20VC：Musk 的万亿美元 $TRN 薪酬方案拆解｜Ramp ARR 达10亿美元、Brex 达7亿美元：谁赢｜OpenAI 100亿美元二级出售｜Atlassian 以6.1亿美元收购 The Browser Company｜ASML 领投 Mistral，估值140亿美元

The real truth is, the buyer has cunningly eviscerated the brains and the heart of the company and left the carcass, and we're gonna pretend it's real, but it's dead as the dodo. Everyone knows it, but no one's gonna go on the record saying it.

Scale was for sale for $28 billion. Brett's gotta be worth $56 billion.

I didn't leave Hollywood. Hollywood left me. This is venture capital today.

Jeff, it is awesome to have you with us. So Jeff, first off, thank you for agreeing to do this with us today.

Great to be here. Thanks for the invitation, Harry.

Now, we're gonna start with the most pressing, or a big element of news, which was Musk and the first trillion-dollar pay package. It breaks all prior benchmarks in terms of a trillion-dollar pay package. I'd love to hear how we thought about this and whether this is a new normal that we should be expecting to see for your Sam Altmans of the world, or whether this is a one-off exception. Rory, I know you love it when I go to you first.

1. Musk Doubles Down

Yeah, I like this question. It made me think a lot, and I did a little work, and I think there's a lot to unpack here. Let's start with the first thing. I always say this: compensation is how boards reveal their real priorities. Nothing else matters as much. So you can tell everything about what the board wants in terms of how they structure the CEO package.

There are 2 or 3 things here. I read the proxy about it—it's 322 pages. I didn't read all of it, but I got through 100 or 150 pages. What's clear here is this: the board wants the Elon bet. That's just super clear. We feel they owe him the past, and they're gonna give him the future.

The second thing is they really did believe that if they didn't give him the extra 12% on top of making good the stuff that was disallowed in 2018, he might walk. That's in the thing every time. Rightly or wrongly, you can argue that, but that's clearly what they believe in the proxy. And then the third, I think the most interesting thing, is they're really paying him to double down again.

The trillion-dollar headline is obviously a big headline, but some of those operational and market-cap goals are huge. It's basically the board saying, “You know, we're the 8th- or 9th-largest market-cap company on the planet. We'd like to double down again and be by far and away the largest market-cap company on the planet. That's the bet we as a board wanna make, and Elon is the way to make it.”

When you look at it, you've got the market-cap metrics. I think $8 trillion is the maximum cutoff, so you've gotta make something with $8 trillion. You've got EBITDA criteria. You've gotta make $400 billion in EBITDA. For context, Google, the most profitable company this year, makes $100 billion, so you've gotta make 4 times more than Google.

And then I think the 4 hard metrics—which are what I always think of as the most interesting metrics, not the money metrics, but the “what are you gonna do?” metrics—are 20 million total cars. That's doable; they've already done 10 million. 10 million in FSD feels doable, because why would you have a Tesla and not get FSD? I love my FSD. I'm a crap driver.

But then the other 2 are, I think, 1 million Optimus robots and, I wanna say—don't quote me—about 1 million robotaxis. When you look at those criteria, they're basically saying, “Double the existing business, but on top of that, build a whole new business on top.”

This is a board doubling down on Elon. This is the bet they wanted. I mean, we can discuss whether they should want it or not, but it's actually intellectually very clear. I find it pleasing. If you wanted this bet as a board, this is exactly how you go and buy this bet from Elon. You say, “I will give you a shit ton of money if you take our trillion-dollar company and you double or quit it.” At that level, it was like, yeah, I get it.

Jeff, what was it like on the other side of Twilio with your comp? Thinking back to your comp package—maybe they didn't offer you a trillion; I don't remember—did you have one of these crazy packages? They're becoming more common with startups now. In my portfolio, I see them.

No, I actually never wanted major comp. As a founder, I had ample equity, and I always thought comp, generally speaking, just distracted—all the time spent on that.

At least for me, one of my principles of compensation was always that the more levers and knobs and things you put into a comp package, the more opportunity there is for someone to just think it's unfair. The idea was, once you pass the bar of fairness, this is Daniel Pink's whole philosophy from his book *Drive*: once people believe they're paid fairly, they focus on the work.

The only thing you can really do with all these knobs and levers and variable comp packages is take someone who thought they were compensated fairly and suddenly make it feel unfair because, “Oh, we missed that metric,” or, “I did my part; they didn't do their part,” whatever.

And so I always wanted compensation as simple as possible for the team, and that also went for myself. Whatever they gave me, I just said thank you, and that was that. The more complicated you make these comp packages, the more shit can go wrong.

Dario, Sam Altman, Mike Truell at Cursor—you name any of these great founders who are very pivotal to a business—could look at this and go, “Well, this is a new benchmark.” Do we see this as a turning point in how we incentivize CEOs at scale, given how central they are to businesses, or is this a one-off with Elon?

I think this is the new standard for anyone whose board consists of their brother-in-law and other relatives.

It's a good point. I find boards are more and more created by founders. They're more and more great jobs, and everyone wants to get into the deal. I think everyone's got, in a sense, their brother-in-law and ex-boss on the board.

In almost all my portfolio companies, the founders control the board, not just from a cap table perspective but from a relationship perspective. They control it. I don't know what you guys see. I see all these deals happening.

Once you've crossed a unicorn, which now is like a Series A, every founder CEO is getting some massive upside package with massive goals. Instead of a top-up for 2 or 3% after you've struggled for 5 or 8 years, they're getting 7 or 8% or more, but you've got to have a massive outcome: $10 billion, $20 billion, $100 billion. It's becoming the growth VC playbook, for right or wrong.

And it's definitely happening. If you go back to the first Elon comp package in 2018, there definitely was a wave of wannabes that copied that in the 2 or 3 years after that, through the end of ’21. It wasn't everyone, to your point, Howie. My guess is 10% or less of CEOs went for it.

The first thing is, it was typically the CEO asking for it. For some CEOs, something like this becomes important and motivating, and you can't ride every horse the same. Some folks are like, “This is the core of who I am,” and I think Elon is the ur-example of that. I don't particularly love that style, but it's like, “It's all about me, and I am the most amazing person in the universe. Compensate me accordingly.”

To some extent, as a board, you're left saying, what do you do in that circumstance? Now, to Jeff's point, you could have done the bluff game and said, “I don't think you'll leave if we won't pay you.” My guess is they were angsty about that. I'm not saying rightly or wrongly, but that's the thought process.

Once you think someone can leave—and, by the way, the one person who can leave is someone who has 3 other gigs that are equally exciting, which is why your leverage is lower in this particular case—you've crossed those 2 doors. You have the person who wants the egotistical win, and he might leave without it, and then you're left with a negotiation exactly like we just saw in Elon's case.

I don't think that's the norm. That will be the high-water mark, not the norm, but I definitely think, like all high-water marks, it will push up other people's demands and aspirations.

Rory, have you thought about the fact that maybe this isn't about upside? You made the upside case, that this is all about building a multi-trillion-dollar business now. What about the downside case? The downside case is that Tesla is overvalued and that it's all the cult of personality of Elon Musk that creates that value, and if he leaves, the house of cards comes tumbling down.

I totally agree. There are 2 risks. One is, if you were valued as a car company, you'd be valued at around 25% of where you are now. Maybe this is a car company worth 25% of our current market cap, and then Elon's special sauce, which is the other 75%.

