, the dispute between Sam Altman and Elon. Just on many different levels, it's generating new work in addition to new types of issues. But I think it's fundamentally going to change the legal profession and the way law is practiced. I mean, most law firms bill by the hour, and if I can press a button and get a work product that's 80 or 90% there, think of all the time, the hours that have been saved. And that really needs to be passed on. The client expects that. They're expecting to see that savings.

You said big law is doing well right now financially?

Yeah, I mean, the law firms, the biggest law firms are doing very, very well.

Doing very, very well. What was the hardest year financially for big law, or Quinn specifically?

So, we only do disputes work.

Yeah. So, I imagine it's not very cyclical, right? There's not. It's not.

A lot of people think that litigation is cyclical. Yeah. There's more litigation when there's a downturn in the economy. Yes. And less when things are going well. I don't think that's true. Okay. I mean, you'll have more of a certain kind of litigation when there's a downturn, distress-related, reorganizations, bankruptcy-related litigation. But when there's a downturn, —people can't perform their contracts. So you have those kinds of issues. But when things are going well and there are profits, companies will bring claims that they might pass on. They'll pick up the nickel on the floor and say, OK, we're going to pursue that. So I mean, the disputes world in America is generally pretty good. I mean, we live in the most litigious country in the world, our legal culture. Is— leads the league in that respect. Is that good? I mean, depends on who you ask. I think that it's a feature of our individualism. It's a feature of people's feeling of entitlement that I have rights, you know, I can get my day in court. Yep. I can get a hearing. Yeah. So those are good things about it. But on the other hand, there's a point of view that it's really a lot of litigation is a tax on doing business. And we know that there's a lot of frivolous claims that are brought as well. So you ask different people, you'll get different responses to that.

Mm-hmm. Right now, you have these, you know, gigawatts of compute being brought online. And in a lot of instances, there will be like a hyperscaler that is signing up for a long-term contract. When you look at some of these NeoCloud businesses, a lot of people like to look at them and say, "Well, look, this is a contract with a major trillion-dollar company. That trillion-dollar company is definitely good for it." Nvidia's back there somewhere, guaranteeing performance as well. Nvidia's usually involved. But in actually playing some of these things out. Let's say in the future there's a compute glut and there's— I think you should debate when that will happen. Maybe it happens in 10 years, maybe it happens in 1 year, maybe it happens in 2 years. No one knows. There's a lot of indicators that, you know, you're just gonna see demand outstripping supply for a long time. But what do big companies do when they're in a contract that they really don't like and they want to get out of it even though they signed on the dotted line and it's a bulletproof contract? Because it's not like, you know, the biggest company in the world that people are, you know, has amazing credit right now that signed the contract, people are like, "Oh, it's fully guaranteed." But there's a lot of things that companies in that scale can do if they're in a situation that they don't like.

Well, I mean, It's tough. They can— look, if there are big obligations on both sides, big commitments on both sides, and one side doesn't want to perform, a lot of times that precipitates a negotiation because they need each other, right? They each made commitments to each other. Yeah. So if you and I have one of these contracts and I have a problem, then you have a problem too. Yeah. So we got to try to find some way to work that out. And oftentimes that's what happens. Happens. But I think a lot of these structures, financial structures, especially for data centers, are incredibly complex. You have an SPV at the middle of it, has a deal with a tenant. You have all kinds of financing, bond financing, securitizations, private credit with cross-default provisions so that if somebody defaults, you know, the whole thing potentially goes down. I think on some of these data centers, we're probably going to see some reorganizations. You know, everything, everything— look at all the money that's going into this and all the data centers that are being built.

Yeah.

Are all of them 100% going to come online when they're supposed to come online and everybody's going to do exactly what they contracted to do? Yep. You know, I don't, I don't think that's going to happen. So there'll be some workouts and the challenges that those data centers face from from, you know, I've been told by one neocloud that the biggest problem they face is electricians, just getting the electricians to do the work. So, it's, there's a shortage of everything from electricians to power to chips, you know, to compute. So, I mean, it's not going to end perfectly for everybody.

Yeah, but you would predict billions of dollars worth of billable hours dealing with all of that. I think— is that a safe prediction? I don't know. If you have $1 trillion of this kind of like complicated CapEx investment, then you're probably going to get at least a billion, couple billion of billables.

Those are a lot.

Billions are a lot of hours. I know, but it seems like, you know, you have thousands and thousands, you know, of these projects.

Look, and if some of them go into reorganization and the equity gets wiped out, there will be negotiations. There will be what are called liability management exercises. There may be bankruptcies as well. And yes, lawyers will be very involved in those workouts.

There was some reporting recently that one group that is pushing back against the rollout of autonomous vehicles are trial lawyers, or, or sorry, not trial lawyers, sorry, personal injury lawyers who seem to be worried that in a world with, you know, full autonomous driving across the United States, there would be less accidents. Does that surprise you at all if that were to be true?

I hadn't heard that, but I guess nothing would surprise me about what the plaintiffs' tort bar might do.

Yeah. How would you reform? If you had— if you—

monopoly, so no corporations can sue each other.