As a board, you probably feel a huge amount of pressure to keep that person. You're exactly right, Jeff. And the other thing is, yes, because the other option you could take, which they're clearly not, is to say—I mean, this is going to sound really hard-nosed—“You've built an amazing car company. We're a car company. Let's manage it like a really great car company. Let's accept that we don't want these future bets,” and let it go. You could get a different person to manage that company.

You're right, the problem with that “negative vision” is that the stock would be down 75% the next day. The individual shareholders—who, let us remind ourselves, have re-voted the prior comp plan twice when they didn't have to—are like, “The people who own this company want to make this bet.”

I personally find it a little terrifying. It's such a risk-on bet that it makes my head hurt. That's what they want to do. And you're right, if that bet were canceled, if the board said, “We're not going to make that bet; we're going to play it safe,” Jeff's exactly right: the stock would go poof.

As a board, you would be dealing with lawsuits from here to the end of human time. So it really is a prisoner's dilemma. It's kind of a scary board to be on. I admire their courage.

You get compensated well. I think they have one of the best board compensation packages on the planet, but it must be a really odd dynamic negotiating with Elon, knowing that, as you say, Jeff, if you try to demonstrate resolve and he threatens to walk, you're down 75% the next morning. That would be tough.

2. Missionaries Versus Mercenaries

The tough thing is, we've changed so much in tech in the last 18 months. AI—the AI greed, which is not all bad.

Yeah.

What Jeff said is how I felt as a founder. Our generation—it's not that long ago—I've got enough. If I'm in the double digits, don't get me wrong, I'd love more. I wish I hadn't taken all that dilution in the seed round. But this is about a team. I'm driving a team on a huge journey. Leave me alone, take care of my team, I've got enough, right? It'll work out.

I think that's what Jeff said. That's how I felt. I don't hear that too often anymore. I think we've changed. When you're Cognition and you go—and this is amazing—you go from nothing to $10 billion in, what, 18 months, Harry, or whatever.

This isn't the Twilio grind or the StubHub—I mean, Jeff was at StubHub before that. They're finally IPOing now. This is a different world, and folks are building great teams, but they're mercenaries. “I want to cash out at OpenAI for $10 million after 12 months.” It's not bad; it's just I don't hear what Jeff said much from the kids these days.

Well, can we even just talk about the idea that founder CEOs quit to go join Meta?

It's crazy, right? Or OpenAI.

What is your take there, Jeff?

What's the policy on profanity on this podcast?

Are you allowed to say anything you want? I'm British, so we swear all the time. It's good.

I'm accustomed to founder CEOs being the most committed, most long-term-oriented, and most visionary of the group, and that's what makes startups great. And so when the person turns out to be just a mercenary and will go anywhere for a higher paycheck and leave the rest of the company that they started and run to flounder, I'm like—as Jason said—this is a whole different world of why people are in the startup world to begin with.

For a lot of folks, and I put myself in this bucket, it was missionary. It was like, you believe the world needs this thing, and I believe that that's the best reason to start a startup because if you just want to make money, probability-adjusted, you should just go get a job at a hyperscaler. Probability-adjusted, you'll make more money.

So you don't start companies to make money; you start companies because you love what you're doing and you think the world needs to have the thing you're building.

Do you think then Alexandr Wang is wrong? I mean, his investors made a huge amount of money. It was a great deal for them. It was a great financial deal for him. Was he wrong to do that?

I think it just shows that it's a mercenary move.

It is, but are we old?

Maybe. Maybe.

But what about being—

Because I don't see it anymore, Jeff. I haven't seen it since 2022. I don't see it.

Okay. So first of all, yes, we are—yes, you are old, Jeff. I mean, that's an objective fact. I didn't think I'd be defending the mercenaries here.

No, that is literally subjective.

Okay, we're going to have to put a pin in that one. But you're right, Jeff, you are right. My age is a fact, but whether it's deemed old is subjective.

Yeah, because from your perspective, I'm young.

You're totally right. By the way, that was another dig. You're 2 up. Okay.

And from Harry's perspective, we're all old.

Okay.

See how this works?

Let the record show, 20 minutes in, we decided gloves are off. I'm going to defend those 2 transactions.

First of all, I love your framing that you should start a company because you're not doing it for a rational risk-adjusted return; you're doing it because you have a mission, you want to change something. I would argue, from both Scale AI's perspective and Windsurf's perspective, the objective facts were that the offers being made—let's call them the attempted acquisitions, because I think in both cases the acquirer would have just bought the company if they'd been allowed to—were well in excess of the company's worth at the time, and probably, in my view, their worth at any time in the future.

So it's their highest and best exit, and to some extent they should think about taking that for themselves or their investors, for as many of their employees as possible.

I think it was primarily because of antitrust. In both those cases, they couldn't do the clean, "Here, we just own the company" thing. We had to do some kind of bullshit structure, and in each case it was weird.

The casualties from that were that you did blow the social contract for a number of the employees. I do believe, and I think the data has come out, that in both cases there was residual money left to make the payout. Remember, the people who get left behind aren't the longest-tenured engineering employees. It's typically people who joined in the last 1.5 years.

Maybe the total ownership is under 5% of the cap table. It would've been entirely possible to take care of them as if they'd been acquired while still doing this deal, and I'm not sure whether they did or not. It's in the murkiness of the underground chatter. If they did, then I would argue those founders did the right thing.

But if you look at the people who did really well from that—say, your Accel and Scale—they have the University of Wisconsin or Michigan, the Cystic Fibrosis Foundation, the Children's Hospital of Atlanta, and all of these amazing institutions that got back a load of money and are able to do things now for scholarships, education, and medicine that they couldn't do without that money. And 1,000 people at Scale who were in marketing or sales will now have to go and get another job at Cognition in the Valley.

I think that's fair from the VC's perspective. I think what all the other board members who are not the CEO and founder have is the pure fiduciary obligation to do the smart thing. All those venture guys did the smart thing, and I'm sure they're glad their LPs are happy, and I'm sure they're glad they're happy. Let's get real here, people.

I think what Jeff is saying is true: it is different for the founder. Even though legally you have the same duties and obligations as a board member as everyone else, I think the interesting question that a lot of this corporate-law stuff raises is—and I'm thinking about this in terms of these 2 deals, in terms of going public, and dealing with post-public stuff—it is their baby.

I always feel in my head, even though you can't articulate it, especially in a public-company boardroom, that the founder has the right to be slightly different and pursue their vision. They have a little more leeway to say, "This is what I want to do." I do believe in both those circumstances that if the founder had said, "No, I believe we should go on here," I think the VCs would've gone on. And if the founders say, "I want to fold," I think the VCs fold. So the practical reality is that it is a founder decision, with everyone acquiescing, right?

But what is this scenario? Is this a fold-or-go-on situation when it's like, "No, no, no, the company should go on, just not with me, because I can go make more money elsewhere"?

No. In both of those cases, can I just be very clear on that? That's a pure pretense to get the government off our ass, right? Everyone knows both of these companies are toast. In the case of Scale AI, in theory, Meta owns 50% of it, and in theory, its business is selling to everyone but Meta. Of course, no one's going to buy its shit anymore.

In the case of Windsurf, the carcass was gone 3 days later. The CEO who's selling out has to pretend, "Oh, this company's going to go on without me." The real truth is that the buyer has cunningly eviscerated the brains and the heart of the company and left the carcass. We're going to pretend it's real, but it's dead as the dodo. Everyone knows it, but no one's going to go on the record saying it.

Except Rory.

Except—I don't have—I don't have... Look, let me give you a clue. If it was my billion-dollar cap gain, I would be quiet and schtum too by not admitting it. I would just be following my NDA.

3. Ramp Versus Brex

Guys, if we cross over to the private markets a little bit, turning tack on this conversation, there were some pretty astonishing announcements this week. First, Ramp hits $1 billion in ARR, and Brex hits $700 million in ARR. Brex is growing 50% after a bit of a rough patch, but seemingly back on now. Is everything just booming? A billion and 700. Is everything just working?

No. I wish they were. I'd love to say all tides are rising. I'm not one of those VCs who goes, "Everything in our portfolio is killing it." No, all tides aren't rising.

I think those 2 businesses are good businesses at scale. They've regrouped, in the case of Brex. The kind of business they are, they're selling money, and they get interchange revenue. It's possible to ramp those businesses very quickly, so I think they're perfectly good businesses in a good place.

I don't think everything's growing at 50%. They're good businesses with interesting dynamics. They're not really selling software. Most of the time, they're selling companies a credit card, which means extending 30-day credit in return for interchange, which they share with the companies. That means your margins are much less than typical software.

But if you're willing to lend money and ramp aggressively—no pun intended—grow aggressively, you can make revenue grow. They're cleaning Amex's clock.

Rory, should they be valued like traditional financial-services businesses, or should they be valued like a new technology-first provider?

I would think that's a bullshit question, because in the end, everything should be valued on the basis of risk-adjusted free cash flows. Start with that. But what you're really saying is, in the absence of free cash flows, what's the best rule of thumb to value those things?

The truth is, they have the margin profile and core dynamics of a financial-services company, but they have the growth rate of a software company. You have to adjust and come somewhere in the middle, with the expectation that—this is the key sentence—once the growth rate slows, they will be valued just like... If they're growing at the same rate as Amex, they will be valued the same as Amex. The growth is what's saving them.

I do think Harry's point is that the AI boom is filtering further and further down the stack and wider. We're seeing Broadcom explode. Cisco—Cisco, that's where our grandpa learned to be an engineer. It is accelerating, right? Twilio's seen some acceleration from AI overall at an Uber level. It's not an AI company. No need to talk about Twilio per se.

But I think we are seeing it, and I do think if you're a B2B company and you're seeing nothing, you're not seeing any boost from AI, you didn't get OpenAI or Anthropic as a customer, and you're not seeing any benefits, you get an F. There's so much money flowing through this system, and OpenAI and Anthropic alone are spending so much of that money. You've got to get some, guys.

It's like fish food at the top. It's floating almost down to where it's dark in the ocean now. It's embarrassing if you can't get any of it.

Dude, sorry, that's the metaphor of the day, right? Top marks for that one.

Yeah. Do you have thoughts on that?

I think we hit a bunch of things here. The first question is whether the growth of, say, Ramp or Brex is indicative of something bigger. I don't know, but I think you're right. If it's deposits—basically, money getting spent—then it's not really about Ramp and Brex. It's about how much venture capital has been deployed in the last 18 or 24 months, and when you look at it, there's a fair amount of it.

It's got to go into some bank. So are they winning some market share? Probably. I know I use Ramp for my most recent ventures, and they've got a great product out there, which is fantastic.

Are they winning because all their customers are crushing it, or is it because there's just money out there? Are they winning market share from legacy companies? That could also be part of it.

And lastly, if their revenue is based on this debt product, then great. Maybe companies having debt on their books is a sign of not-awesome things happening. All that is to say, I think Jason's right. There's clearly a boom that's going on because of venture capital fueling it, which just pushes the question to: great, when will there be the returns that everyone expects, and on what timeframe? That's the big open question now.

From an infrastructure-provider perspective, we certainly saw this at Twilio. We had customers spending a lot of money on Twilio during the mobile boom. A lot of them didn't make it, but that didn't mean they didn't pay us millions of dollars along the way. That's just what it takes to figure out who the winners and losers of the boom are going to be.

We saw a lot of those along the way, and that's one of the benefits of being an infrastructure provider. It's also the risk, because if all those companies didn't make it, that revenue went away for Twilio, and we had to replace that revenue with somebody else.

So it was either going to be more durable revenue or just the next thing that grows really fast and might be the hit thing—and maybe not. We'll see what happens. But in the mobile boom, there were just enough companies coming constantly that even if some of them ended up fizzling out, you had another batch that was the next one that could replace the revenue. That's probably a decent amount of what's happening today with the AI boom.

4. Sierra Ten Billion Bet

When we look at Brex, the last round was $13 billion, and now it's at $700 million, growing 50%. Then you look at Sierra. I love Brett Taylor—a phenomenal operator. I interviewed him before. Not competing with Brett Taylor was the takeaway I had.

It's valued at $10 billion at $100 million of ARR, with Greenoaks leading it. We love Neil Mehta, one of the best. Is this market going AI nuts again, to our last point? Or is this an extremely rational bet, given the operator and the growth trajectory they've been on to $100 million?

To me, when I looked at this, I'm sure there's a spreadsheet that justifies it, right? The 100X. But Brett Taylor—if you buy Sierra, you get everything, right? You get the ex-CTO of Salesforce and Facebook and his team. You get all of it just like you would if you were buying one of these startups, and you get a potential leader.

So my thought is, “Look, worst case, we make $20 billion. All these other deals are happening. Worst case, I make $20 billion on the deal if I'm Greenoaks or whoever,” right? This is a generational guy. This is one of the top 10 guys there is, right? And this seems like a better deal than buying Scale AI. I'll buy him.

If Scale AI was for sale for $28 billion, Brett's got to be worth $56 billion. I think it's part of the math because you might not get it if it was Harry and Jason's company with the same metrics. I think there really may be downside protection here.

What's interesting about the bet is, if you have a mental model of all these bets, this ticks every box but the last box. My mental model is always—we say this internally—“Is there a category that can support a big winner? Are these guys going to be one of the winners in the category, and are you getting paid for the risk?” You could argue that's the sequence of questions you have to ask every time you look at a deal.

Is there a category here? Absolutely. Other than coding, which is the infrastructure play, at the app level, customer support and customer success are the No. 1 use case for AI because it's just so obvious. You have lots of people answering phone calls and answering emails. You can do 70–80% of it with AI. It just saves a ton of money. It's a cost center. This is going to happen. It's a thing.

Are they the winner in the space, or a winner in the space? They're clearly one of a small number of people. They've got a really nice position here. They're dominating the high end, and you've got a person running it who I agree with. I had listened to the Latent Space podcast with Brett, and I remember thinking, “God, that guy's smart.” He was talking tech and business and could move between them.

So if you're an investor, you're like: tick box 1, tick box 2. There are so few deals that tick both those boxes that you're just so tempted. So the only box left is, are you getting paid for the risk? There's only 1 question left, right?

The bigger the market size, the more you can squint and say, “Well, at some point, this company will be worth $20 billion or $30 billion, therefore I can do it. My downside is limited to a low IRR.” In the limit, that actually can be a fatal mistake because you just over-extrapolate too many things.

If you were to say to yourself, “I'm going to commit this, quote, investing sin only 1 time every year,” which, of course, isn't how sin actually happens. Once you do it, you do it all the time. But probably at the app level, this will be one of the ones you think about because you're like, “A great guy in a big market at a terrifying price. Okay, I'll close my eyes.”

So I see how they got there. 100 times ARR is pretty steep at that stage, but I see how they got into it.

The only thing I think through, actually, is just the opportunity cost of the capital for Neil. You're like, “Okay, $350 million there. Yeah, he's probably going to be doing $275 million in a $2.75 billion fund.” That's like 10% of the fund going into that next check, which is the second check into the company.

It's just an interesting one for me, which is: hey, when he looks at the opportunities on his desk and where the upside is, he sees this as one of the top ones. That's interesting, and given the percent of his fund that this will be, that's notable.