Um, how would you— what kind of reforms would you push for if you were, uh, king of the United States for the day around specifically around personal injury law? Because when you drive around LA, it's honestly very depressing. It's like If you look up, you'll see a billboard that's basically saying, hey, do you want to sue someone? Yeah. And that signal to me is that there's just billions of dollars of revenue. Look, there's no doubt.

I mean, there are personal injury mills, lawyers that process high volumes of claims. I've never worked in one of those factories, but I sometimes you wonder whether lawyers— there's ever lawyers' eyeballs on what's filed and the discovery and the like. I think a lot of those cases are brought to settle as quickly as possible. So sometimes you wonder whether the client's interests are really number one. But I don't work in that field, so it's my speculation.

And it does feel like it would be a very different situation if every automobile accident involved going up against Waymo with perfect camera footage from every angle for every— there's no testimony of this person said this person, this person did this, this person said that. You have the perfect footage, the GPS, everything that happened in that moment.

Well, I think, I think AI, I mean, you know, there are early-stage companies, early-stage, maybe somewhere beyond early stage, that hoover up all the data available. Yeah, like permits, ingredients. Sure, sure. You know, compliance certificates, whatever. Hoover all that up. Yeah. And they also ingest all the regulations and rules, and they will identify claims and class actions. Yeah. And you can subscribe to this and they'll serve it up. Yep. So, I mean, AI is in itself generating more of that kind of work as well. But I also have the hope that it'll mean that cases are resolved sooner because there's more transparency sooner, as you were saying, like the Waymo situation where you have the record. Yeah. So people hopefully will be able to identify the merits of their claims. Everything's more transparent, identify the merits of the claims, you can underwrite to risk, and people may be able to get together with settlements sooner.

On the next generation of lawyers, do you think aggression can be taught?

I don't know if aggression can be taught. I don't know if it should be taught. Aggression in the law and litigation isn't an end in itself. I generally think it's good to be the first mover. I mean, we have a saying at our firm that the side that figures out first what will ultimately matter wins. So, as a default position, I believe you should try to get ahead and stay ahead. Now, that doesn't necessarily mean you're aggressive. But in some cases, you don't want to be aggressive. I mean, you might want to adopt a rope-a-dope strategy. You know, come get me. You know, and let the other side play their cards first.

How can you tell if someone's bluffing?

If you know their— if you have a sense of what their case is and what it's worth and what the strengths and weaknesses are, you sometimes get a demand that you think either they're bluffing or they don't understand their case. So, you know, it really very much depends on the— I think on the situation.

How did you, how good were you at reading people early in your career? Not very good. And how much did you improve? Like, did you have some sort of baseline above average ability, you think, and then you got significantly better?

I think with experience I got better. I think as a younger lawyer, I was probably too aggressive too often. And didn't sort of modulate my approach. Over time, I think I got better of understanding, not just understanding the other side and reading them, but myself as well, where I was coming from and what my strengths and weaknesses were.

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

Thanks. Thanks for having me, man.

Wish we had more time. Let's play you off with the final outro to Moskva Lawyer. Lay higher, John Quinn. The war is on and the verdict burns right here. I think every one of our guests needs an intro like that metal song. I think we're— I think that might be a new bit for us. First, let me tell you about Figma Agents Meet the Canvas. Your AI agents can now create and modify your Figma files with design system context. We've been keeping our next guest too long. We got Nikhil Reddy from Kaizen coming on the show. Let's bring him in. How you doing?

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

Legendary outro for John Quinn. Yes. Hopefully you don't have to hire him, but if you do—

Honestly, I was going to say hopefully you get a chance.

Yes, the opportunity to work with him on your side.

No, it's hopefully you don't have him sending you emails.

Ever.

That's for sure. That's for sure. Except maybe a client intake form. Yeah.

Uh, well, thank you so much for taking the time to come chat with us. Uh, kick us off with an introduction on yourself and the company.

Yeah, man, thank you guys so much for having me. Jordi, John, big fans. Um, quick background on me: I started Kaizen in 2022. Before that was an early engineer at Anderol, joined from the Bay, joined Anderol right out of school in 2019. Um, spent 2 and a half years there as an FDE, so worked on the web interfaces for a lot of the early hardware products that are now flying and doing wonderful things. And before, needed a UI and UX to sort of power some of those assets. And it was amazing opportunity to just go to a bunch of military bases and build really quickly for the warfighter. And then started Kaizen at the end of 2022, specifically with the, the fire to modernize all constituent services. I think that experience showed me how broken and disconnected government application layer systems were. But when you think about it beyond the context of just defense, I know we're here announcing a DOW project, but globally across all governments, all public institutions, I think the worst part of everyone's day is that 8:30 AM when you go to the DMV and you have to use the tech stack to renew your driver's license or pay for a permit really quickly or do the basic fundamental task. And if you can make that experience 10 times better in order of magnitude better, then I think everyone sort of leaves those buildings feeling like their taxpayer dollars are being spent appropriately and that they're being taken care of. And that was the vision, was to go to every single public institution and modernize their application layer that citizens and individuals interact with, I think more in a volume perspective, more than any other government service and make them 10x the quality. And that's sort of what we've been up to the last 4 years, selling to city, county, state, and now federal government. Announcing a big project with the Department of War.