That's a very relevant way for a venture capitalist to look at their portfolio allocation, right? And you get dangerous when you get that much concentration. I'm not a venture capitalist. I'm an entrepreneur.

The way I've always looked at it is, when I start a company, all my prior ventures—that was 100% of my capital allocation for myself, for my life, for my time, for my bank account, for everything. And so the whole idea that an investor would have concentrated risk with 20% of their portfolio, that's easy.

This is why it's great having operators on. This is your reminder: don't say you're brave when you put 10% of the fund into 1 deal. On the other side of the table, they're putting 100% of the fund into 1 deal with no way out. Nicely put, Jeff.

5. Late Stage Venture Returns

Speaking of where to put funds, I thought one of the most interesting venture deals of the week was Kleiner Perkins investing $100 million into the $13 billion Anthropic round, priced at a $183 billion valuation. It's their first investment in a model provider. Does every large fund have to have a model investment, No. 1? And then No. 2, is this actually just an indication that the best way to make money and stay in the business is to do late-stage AI when winners are confirmed?

How big's that fund?

$1.5 billion.

So it's probably a logo deal, right? You can't walk into the partners and not have Anthropic or OpenAI on the website. It's not enough. I think it's a logo deal.

I think at $100 million, no one just does a logo deal.

Maybe.

We did the math last week on the fly, and I eyeballed the math. I started off going, “No, of course you wouldn't.” And then you run the math and you go, it's not a crazy bet at all. First of all, in the abstract, as you say, if you had 1—

It's not that it's a bad investment. It's just not venture capital. It's a logo deal because you weren't there at Twilio in the seed round. You weren't buying and dealing in the trenches with Jeff. This is throwing in $100 million at $13 billion. I mean, do you even get a meeting with Dario, or do you just get—

You probably just get to—

You probably don't even get to go in the office. Seriously, you probably don't get to go in the office.

You made a comment here: this isn't venture capital. I think I've quoted Gloria Swanson before: “I didn't leave Hollywood; Hollywood left me.” This is venture capital today. Most of the money in, quote-unquote, venture capital is—venture capital is 20% old-school venture capital and 80%, plus or minus, late-stage, what would have been Fidelity growth public investing. This is where most of the dollars are going today.

So first of all, objectively, it is where the money is going, just because that's a fact. And then secondly, Harry said something insightful. It must have been an accident. He said, you know, “Is this not only the main place it's happening? Is it the shrewd play? Is there a point on the board regardless?”

I mean, the thing about this is you have a chance to matter and be relevant. I can totally—Kleiner doesn't need to matter. I think Mamoon's awesome, and I think they don't need anyone to matter because they have Figma. They're glorious.

But I totally get the idea of sticking some money in some ultra-late-stage, $150 billion pre-round just to feel you're relevant in the space. It's not crazy out of multiple dimensions. It's not the business you probably sold to your LPs 4 or 5 years ago, but it's not necessarily absolutely wrong.

I mean, it would be pushed to the extreme where it would become wrong. When all the other risks evaporate, remember, the first 2 risks I raised are, kind of, “Is it a category at all?” The only risk left is valuation.

In the end, valuation risk expands to fill a vacuum. So in the end, what will happen is people will over-extrapolate and a bunch of these will be overpriced. Then people will go, “Oh yeah, that's why you don't overpay.”

But along the way, there'll be some great companies, and maybe this could well be one of them, where even these rounds math out. You're not in the trenches with Jeff at the seed like Bessemer were, or any of the deals, but maybe you putting $100 million in and getting $300 million—that feels like an easier way to make a buck.

So you're saying it's kind of like going to the mall with your parents' credit card as a teenager? You spend a bunch of someone else's money to feel relevant?

On a bad day, I mean, that's a bit harsh because if your parents were grading you on the quality of your purchases, then yes. So we've chosen not to do that.

Harry is literally hiding behind his microphone.

I just love Mamoon. I don't want Mamoon to hurt me.

Well, bring him on the podcast.

I'm just trying to make some good entertainment here.

He's one of the best of all time. But the $100 million can't 3X the fund on its own, can it? I mean—

Well, Rory, I would argue that you should.

Yeah, I agree. In other words, you're basically on my side, but I'm saying it's not crazy. Jeff is being—I don't say this negatively—a little pejorative about it, and you're saying, “Not only am I right, but I'm actually being a wimp by not doing it myself.”

Yeah.

Which is another way of saying, to put it more directly to Jeff, that you're disagreeing with Jeff. You don't think it's just kids buying with their parents' credit cards. You think it's a rational strategy in 2025 for venture funds to put a big slug of their money into ultra-late-stage investments because, risk-adjusted, the return might be the most attractive.

But you know what? There is another thing, in all seriousness. Mamoon's one of the best there ever was, and Figma's—I mean, that was his first deal at Kleiner. That's multiple billions back.

Figma's a $25 billion company. It feels small, niche in this weird world. It feels small compared to Canva, when we had Cliff on last week, and it feels very small when we're talking about Databricks just crossing 50% growth at $4 billion. It feels small compared to Anthropic and OpenAI.

It feels small, and as great as Figma is, when the 19-year-old founders walk in and all you've got is Figma, it's, “Well, where's your AI one?” Figma's great. My old team used it, but that's just a little niche SaaS application at $20–35 billion.

I know it sounds facetious, but listen to this—the numbers are so big today, and this is the greatest wealth creation, wealth hunt, greed hunt, venture hunt ever. These are orders of magnitude larger. A little $10 billion company isn't enough today. Look at Brett Taylor; he's just getting going with his AI.

6. OpenAI Creates Millionaires

Staying on something very notable there around the space is OpenAI and the secondary that they did, which is $10 billion. It's expanded more and more. How does this change the surrounding areas? It can be anything from San Francisco's real estate market to the retention of those employees to the number of angel investors. The Valley is about to get a lot, a lot of millionaires that it didn't have before. What changes?

I was just going to say, I remember feeling that way before Twitter's IPO back in 2014 or 2015. I was actually looking for our first house around that time, and I remember thinking, “Oh my God, I have to buy a house before the Twitter IPO because everything's going to go nuts.”

The question is: did it? Yes, it did. Was it because of that particular batch of people who finally got some liquidity? I don't know, but I hope that San Francisco—and I know some of the leaders now in the Bay Area are focused on abundance and a growth mindset in terms of housing and building the capability to absorb new wealth without having to displace other folks.

I think that's the mindset of folks in office now, and I think it's a good time for it. On the flip side, it'll create more entrepreneurship, so you'll probably get more founders spinning out of OpenAI once they get liquidity because they're afforded the ability to take that risk. That's another upside.

I agree. You made a comment that's not correct. You said, “It's unprecedented.” A private secondary of $10 billion is unprecedented, I agree. But if this company were public and worth half a trillion dollars—just as a reminder, Apple was only worth $800 billion in 2018 and was the largest market-cap company—20% held by management would be $100 billion.

The headline could be portrayed as, “People who are very wealthy choose to sell 10% of their total holdings to slightly diversify as an entirely rational move.” Much less dramatic. I'm willing to bet that when Apple was worth half a trillion dollars, this kind of money flow was taking place every year, because people would be crazy not to diversify some of their holdings.

It's only anomalous because it's private. A company with the same market cap in the public markets wouldn't make it as obvious what's going on, and we'd digest it just like Jeff said: they digested Twitter, you digested Meta, they digested Google. It's not as anomalous as it seems. It only is weird because it's private and relatively early in its life.

I don't even think it's that early.