Yeah, congrats. Okay, very cool. A lot, a lot I want to understand on that side of the business. But yeah, talk about the contract. Yeah. And, and yeah, parlay this into what you're announcing today.

Yeah, so we're announcing the launch of the counter-drone marketplace that is run by JIADA 401, the counter-drone task force within the Pentagon. The problem that we're trying to solve is that counter-drone assets are being manufactured by American OEMs, many of whom that you've spoken to and have had on the show, and the threat profile is changing so quickly, the hardware is being manufactured at lights-out pace. But then actually think about the process. Now that there's an amazing OEM product that needs to be fielded out into a forward-deployed region of the world, think about the paperwork, the process, the structure, the 50 different software platforms that someone within the building needs to use to authorize budget for it, get contract vehicles established, then go back and forth with the KO, negotiate that contract, then find the right logistics and management partner to get it on a pallet and get it delivered. You're talking about potentially weeks to months of delay when the warfighter needs that capability and that asset delivered on the pallet tomorrow, so they can actually use it in the fight. And that was sort of what for us was the most exciting problem to solve. And 401 put us on contract in 9 weeks from a blank slate, we designed, developed and deployed a true two-way e-commerce marketplace that looks stunning, rivals whatever you'll see in the private sector where American OEMs can sign on, establish their vendor profiles, upload their products, sell them directly to 3 different buying groups: Department of War capability offices, people with budget within the Department of War that want to buy now; state and local law enforcement, because think about the intra-government or intra-United States case where you're a state and local law enforcement leader and you're protecting MetLife Stadium for FIFA World Cup, you also need to buy counter-UAS. So it's a centralized platform for those buyers to log in and purchase those assets. And then partner nations, allied nations want the best technology. American OEMs make the best technology and they want to be buyers on this platform as well. So it's a two-way marketplace where buyers can come in, register and validate their profiles. Vendors can do the exact same thing with their products back and then being able to have the Pentagon have one core centralized source of truth. We've built the entire front-end architecture for vendors to post and publish their products, buyers to come in, purchase their transactions. A ton of agentic tooling where even if you're not the expert, you open up our capability planner, you specify in free text that you're protecting a stadium for FIFA World Cup. Here's my threat profile, here's how much money I'm willing to spend, and it immediately suggests a bunch of assets that you can purchase, add it to your cart in one click. And then on the backend, we're working to simplify all the tough, difficult, very frustrating procurement processes to actually get hardware on a pallet sent to the warfighter. End to end up and running in 9 weeks.

Uber for drones.

As a taxi driver. Sorry, exactly.

I had to say that. Every once in a year still, I'll see a pitch that's just like, we're building Uber for X. Wait, really?

You've seen an Uber for X?

Still see that. Wow.

Throwback.

It's going to come back. Every now and then. Dutch East India Company for drones.

How about that? As a taxpayer, do I want you to win or do I want a monopoly in software delivery for the government? Because everyone's run the business and they've had a bunch of different point solutions and there's a different world here where I'm like, ah, like this is cool that you're doing this. It's good that I'm happy for you, but also like, couldn't this just be like one extra module in a Palantir system that's already deployed. Like, why do we need— obviously, besides the competitive dynamic, just from a technology perspective, is it okay that we're gonna have proliferating systems? Is systems integration here not a particular problem? Are you addressing some problem that is too low for other contractors and primes to go after, so it would get neglected, and so that's why you're doing this project, like, help me understand how you fit into the landscape of other, other prime contractors. Yeah, fantastic question.

I mean, like, the answer is actually a little bit of all of the above, actually. From the data layer perspective, our goal is to build application software that is capable of integrating into a seamless, unified, coherent data fabric. I want to write data to a Palantir or a Databricks or AWS. Because that fabric is well-built, securely constructed, secure, compliant. But then you actually look at the market landscape. You're right, most of the market, I think it's actually dominated by two extremes. When you're thinking about the city, county, state level, largely what we see is super old, janky, private equity-owned point solution for blah that hasn't received an update in 15 years. The engineering team is a bunch of Java developers that found out—

some of those businesses are real cash flow machines. Okay, so yeah, let's not take— they're good businesses. No, no, no, but as a user, as a user and not a shareholder, uh, I, I, I share the frustration with the DMV system every once in a while.

As a taxpayer, you're, you're, you're asking yourself like, what are we doing here? And then on the federal side, most of those platforms are run by a bunch of low margin services companies that are selling hundreds and hundreds of heads and overcharging the government hundreds of millions of dollars. And then we'll spend 18, 24 months trying to build a thing deployed as a pilot, largely accomplish nothing. And I think the model that we're trying to apply is, yes, we should own as many application layer solutions as we can. And that's obviously sort of why Kaizen exists. But our goal is to charge based on the outcome and quality. And I would actually argue if we can make those application layer systems and an order of magnitude better for the taxpayer from a user-friendliness perspective, easier to use, more simple to log in, adjudicate your information, have it approved by an administrator on the back end. And then for the administrators that are spending 8 hours a day logging into these platforms, trying to triage and do all the paperwork for them, if you can make their lives an order of magnitude better, I think we're actually solving the real problem. This is why taxpayer money is appropriated to these kinds of services so that we can actually provide a quality of service for our citizens they deserve. And so for us, the number one goal is make every application layer product across these institutions as good as possible and move away from the point solution for X model and rather go to our customers and say, what is your need? Are you actually trying to make the DMV licensing process better? Or in JATA 401's case, they need a Shopify-style system for counter-drone procurement. And then we should have the technology stack and the set of modules that we can use to rapidly configure against those capabilities and deploy in 90 days or less. And still, I think you can move very quickly and build very quickly as we're seeing in the world today, but also hold a high standard for compliance, security, good guardrails, testing. I think you—

we live in a world where I think you can have both things.

And I think that's sort of what's been super exciting about our federal progress.

Love it. Well, congratulations and thank you so much for coming on the show. Thank you for your service. We'll talk to you soon. Thank you so much, guys. Yeah, goodbye. Up next, we have Art Levy, the new Vice President of Global Partnerships at Cognition. We got Russell Kaplan, the President of Cognition. Um, they're coming in to the TVP and UltraDome. Very excited to have some folks joining. Congratulations.

There they are, the trade— in the corner office, in the corner office, fit for a new Vice President. It is a corner Office.

Yeah. Where are you guys? In the South or Bay Area? We're calling from Austin.

Yeah, calling from our Austin, Texas office. Austin, Texas. We got a great team out here. Yeah, super excited to be having Arda on the team.

Okay. What is a global partnership? Does it have to be global? Is it just a sales deal or is there a deeper partnership that happens with these companies like AWS, Apple, Slack, Navant, Oracle, Capital One? Break it down.

John, you nailed it. Global partnership means it's not just a sell-to deal. It's something where we're both selling to the business. Potentially there is an investment. Potentially we're co-selling to their customers. We're selling to them, we're selling to their customers, we're doing co-marketing, we're doing a, a press release. Potentially there's a product integration. It's about deeply connecting two businesses who share a common goal around, you know, acquiring the same customers, or we build inside of them a product that is just better for their customers. And then both parties win. You know, classic partnerships, 1 1 3.

I think of Cognition as an AI lab, maybe a Neo lab, a harness developer, a consulting firm in some world.

I think of it as a source of geopolitical power. I guess. I guess. That's just me personally.

Yes, it's birthing a new god as well. But what is the conversation that you're having Fortune 100, these huge companies. How are you positioning your role in the AI ecosystem, in the business ecosystem? What is this full scope of the value that you can deliver to customers?

Yeah, I think as we've grown up and evolved, you know, I think we've really realized our role is to be the independent agent lab. And so, you know, we're an applied AI lab. We're focused on building agents and we want to give every company the best possible agents to, to go solve all of their software engineering problems. And so, you know, whether it's a Fortune 500 or it's a, you know, it's a small, fast-growing startup, you know, people have a lot of needs for, for software. And Devin from the beginning has always been designed to work with all the other tools you already use. Right. It shouldn't be, you know, it's the Devin way or the highway. Like Devin should join you like a teammate. And I think that's one of the reasons I'm super excited that Art is joining is Art built an amazing alliance and partner ecosystem at Brex over his like very long career there from employee 30 to chief business officer. And we see an opportunity in combination to bring not only, you know, it's like it's not just, hey, here's the tool, go figure out. It's like, how does this work super well with every piece of technology you have inside your organization?

Are large corporations that are maybe they don't have an in-house AI lab Are they generally of the view AGI is real, AGI is soon, but we're going to go through a sort of slow takeoff? Diffusion adoption will be a big problem and Cognition can help sort of realize that thesis.

I think the pace of progress in artificial intelligence is dramatically over, over the pace of progress of internal organizational redesign. For sure. And so, you know, and by the way, this is true not just at the big Fortune 500s, even the sort of startups or tech companies we would consider as fast movers. Everyone is asking, you know, do I really need to rethink everything now because of AI? And I would say it's actually one of the, one of the things that's been lucky for us at Cognition is we only incorporated in January of 2024. So our internal, you know, team setup and structure is it's, it's very native. But every company is trying to figure this out for themselves. And I think if you, if you froze capabilities today, you'd have at least a decade of crazy progress in the, in the full enterprise world just to get this stuff really rolled out and operationalized.

So what are it like when you're talking to an enterprise? Is, is, is it better to position as the, the, you know, you're going to help with this diffusion, this value creation, or are we still in that sort of, maybe it was like May, June, July, like, whoa, we went a little bit too far with the token maxing. Maybe Cognition can help me rein in this budget, actually make sure that we're getting ROI. What's resonating there on the cost to benefit position?

I think it's both. I think the third too. So first, you know, all the models have been getting better, right? And so as the models get better, uh, the capabilities keep improving. Uh, yeah, people are starting to ask the question, hang on, like, do I need to drive the Ferrari to the grocery store? Uh, you know, can I, can I use an open-source model for parts of these workloads? Can I use cheaper model, faster inference?