True. You're right; they are from 2016. Good point. It must feel early if you only joined 2 years ago and you're getting $10 million. You'd feel pretty good, but, yeah.

Of course, the other fun thing, Jeff, is that if you think about NVIDIA, the guys who peeled off from NVIDIA at half a trillion dollars about 4 or 5 years ago are probably like, “Oh, I took $10 million off the table. It could have been $60 million. Bummer.”

So if you believe in the journey, and Sam Altman clearly is articulating that journey, you might be leaving money on the table. Just saying.

Listen, maybe, Harry, you want to move on. I think the biggest difference from the Twilio time Jeff was talking about is, oh my God, there's so much liquidity, right? Compared with other times, the impact on recruiting is so much bigger in this generation.

If you're running a boring B2B company that's only going triple, triple, double, double, even just 36 months ago you would have been S-tier, right? Today, how do you compete? You're not going to get a lot of people.

When it comes to engineering talent—and we've asked a lot of folks on this show, and we haven't gotten great answers from CEOs on this question—how do you compete? The answer has to be: we don't. We don't compete, we're not trying to hire those folks, or we're not building an LLM. It's tough to get AI talent.

Yes.

It's just tough when everyone's making 8 figures, with that money handed out like candy.

You know what you should do, Jeff? Have Jim Farley from Ford on the show and ask him, “How did you recruit developers during the teens, when they could have gone and worked at Twitter and Facebook?” It's the same problem, right?

Nice.

Yeah, or the NSA, I wonder, too. They do get them.

Who won in that era against Silicon Valley?

Well, Tesla's the only one that can really do what it does.

Amazingly, actually, I point to Domino's Pizza.

Yes. Best stock. A 10-year killer stock.

They built a great tech operation in Ann Arbor, Michigan. So maybe the key is to get out of Silicon Valley.

Interesting you said that, Jeff, because the other thing that turned out—or maybe it's correlated; it probably is correlated—is that they also have been a stunning 10-year stock.

Oh, yeah. The best return. Better return than Google over that time period.

Which I just love, and I did not know that they built a great tech operation. Interesting. You're exactly right. You probably shouldn't be competing for the same people in Mountain View, but there's lots of people who don't want to be in Mountain View. Can't imagine why. As you say, go to Ann Arbor, Michigan.

Well, and I struggle to think of a single legacy company who said, “Hey, we have to get in on the software thing,” opened up its Silicon Valley office, and actually—

Made it work.

Yeah.

Yeah, interesting. Because Walmart did it for a while. I think they closed it. GE did it with that whole weird thing that totally blew up. Yeah, no, you're probably right. Interesting.

7. Anthropic Copyright Bill

Going from employee payouts to one we didn't expect: author payouts. Anthropic paid out $1.5 billion to authors. Is this a one-off prayer for forgiveness? Is this a continuation or a new business model? How do we analyze this?

Easy and super clear. If you read the judgment, it's really interesting. The judge said the following: “If you bought the damn book once and used it to train your model, and provided you paid the $15 per book, that's totally legal. If, however, you downloaded this corpus of books, didn't pay anything, and used it to train a model, I'm going to fine you $3,000 per book.”

That's how the fine was arrived at: 500,000 books at $3,000 a book. So there's actually a fair amount of clarity here. What it says is, if you want to train on 500,000 books to build your LLM, what you actually have to do is buy the book, slice it up, OCR the whole damn book, and you can legally use that.

But if you don't do that and you just don't pay the $15 per book, you get fined $3,000. So I thought it was actually a fairly coherent legal opinion that said this is the cutoff between fair use and non-fair use.

I think Anthropic just made the mistake way back when of not doing that and got caught for it, but it's cheap at the cost. Probably like, “Yeah, we should have done it. It's not a crime. It's like, we shouldn't have done this. We're going to pay our $3,000 per book. Wish we'd paid $15 a book. Life goes on.”

So the future is you're going to go to the bookstore and you're going to buy a book, and it's got a steel bar through the cover, but the version without that costs $3,000.

It's cute. That's funny. I assume, for example, that there will be a much more efficient way than that of doing it. You're exactly right. I'm sure, for example, that there will be a corpus available containing a purchased copy of every book.

I'm sure that one of these AI lab guys will say, “We have bought for you, and just for you, 500,000 books, scanned them just for you, so we have a legally compliant book set that you can use for training.”

Yeah.

But yeah.

Having said all that, this is pretty bad. I think begging for forgiveness—the classic startup thing—is interesting, but they downloaded this from pirate websites.

Yes.

Okay? This wasn’t cutting a little bit of a corner, okay? This wasn’t claiming that something that wasn’t quite open source was open source. This is literally, “Guys, we’ve got to get this rocket ship going. I need a trillion books. I’m going to the 2 places where I can download them.” Pure piracy.

This isn’t even stealing YouTube videos like OpenAI did. This is as bad as it gets. You can’t defend it. You can’t defend Pirate Bay for books.

No, you can’t defend it, but to be fair, they just paid—I mean, you know the concept of triple damages? Triple damages would’ve been $45. They just paid 200 times in damages.

Yeah, and they may end up paying more. It’s not over. I’m not saying it’s bad, but this is as bad as begging for forgiveness. This isn’t just pretending I’m not using someone’s API.

Which is why, again, I admire the coherence of the judicial ruling. And, again, these guys think differently from some of the other branches. They just said, “Look, if you’d done this, this is what it would’ve cost. You didn’t, and we’re going to charge you 200 times as much.”

You’re right, it’s a big fault, and no one’s going to make that mistake again. You could’ve picked a number, or you could’ve said you enjoined them from using it, but that wouldn’t make sense in the context of having a damages claim. They said that because you’re not directly reproducing the book, it’s fair use. So your only damages claim is $15.

Now, the interesting case is where some of these artists are saying, “It’s not a question of just using my art to train a generic model. When I go onto the model, I get effectively my art or my sentences back.” At that point, you go from $15 a book to a much bigger damage.

So I think there’s still litigation to be had and decisions to be made in terms of how fair use manifests itself in the AI age. But I thought this was kind of clear: that’s 1 piece of the puzzle established.

We’ve discussed OpenAI; we’ve discussed Anthropic. Mistral announced in the last few days that ASML has become its largest shareholder at a $14 billion valuation. Everyone is slightly scratching their heads at this, if we’re being honest, going, “Did every other venture investor turn them down? Why is ASML funding this?”

$14 billion is a huge amount of money. How did you guys analyze this? Help me understand what is going on here.

I don’t know if I can, but just so everyone knows, ASML is a semiconductor capital equipment company based in Holland. It’s 1 of the 2 or 3 most important capital equipment companies on the planet. The machines they make and sell to TSMC make pretty much every semiconductor possible.

It’s 1 of the most strategically important companies out there, and I think it’s 1 of the largest market-cap companies in Europe. So it’s far removed from AI software. It’s at the top end of the value chain. If you think NVIDIA is complex, 1 level below NVIDIA is TSMC, but 1 level below TSMC is ASML. So it’s in the AI value chain, to use Jason’s metaphor earlier, but much further upstream.

That’s just the context of what it is. As to why it’s doing this, I have no clue, other than some kind of European—maybe the biggest tech company in Europe should support the biggest tech AI and LLM company in Europe. It is the biggest and most successful tech company in Europe, so maybe it should support the biggest AI LLM company in Europe on some kind of Euro-conglomerate basis. I don’t know.

So they have less of a right to do this than Mamoon, in your opinion?

Almost everyone has less of a right to do things than Mamoon. He’s done so well. But yeah.