And so that's definitely one— driving a Ferrari to the grocery store, but continue.

If only, if only we could all Be Jordi Hays. But, but, you know, I guess one part of it, I also think there's a separate thing that's emerging, which is if you actually want to have true frontier capabilities, you know, different models have different spikes, right? Like we've seen, for example, we've seen the OpenAI models, for example, are, are the best in the world right now for, for recall of security vulnerabilities. Like if you want to catch, if you want to catch as many as possible, the OpenAI model series is best. There might be others where actually, you know, the Anthropic ones are a little bit better on precision, you know, getting it really, really tight in. So, so I think there's, in addition to the price performance angle for enterprises, there's this element of, well, for how do I use the best tool for the job, you know, for each job to be done. Yes. And then there's the people, the people change side, which we try to help with too. I'll give you one, one story on this. You know, we had a customer, global, you know, Global 2000 customer. Tens of thousands of engineers, and they deployed Devin across the entire organization. And they were seeing, you know, crazy, crazy productivity gains. But one organization they deployed to was the product management organization. And so every product manager now had Devin. And what they were doing was this organization's job is to basically take in customer requests and turn them into Jira tickets, roughly. It's like scope out what the customers want, turn them into Jira tickets. So they were using Devin to— they took those requests And then they had Devin actually go prototype the fix. And then once Devin prototyped the fix, they had Devin say, they said, Devin, can you summarize this fix in a Jira ticket? And then they deleted all the code. And we asked them, guys, like, what are you doing? You already had the fix. And the product manager we talked to said, that's not my job. Like, I'm part of the product management organization. I'm not part of the engineering organization. And so I think like for really large companies who've been doing things the same way for a long time, there is like a total rethink that needs to happen. It's just literally how do you design, you know, how do you design your structure?

Yeah. How are you guys thinking about M&A? You bought Poké recently. I think, I think, I think it was surprising but made a lot of sense. It was surprising to me because I figured someone like an Apple or something would, would see Poké's, you know, talent and their kind of consumer insights and all that kind of thing, try to pick them up. But how are you guys thinking about M&A from like where the value really comes from? You guys are very good at building software. That's your whole pitch, right? Is we help you build software. So I can imagine a world where product acquisitions are like less interesting because you're looking at your own roadmap and you're looking at what someone's built and thinking like, well, how quickly can Can we build this ourselves? Is it, is, are we, should we expect like more sort of like talent style acquisitions? What, where, where, what's your general framework and maybe what's, share as much of your mandate as you can, Art.

Yeah, I mean, I would say all of the above. I mean, we have today at Cognition, 97 former founders working in the company, 2.5-year-old startup, and we are, we are like aspiring or actual hiring. Why didn't one team a week probably right now? So if you're, if you're a strong founding team and you want to be part of a rocket ship, you know, get in touch. We love bringing on founders, but we also think there's, you know, there's bigger companies out there with, with products, with distribution, with, with learned insights that are really useful for us that we want to bring in as well. And I think how you do that successfully, a lot of companies screw up M&A. We kind of learned, learned it trial by fire acquiring Windsurf. You know, in the course of a weekend as a company that was 4 times our size. And, you know, from like first call to, you know, execute a greeting from Friday night to Monday morning. So it's something that I think we're only going to get better at. And one reason I'm really excited Artis is here is, is to help us kind of level up that process and be the best destination in the world to be acquired by.

Yeah, I think one of the things I've seen work well in my career is a lot of times partnerships can be a preempt an M&A. So a bit of a try before you buy. You make sure that the cultures really make sense, that two companies build an integration or a partnership. And then suddenly in those conversations or after you delight a few customers, you both decide, hey, actually maybe it'd be better if, uh, we work together. And I think given the scale of Cognition, our rate of growth, the impact that founders can have when they come here is really gonna be generational. So I'm really excited to, you know, talk to a bunch of founders and hopefully bring a bunch more of them onto the Cognition team. Awesome.

What country internationally outside of the United States is punching way above its weight in terms of AI adoption, AI diffusion? You always hear that story about Estonia rolling out high-speed bandwidth to the entire country, and they just went way, way forward in terms of GDP per capita. Uh, what country are you having like surprising success in Japan?

I would say Japan, number one. We love Japan. Look, we're a global— we're a global company. So we have people all around the world, we have customers all around the world. But I got to— I got to shout out Japan in particular. You know, even from, from the beginning of when Devin first became available self-serve, when, when cloud agents were like just at the edge of possible, we had an enormous uptick of users in the Japanese market. And I think there's a few, I think there's a few reasons. I think one of them is, you know, our Japanese customer base, everyone is incredibly conscientious and detail-oriented. And whenever we get a support ticket from Japan, it's a very thoughtful essay of every single issue that has been tried, every piece of debugging information you would need. And like, they just found a bug. They just found a bug. If they could have solved it on their own, they would. And so I think the user base in Japan is super sophisticated, really early adopters, and, and it's been exciting to see all the growth there.

I prefer the American style of just fix it, fix it. Yeah, it seems, it seems like really beneficial. Sorry, you were going to say something?

No, I was going to say, as I'm digging in also, ironically, on the channel sales side, we've actually had the most success with sell-through with partners in Japan who seem to just deeply understand the product and the value prop and be able to sell it to their customer bases.

Interesting. Do the Japanese love otters by any chance? Could that be a factor?

That's a good question.

We definitely have a lot of otters. We see the plushie.

Yeah, we got a lot of otters. There was a big debate in Japan of the right honorific for Devin. You know, is it Devin-san, Devin-kun? And then we learned that the community in Japan has settled on Devin-kun. As the right honorific, you know, your, your friend, your bud, your, your sidekick who's helping you out. I think it's a, it's a fitting name for—

and that came from the community, this was not internal? Came from the community, yeah.

We were informed, we were informed.

Ground up. Amazing. Okay, well, thanks for breaking it down. That's very— love it.

Well, I'm excited to see everything that you guys do together. Yep, deals, deals, deals.

Very excited. Great to see you, John. We'll talk to you soon. Great hanging, guys. Take care, Dad. Congrats. Goodbye. Let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents, whether you're writing code, analyzing data, creating content, or automating business workflows. Codex helps you move projects forward from start to finish. We have Brendan Carr from the FCC joining the show shortly. It's his first appearance, but we met years ago. I've always been fascinated by his perspective on everything that the FCC does. They're obviously expanding 5G. We talk about satellites and space internet, what it takes to launch a rocket. I know you've been thinking about the TBPn satellite cluster.

He's the guy to talk to. Perfect. TBJ over on the X chat says literal art of the deal right now. Wow. Insane nominative determinism. Of the deal.

While we are waiting for Chairman Brendan Carr to join, we got to go through this crazy historic analysis from Ed Zitron.

He— Ed Zitron, July 29th, 2024, gave Sam Altman the perfect roadmap. He said, I am hypothesizing that for OpenAI to survive for longer than 2 years, It will have to. And then he laid out pretty much— Exactly what happened. Exactly what went down.

But looking back on them, they did seem like crazy, crazy things that had to come true.

I think Ed should join an AI lab and— With these predictions. Maybe as like a chief strategy officer. Maybe. 'Cause he's imagining everything that could go wrong, but through that, he's imagining everything that could go right. Well said. So if he put his optimism cap on, he might make a great chief strategy officer. Yeah, maybe, maybe at DeepMind. I think that'd be fun. Maybe they could pick him up.

That'd be great. Well, Chairman Brendan Carr is in the waiting room. Let's bring him into the TBP and Ultradome. How are you doing, Brendan?

I'm doing great.

It's long overdue. Thank you so much for joining.

How are you doing? I'm so glad to be with you guys. I love the show. Love what you're doing. I dressed down a little. No tie. I kept one extra button compared to you. I didn't go the full—

there you go.

You know, preview, but take 3 down. The Cougan preview.

The Cougan preview. That's an amazing line.

As a bureaucrat, you know, someone that regulates broadcast airwaves, not you guys, but broadcast airwaves, you know, decency still matters where I come from.

So we wanted to have you on because we wanted to push for podcast safety. Yes. We think that it might be a little bit irresponsible for the US government to allow just anyone to grab a microphone and get on the airwaves.

Yeah, what's it going to take? I want a multi-year approval process for new RSS feeds.

I do. Maybe like annual, you should maybe need a million-dollar fee annually to sort of maintain that license.

Yeah, I mean, TPN is doing so well, we need to build a moat around you guys, you know, make it harder for others to do this. Yeah, I think they call it regulatory capture.

That's what I want to know. Let me—

we're not here to talk about podcast regulatory capture.

No, we are here to talk about robots first. Or where do we want to start? China, different supply chain pieces. What's newest in your world?

Well, probably some of the newest stuff we've done. When we started, we stood up a new Council on National Security. We've been doing top to bottom review of all bad actors, foreign adversaries. The most recent thing we did just a couple of days ago was we added to our covered list So the FCC, there's no piece of electronics that you can use without going through FCC processes. Yeah, putting on the covered list means new models of that thing can't be imported or sold in the US. So we've added a couple of things to that. We did advanced robotics. So think humanoid robots, think those quadruped robot dogs. And the idea is, if you look to the future, those advanced robotics are going to be key to our economic security, to our national security. And President Trump is looking around the corner and saying, you know what, probably a good thing if we don't become dependent from a supply chain perspective on a foreign adversary nation or a bad actor for that stuff. So we're trying to onshore that. We've done similar stuff before with drones, for instance, at the end of last year we added all foreign-produced drones, but it's a balanced approach, meaning it's new models. So we're not going to take anything out of your hands right now. You can still buy, quote unquote, new versions of existing models. But it's a way of signaling to the market like it's time to adjust, guys. We need, you know, a safe domestic supply chain. So it's good from a NatSec perspective, but it's also great for investment and jobs in terms of onshoring domestically as well. Yes.