One thing, and I don’t know how European GAAP works, Jeff may have some thoughts here from Twilio. When big companies with a lot of cash make corporate investments, it’s weird because if you’re generating massive amounts of cash, it’s orphaned on your balance sheet. You can’t just go hire 1,000 engineers; it hurts your EPS, right?

But if you can swap 1 asset for another, and that asset is not impaired or it’s impaired many years down the road, it can basically be free. There has to be some synergy here, don’t get me wrong, but it doesn’t have to be a VC synergy, right?

If the asset isn’t going to decline, if they’re looking at all the AI revenue that TSMC and others have and they think they’re not going to lose money on this—I remember, a few years ago, someone who used to be high up at Salesforce Ventures said it to me, and it resonated with me. We said, “Mark, our job is to make money at Salesforce Ventures, but it’s more important that we don’t lose money.”

Because if we lose money, we may have to take an EPS hit or an impairment charge. But as long as our investment doesn’t go down, it’s pretty much okay. So the motivations here have to make business sense, but just not losing money might be okay because cash is locked. It’s hard to do anything with it. You can repurchase your shares; that helps. You could invest, and that’s about it.

But it’s kind of like having the entity in China, which is like, okay, well, it may go make a bunch of money, but is it your money? No. All you can do is then reinvest it in the next thing in China and the next, and you’ll basically never have that money back.

That’s kind of what the VC thing is for companies. You’re right, if you’ve got this money burning a hole in your balance sheet, now your investors might say, “Well, give it to us and let us make those investments.” That’s the argument.

But if not that, then you’re right. They can feel free to go make this, and the income they make from that will be discounted, but it won’t be discounted 100%. They’ll get some credit for it. But again, now you just made your problem bigger. You’ve got more cash on the balance sheet. You go, “Hey, do the next investment and the next.” So it’s kind of a wash.

Really, the thing I would say if I’m, say, Salesforce or a corporate investor like this is: is it giving my core business some sort of fundamental advantage? In the Salesforce world, the answer, I would say, definitely is yes. Obviously, it cemented their role in the center of an ecosystem. They ended up making acquisitions. They have more information to make their product decisions on. All sorts of benefits accrue to Salesforce, and I don’t know if you could say the same of ASML.

I think you’re right. I love the comment on cash because, just as a reminder, if you think the software business is hard, the semiconductor business is way more cyclical than the software business, and you have to be tough as nails to run a semiconductor business.

Who thinks the software business is hard?

Hold that thought. Leave it.

Just hard to stay on top.

Okay, we can come back to that. But if you think the semiconductor business is hard, the most cyclical business on the planet, almost, is the semiconductor capital equipment cycle because it’s kind of leverage on the semiconductor cycle.

So to your point, Jeff, I doubt it because they’re so cyclical. But there may come a day when you need that $1.5 billion, and sometimes you just need cash. I’d always be wary of tying up capital, so you do wonder about that. And you’re right, the strategic value isn’t obvious to me.

I don’t know if you need to own the models to sell the capital equipment. I don’t know if anyone’s ever seen a picture of it. This is the most complicated machine on the freaking planet. I kid you not. They are huge. They are enormous. They take months to assemble. They make a Boeing jet seem trivial in terms of their precision.

These guys are not dummies. They perform the single most complex engineering feat on the planet, and they make a lot of money doing it. But I agree. I just go, “Maybe it’ll work. Maybe it’ll make a 3X. I don’t know.”

I think a lot of it could well be just knowing Europe. It has been interesting to see this whole dynamic of non-US, non-Chinese regions now feeling the need for some kind of local champion. The combination of the hubristic talk about AI, coupled with the hubristic nationalism and behavior of the US and China, means that if AI is terrifying and these other countries are very aggressive about enforcing their stuff, maybe you do need a national champion.

Maybe some element of this is behind-the-scenes stuff, just like in the Middle East, where you’re seeing that. I’m not saying I agree with that even slightly, but it’s what’s happening.

Rory, can I ask you, Rory, when has sovereignty ever been the sole driver for a company’s success in the past?

The British East India Company did pretty good. They just went over and took everything. But I agree, I’m not a believer in the tech space.

Look, I said I don’t believe in it, but I’ll give you an example. If there was a free market in trade, then the national champion of any tech makes no sense. You should have a couple of companies competing on a global scale.

But let me give you an industry where there absolutely are national champions that’s high tech: defense. When people are afraid that other people won’t sell them guns or weapons, they make their own weapons. And what’s been interesting is this perception, rightly or wrongly—I think wrongly—that AI is caught up in that.

You start having this perception of a national champion, not because it's the best solution, but because it's a suboptimal solution based on concerns. I think that is true. The Europeans make a whole load of defense equipment that they have no business making from an economies-of-scale perspective. They simply do it because they don't want to rely on the Americans, and this is the AI version of that. End of.

I agree. The ironic thing is, if you go anywhere in London right now, Rory, the only thing you see is Anthropic billboards everywhere.

They've done a great job of seeming European in Europe. I actually thought that was one of the slickest things they've done: establishing the local subsidiaries and talking the talk in a way that some of the other vendors haven't been able to do, where there's been talk about disabling advanced features and a lot of Europeans are holding off buying. I'm not sure if it's the F-16 or the F-35, but yes, Anthropic's done a good job. They've had to do it.

And they made a super-strategic acquisition in Australia, which also made them a lot more Australian to the Australian government and people on the ground. An Australian company incorporated there made a lot of sense. In terms of going back to corporates investing and the benefits that come, you mentioned Salesforce there and how it put them at the center of the ecosystem, Jeff.

8. Atlassian Goes On Offense

Atlassian's M&A team is just popping corks these days. These guys are going on a tear. They acquired one of my companies, Cycle. It was a small acquisition, like $21 million in cash. Great. Thank you, as a seed investor. The Browser Company, $610 million in cash. Josh is amazing, with a fantastic product team. $610 million in cash is a lot of money. How did you guys analyze that, and were you as shocked as I was?

I read the thesis behind it. Mike's always been a real forward thinker, but I would say the thesis didn't really resonate with me in terms of, “We need a different browser for work.” I could imagine some upsides, but are there enough upsides to actually change behaviors? I don't think so. But did that thesis make sense to other folks?

I think we're at a moment in time where everyone feels like they've got to make a play, right? Maybe, Jeff, you've lived it. Maybe you don't really have to make a play when you feel like you have to make a play, but I think everyone's itchy in the seat, right? Whether it's ASML leading a round into Mistral, and Atlassian is one of the greatest of all time, but it hasn't seen the AI. Would this be the play? I don't know.

Sometimes when we're itchy—and it's true for investing too—it's not that you make the wrong investment, but you might not make the ideal investment if you're not itchy. If you've already gotten 3 deals done by September, you might just phone it in for the rest of the year. But if you haven't gotten a deal done by this point, you just might throw in $100 million into the last round because it's the best idea you have.

We can call Mike and ask him.

I'm sure, as a public-company CEO, there's nothing he'd enjoy more than an unscripted conversation with this group of idiots about one of his products.

But when Jeff puts it that way, look, if that's the bet, then listen, Atlassian has a massive footprint among knowledge workers, both enterprise and developers, right? We're going to push this browser and it's going to give us an AI play. There are worse bets.

I watch Jeff doing his M&A.

Yeah.

And I watch Mike. It seems like, with Atlassian, they do what Mike thinks works. So Loom and Trello made a lot of sense, but maybe they weren't impactful to Atlassian at the other end. When I watched Jeff, I'm like, “This guy isn't wasting time.” He's like, “I'm going to buy Segment. I'm going to buy Zipwhip.” This was a man on a mission. He wasn't going to wait for these new things.

This is my view as an outsider. He wasn't going to wait for little things to germinate 8 years later. He was going to put points on. I loved your M&A strategy, even if it had risks, right? Because you weren't waiting, were you?

No. Here's the thing. We were never under the misconception that SMS would be the most dominant way of communicating 25 years from now. We knew that at some point—SMS was already legacy tech when we started the company—but we breathed new life into it. But at some point, that will no longer be the case. So we have to parlay our success in that world and the amazing customer base and amazing revenue base we have—

Yeah.

—and parlay it into the next era. So the question is, we don't know how long that timeframe is, so we better get busy doing it. That was basically our philosophy.

The messaging business is a great business for us, but it was always seen as a bridge to an even bigger play that, at some point in the hopefully distant future, we'll be glad we did. I always liken it to Intel going from memory to CPUs, or one of those. In the fullness of time, people will say, “Oh, remember Twilio started doing SMS messages?” And you'd ask Grandpa, “What's an SMS message?” “Well, let me tell you.” That's how we thought about it, and that kind of urgency.

The thing I would say, though, about any company in SaaS today—and Atlassian is a prime example of this—is they are primed for disruption right now because AI is going to decimate their seat base for their products. It'll decimate the roles people are playing. AI will do the jobs that people are sitting there in Atlassian products doing today.

So the question is, what are they doing? Now I look at The Browser Company, and I'm like, I'm not sure that's the answer to what's potentially going to replace a whole lot of revenue if AI is taking over these jobs that humans are doing in Atlassian products today. I would skate directly there and say, “Great, what is a job that humans are doing in Atlassian products? And here's the AI version of that today.” That's what I would be doing, I think.

It won't be the soundbite of the show, because I know Harry, but that should be. You're exactly right, Jeff. I thought it was a super-insightful set of comments from you two, Jason and Jeff.

To some extent, you were like, “This is a good business, but we have to add on top of it.” It's a lot tougher now when you're like, “My existing business could go away. I better do something.” You have to call those shots.

Maybe this shot didn't resonate, but you're probably sitting there as a SaaS CEO, saying, “You don't have the option of just letting the existing thing compound, because it's not going to add so much to cash flow. It could start declining.” To Jason's point, I love the descriptor: you do get trigger-happy.

What you're really saying is, “This might not be the best deal ever, but it's the best deal of the 3 deals on my plate right now, and I need to do something because I feel the imperative to act.” It's probably a very honest, in-aggregate reflection of the dynamic right now if you're a CEO, and frankly also if you're an investor. If you're not in, you can't win. But, oh my God, it's hard to know.

Well, you can't buy Sierra.

You can't buy the things that are great. It's a weird world. Turns out making money is hard.

I don't know how Jeff thought about Twilio, but when you listen to folks like Benioff and others, they want them all to work, but actually they have a loss-ratio calculation just like VCs, right?

Yeah.

There's a loss ratio. And so, of course, it was $600 million or whatever, but if there's even a 40% or 30% chance it's truly impactful to the customer base, that's probably good enough.

Well, I think the other thing that is conventional wisdom, especially at companies that do a lot of M&A, like Salesforce, is you don't worry about the deals that didn't work out. The thing you regret is the ones you should have done that you didn't.

The whole mantra generally becomes: it's worse to miss a deal you should have done than to do a deal that doesn't end up working out.

Can I ask you, Jeff, what deal did you miss that you feel you should have done?

Probably can't talk about it.

We didn't warn you that he does this, Jeff.

But was there one? Is there one you still think about, or is it behind you?

There is one.

You can see the love in his eyes. You can see the desire. He ain't going to cough it up, guys. He ain't going to cough it up.

Right.

But you can see it's still there.

When you look at Mike on the offensive today, whether we get the thesis or not, and when you look at a couple of the other players in this realm of market cap, do you wish you were a public CEO on the offensive with a big-ass budget to be aggressive and buy some of these assets, or are you happy not being there?

No, absolutely. I thought this was going to be a really exciting time for public companies to navigate this. Like I said, one of the other things is we were in a different boat than almost any SaaS company because, A, we weren't SaaS; we were infrastructure. So we weren't selling seats, and we had no innovator's dilemma as it related to AI. Everyone who's selling seats—

Totally.

—has this massive innovator's dilemma, and we didn't. The way I looked at it is, we were always trying to crack into SaaS, right? We built a contact center product, a marketing automation product, and it was hard to do.

It was hard to crack into the SaaS market because that’s just not how people thought about us. And so that was frustrating. When I saw all the AI coming, I looked at it and I was like, “Holy shit, this is going to replace SaaS.”

All the incumbents here are going to have an innovator’s dilemma. They’re going to add a feature here and there that’s like, “Oh, we’re going to make your human beings doing the job 10% more efficient because of the AI copilot thing,” when in reality they’re going to want a product that’s like, “No, no, no, I don’t need 75% of these people anymore. Give me that product.” Anybody selling you seats is not going to sell you that product.

And so the amazing opportunity is to come in with that next generation, and that’s what you see with all the AI startups. They’re going from $0 to $100 million overnight. That’s exactly what’s happening. As Twilio, I looked at it and I said, “Hey, we finally got our break here. We don’t have to become a SaaS company in order to build more value. We actually have a new way in that we are unconflicted on, and everyone else is. Beautiful.” So that’s how I thought about it. It’s harder if you’re a SaaS company because you’ve got to disrupt yourself right now.

So would you like to be in the CEO seat of a public SaaS company going on the offensive, having the ability to buy companies like The Browser Company?

I think it’ll be a fun job. Do I literally want to do it? No. A, I’ve got a new venture, but B, I’ve never wanted to be a hired CEO. To me, being a founder is the thing I love. So that’s my point of view.

For a whole lot of folks out there, do I think this would be a great time to be at the helm of a company and navigating this transition? Hell yeah. You’ve got a customer base, you’ve got a lot to work with there, but you also have the innovator’s dilemma to work with, which makes it both hard and super interesting.

How much harder do you think it is to make a bold move when you’re public at a time like this than being private?

You have capital to work with.

Yeah.

And you’ve got shareholders who want a great AI story. For us, we got hit with headwinds for growth. And that becomes the thing you’ve got to fix. The question is: are you fixing that, or are you planning for the longer term? It becomes hard to do both at the same time.

If you’ve got the growth rate right now and you’re a SaaS company, absolutely you should be—

Oh, okay.

—swinging for the fences.

That’s helpful.

The hard part is, if you’re lacking growth right now, it’s hard to do both at the same time. That’s the position that sucks to be in.

That actually makes sense to me, and I’m going to put an addendum to it. If you’re doing 30% plus, you can be aggressive, and you should be aggressive and buy shit as a public SaaS company. What you’re saying is, if you’re doing 10%, you can’t be aggressive because you’ve got to fix the growth story.

And then this is the thing I want to add: even though you probably should be aggressive, at least slightly, because just fixing the growth story, getting it from 10% to 14% or 15% on its own, without getting on board the AI train, probably is not enough.

It is an interesting but obvious point: if you’re selling at the infrastructure level, it is easier to get on the bandwagon. You have to have the right product, but you’re not cannibalizing your seats. If you’re selling messaging, if you’re selling email, it’s easier. Then you’re not necessarily disrupting yourself.

Let’s say you’re Brett Taylor at Salesforce selling contact center automation, and they’ve got Service Cloud, which is, from memory, a third of their revenue or whatever, right? You’re going to cannibalize a third of the company’s revenue. That’s hard to do as a public company.