So you said it's good for the market or the market will adjust. I'm interested to know what is your process for assessing the capability of the American supply chain to actually step up to the call when there's, you know, basically there's going to be a gap in the market around drones, humanoids. Fortunately, it feels like we're not in a situation where I'm talking to entrepreneurs who say, look, I set up this business years ago. I am dependent on buying this particular thing from China or internationally on an ongoing basis. And this ban is going to throw me way, way off, like what might happen with other pieces of the supply chain. So there are— there isn't a whole industry that's dependent yet. But what are you doing to actually understand how quickly and how capable the American industry is to fill the gap that, that will be created by this?

Well, there's a couple of ways we take that into account and I'll walk through them. But just as context, for instance, when we added Drones to the covered list, we saw something like $5 billion flow newly into US domestic drone manufacturers. They're now making drone motors at scale in the US, which we've never really done before. But the balanced approach is this. So again, if you're buying a motor or a battery from overseas today, you can still continue to buy that because the prohibition is only on new models and the companies overseas can continue to produce the old ones. And then again, there's also an exemption process. So for instance, on drones, we went through— the relevant agencies went through and exempted toy drones, for instance, from the ban. And so it's balanced. That's forward-looking. Let's not develop too much dependency. Okay, maybe there's a couple thousand of these, not a huge deal. We don't want a couple million. So the entire regulatory structure allows for exemptions plus transition. And then look, we can come back afterwards. And remove the, quote, grandfathering for existing models. And in fact, we're doing that now for things that were added to the COVID list before 2024. So things like Huawei, ZTE, that type of equipment. Again, one, it's not consumer-facing as much, but two, you had multiple years to adjust to where we think the market's going.

Do you have any optimism around these, you know, frontier AI models have incredible cybersecurity capabilities. Capabilities. I could imagine a world where before there was a lot of fear, uncertainty, and doubt about certain pieces of electronics that are made internationally, potentially having backdoors or spyware. And if you can hammer one of these, you know, one of these systems with every possible test, are we going to wind up in a world where we can more confidently say yes or no, does something have backdoor, or is this just, you know, everyone will just be stuck in a never-ending race and we'll be back to where we were?

Yeah, I do think it's helpful from that perspective. And obviously there's telco companies that are on the early release lists for a lot of these models, help identify any patches that might be necessary. But we're doing a couple of things that are new as well. So it used to be if you wanted to sell electronics in the US, you had to go through a lab. The lab would test it for power levels and interference, very basic technical stuff. We're now inserting national security checks in that lab process. In fact, we stood up something called a Bad Labs proceeding because it turns out that something like 70% of all electronics that ultimately arrive in the U.S. are tested at labs located inside China. And some of those labs themselves were, you know, linked or controlled by or tied to the PLA. Yeah. So not necessarily the most trustworthy. So we're kicking many bad labs out of the testing process, and again, it's helping to onshore some of that testing process. So we do want to make sure throughout all of this that there's much more trustworthiness checks along the way. That's very interesting.

Jordi, do you have anything else on this?

No, I, I was, I was, uh, very excited to see the news from last week, or maybe it was the week before, just because I Even processing the drone industry and even the consumer drones, personally, personally, I think drones are great. I've seen drones that have captured some cool footage of snowboarding or surfing and all these things. But the idea of having, you know, millions of, you know, if you could ask any foreign adversary government, hey, would you like to have millions of, you know, flying cameras in in one of your geopolitical adversaries within their borders, they would say like, absolutely, that sounds great, let's do that. And so the idea that we were gonna let the same thing happen, that the thing that we let happen in drones, that we would let that happen with humanoids, which are much more of a risk to me. They can actually have, you know, they're not very significant today, but they can have a real effect on the world. And the idea that we would allow, you know, millions or eventually even billions of these into our borders just was very concerning. So personally, I'm glad that you guys took action.

I'd love for you to explain the E-Rate program, what's going on there with school internet. I think it's an underdiscussed story. What— it sounds very good on the face of it. You want students in America to have internet. How are things changing? What's the history of the program? What's the story going forward? This one's interesting.

So this country, the last couple years, I think we really experimented with going all in on screen time for kids, and the results of that are starting to show themselves, and the results aren't great. I mean, we have actually gone through one of the most significant losses of sort of human capital from a learning perspective that we've basically ever seen. I mean, going back to almost the Dark Ages, when you look at some of the decline in test scores and to some extent, I don't know if it's correlation or otherwise, but it coincides with this big uptick in screen time. And the FCC has a role to play. We fund internet connections and to some extent internal connections at schools all across the country, multiple billions of dollars a year. And so we've done a top-to-bottom review of our program to make sure it reflects the current research on screen time for kids. And it's not just us. Obviously, you're looking at a big movement around the country. Different states, different school districts are all starting to say, you know, maybe put the smartphones away, maybe put the screens away. I mean, you got young kids, kindergarten, first grade, they're just swiping all day. Uh, the idea—

this, you'll love this— a school nearby me, uh, elementary school, in PE, they spend more than half the time on iPads looking at physical exercise, like diagrams. I'm like, these are children. Maybe just let them run around outside. Study. Yeah. Exercise. Yeah. Odd. I think we've gone too far.

Get us up to speed on what's happening with Starlink and other space-based internet connectivity programs. It feels like, uh, much to the benefit of the American internet consumer. The race is heating up. There are more companies launching constellations. What are your KPIs? What do you want to see? Is it faster approval times for new constellations, clarity around who owns what spectrum, smoother auctions? Like, what is on your to-do list for the next couple of years? Yes, a little bit.

There that we're doing on every the points you hit. So for one, President Trump came in and gave a very clear direction. He said he wanted the, the United States to lead the world again in technology and in space in particular. We're now seeing that. So there's this new technology called direct-to-device where you can go straight from your smartphone right to a low-Earth orbit satellite. You don't even need that little Starlink mini or other satellite provider dish near you anymore. It's just starting to emerge, but we're putting the regulatory framework in place at the FCC for that to succeed. One, that means more spectrum. You need new airwaves for that technology to work. We've been facilitating many transactions to do that. Starlink, that was a big chunk of spectrum. Amazon is buying another company called Globalstar to basically take their spectrum and use that for their own version of direct-to-cell. And then inside the FCC, we're taking our Space Bureau, which historically processed applications for a lot of this stuff, and they would review individual applications and long narratives, be very subjective. We've scrapped that entirely and we've replaced it internally with an assembly line where we give you clear rules of the road. If you hit these targets, your application's been on the fast track and you're going to go because again, look at the launch cadence, look at the amount of mass going into orbit. We got to keep up with that. And this is someplace that I'm just super excited for the country, this hard tech area. These, you know, the kids from the gundo that are now everywhere outside the gundo as well. I mean, what they are doing right now is a I think a huge part of the future of this country, and we want to make sure the FCC is not sand in the gear, uh, when they continue to innovate the way they are.

Does, what, does the FCC have any other roles to play in that? I mean, we've seen like every day there's a, there's a company, a startup that's coming on to announce an LOI or some sort of deal with a different piece of the government. Are there, are there any other ways for for the startup ecosystem or the tech community to interface with the FCC beyond, okay, I got approval for this thing?

Yeah, we're working across the board to make sure that, you know, whether you're big or small, medium-sized company, that you can get a yes or no answer from the FCC. And part of that's just getting rid of deadwood. So we started this proceeding at the agency called Delete Delete Delete, which is about going through every single page of our rulebooks and getting rid of anything we don't need. We've obviously sort of resulted in, you know, thousands of words being taken out. I think 400 pages have been removed from our Code of Federal Regulations already just so that, you know, whether you're small or unsophisticated, it should be much easier to engage with the government. We've cleared the application backlog in space, you know, by about 50%. So we're seeing some good results there for everyone to engage with the agency.

That's very exciting. Jordi, anything else?

Not for now, but—

Thanks so much for coming on the show.

Have a great day. Honored to have you.

Have a great week. We'll talk to you later. Good to see you guys. Thanks. Good to see you. Goodbye.

SpaceX had earnings.

Oh, yeah. How'd they do? Let's see.

First-ever earnings as a public company. Up 10% right now? Nope. Down— well, it was up 9%. It was up 9.4%. Down 4.5% after hours. Okay. First up, SpaceX announces a new partnership with NVIDIA to design its StarMind AI-1 payload, bringing data center class compute into orbit. That makes a lot of sense. Let's find some actual numbers.

So revenue was $7.81 billion versus $6.93 $3.2 billion expected loss per share was $0.09. Average analyst estimated a loss of $0.26. Revenue jumped 92% from $4.1 billion a year earlier. So huge. You know, you're even at this scale still doubling the revenue. And it's the first time Elon Musk's reusable rocket maker will face Wall Street. In this capacity, and investors are jittery. SpaceX stock has dropped 12%, uh, 16% since opening at $150 a share on June 12th. Um, and SpaceX lost $4.9 billion last year, largely due to heavy investments in artificial intelligence, which we've discussed. Uh, the company merged with Musk's xAI in February. CNBC reports saying that, ask, at the time, The vision was to build data centers in space, but the launch business, which counts on large contracts from NASA, is losing money. Most of SpaceX revenue for the year and its only source of profit came from its connectivity segment, which consists of its Starlink internet service. Starlink is sold directly to consumers. So here's how SpaceX performed in the three key segments. For space, they brought in $962 million. Versus $835 million, which was expected. So they beat in space. On connectivity, they brought in $4.29 billion. So the Starlink business is 4— more than 4 times the size of the actual launch business. They brought in $4.29 billion versus $3.83 billion that was expected. So they beat there. And on AI, they brought in $2.56 billion versus $2.15 billion. $8 billion expected. Beats across the board. Very interesting to see that the thing that they started doing, space, of course, launching rockets, is now their third largest line of business. Connectivity is, of course, bigger. Also, AI is bigger. Everything is computer, John. Everything is computer. That is a good summation of it.

That's a good place to end our show.

Let's play you out with the most feared lawyer, John Quinn. Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter at tbpn.com. We will see you tomorrow at 11 a.m. Pacific. Goodbye. Flashback. Feel in the air. Names go pale. The room goes still. Then the verdict. Victory!