Whereas if you’re a pure play just selling the automation, your job is to go steal a third of Salesforce’s revenue and replace it with a smaller number, but it’s all yours. That’s the whole point of being the disruptor in those markets.

And interestingly, if you eat the labor, it might not even be a smaller number.

Oh, yeah, it probably will be. But you don’t need it to be as big a number as Salesforce has to build a grand slam company as a small—

Yeah.

—well, I’m saying—

No, I’m saying it might even be bigger if you can actually—I mean, you know, some of these contracts you’re seeing, you’re getting more—

Oh, in the end, yeah.

Yeah.

I don’t know. I doubt it. I doubt it, because I think the economic argument will be, like, you save money.

Yes, actually, but the argument people are making is you’re saving not just software money, but labor money as well. Can you command some part of that?

Ah, I see. The software-plus-labor market is bigger than the software-alone market.

Yes.

Yeah, that’s fair. That’s fair, yeah.

If it’s a one-for-one replacement, or even worse, as you suggest, a 0.7-for-one replacement, then a whole bunch of venture money is about to get flushed down the toilet. The only way the math works for Sierra at $10 billion is if you don’t just get India—maybe not up front, but over time. You don’t just get the Service Cloud revenue; you get the Service Cloud revenue plus some slug of the labor.

I’m not sure I agree with that, but I think the upside is getting a slug of the labor, too. But I’m not sure I agree that even if it was just 0.7, or 70% of the revenue, you would still build huge companies that could eat the SaaS companies alive.

Jeff, I’m sure you’ve consistently, over the years, spent time with a generation of public company CEOs and founders, from your Atlassians to your Zooms to your MongoDBs to your Oktas to your Boxes to your Dropboxes, all in the same kind of generation. When you look today at that crop of companies, which founder-CEO do you think will be most aggressive and strategic in the acquisitions they make?

I think Mike will be one of the more aggressive on the acquisitions front, because they’ve always been. I think it’s in the DNA of Atlassian. So I do think we are looking at one of those, even if I am not fully on board with the most recent.

I think Drew has an interesting vision for where Dropbox can go with AI. The question is, will they have a right to play there? Dropbox has struggled to expand out of its core sync-and-sharing market because they’ve tried a lot of things over the years. Will AI provide an opening for them to provide a new market for their customer base? We’ll see. I think he’s got an interesting vision there, but it’s hard to break out of the jail you might find yourself in in those scenarios.

When you’re at the Dropbox stage in terms of growth, it’s just when you need acquisitions the most, but you find it hardest to do as a public company because you’re still in the low-growth penalty box. That must be a frustrating place.

Here’s the thing I would say about both Dropbox and Box. Well, Aaron and Drew—these guys are cockroaches in the public market, right? They’ve been through hell and back in the 10 or so years that they’ve both been public companies, and they’ve managed to survive. I know Drew’s got good protections. I think Aaron does, too. I don’t remember.

But they’ve managed to figure out how to compete brutally and continue the path as public companies, continue to invest a reasonable amount in R&D, and advance the stories of their companies. So I would bet on those two to continue to do that.

I wish they were both doing it faster, and I wish they were both able to do it more at scale. Neither of them has relied on any kind of big M&A, really, and that’s probably a function of their presence as public companies. But I also think that their history of being able to plow forward and make it happen will help them here.

Hopefully, what it takes, though, is some kind of breakthrough. They’ll need some little bit of luck. They’ll push through some opportunity that breaks for them, and then it could be amazing. I think both of them have a shot at it, but it probably won’t be through big M&A. It’ll be through product smarts.

Knowing both Aaron and Drew, they’ve both been looking for that opening, kind of like I was as a CEO.

Sure.

Looking for that opening that’s going to let you break out of your jail and expand your product portfolio in a new direction, and earn the right to play in a new area. AI is certainly one of those opportunities.

In what way did you most want to break out of your jail but weren’t able to?

Well, the thing that frustrated me was that our most successful product was our messaging product. As a messaging API, the crux of that product was an API with 3 primary fields, if you will: from, to, body. That’s a text message. Who’s it coming from? Who’s it going to? What does it say? That’s a text message.

And so we had millions of developers who integrated Twilio into their code and specified in their code a to, a from, and a body. Now, in that world, how do you add more value to the customer over time? They specifically said, “I want you to send a text message from this to this that says this.” What do you do to add value? You’re kind of in a box.

If you look at the last, say, 10 years of Twilio, it was all about how do we create a product that allows us the expressiveness to go add value, because the customer hasn’t explicitly stated exactly what they want us to do. Therefore, any deviation from that exact thing is called failure by the customer.

And so that's a lot of what we were always trying to do: create a surface area that allowed us more expression as a product team and as a company.

Think about if you're a file-storage company. Success is, “I stored your file and I didn't lose it,” and failure is, “Oops, I lost your file. Sorry.” You have to break out of the world of, “No, no, no,” because customers want you to add value beyond just, “No, my file was there, thank you very much.” That's the challenge.

Certain product arenas, and the nature of how customers use the product and the nature of the product's promise, give you more ability to expand. I always admired the product surface area that Cloudflare had because they sit at this super-strategic intersection of the world and your website. Then you can ask the question, and it's a dashboard. Once you're inserted into the DNS and you're proxying all the traffic, without writing another line of code, they could add another feature to that dashboard—

Yeah.

—that says, “Oh, flip a toggle to do this and do that and do that and do that, and all you gotta do is flip a switch,” and it's beautiful. That's a great position to sit in because you're at the point in the product where you can just add that feature and make it a toggle switch. That's beautiful.

Super interesting.

To your point about it, it’s funny. I always thought years ago there were only 2 API companies: you guys, who abstracted the complexity of messaging, and Stripe, which abstracted the complexity of money. Both of you were the interface for developers, with a whole bunch of complex shit behind the curtain.

I think what I hadn't realized, and you made clear to me now, is that you probably had more degrees of freedom at Stripe because there are more things you can do with money than you can do with to/from text. You were trying to find the unlock on top of that. That's my takeaway from that.

Yes and no, right? I don't know Stripe's financials, but given the fact that they've fiercely stayed private all this time, I wouldn't be surprised if they struggle with a similar thing and they've been looking for a better answer. I've heard whispers that a lot of their product portfolio isn't really contributing to the business. It really is the core business, which is pretty common. I'd say probably the same—

Yes.

—mostly the same of Twilio, too, right? It's a main product and there's a bunch of other stuff that you hope will break through, but it's always hard to do that.

9. The Developer API Theory

If we want, we can take a detour and I can tell you my theory of all developer APIs.

Yes.

Or we can talk about the CEO of IRL who was arrested for fraud. Oh, when you put it like that, big guy.

I'm passionate about the fraud topic. What's the billboard on 280? “Ask your developer,” right?

Yeah.

Since inception. But I feel like, Jeff, I'm kind of bummed you're not in the game because I'm on Replit 2 hours a day. I couldn't have been a developer before Replit. There is a renaissance of the developer. Everyone's becoming a developer, right?

I literally integrated the SendGrid API in 60 seconds. I couldn't have done that 6 months ago. I'm not stupid, but I just couldn't have done it. Now Replit and I just did it, right?

I do want to talk about IRL, and I know we're out of time, but I do want to hear your theory of all of it because I feel like it's just become a new world for developers.

Yeah, please.

I'll tell you my theory of developers, let's say pre-AI, and then maybe we can talk about how it evolves. My theory on pre-AI developer stuff—and I tell this to every entrepreneur who would listen and think about it—was kind of one of our guiding philosophies as well.

I thought that in 2017 and 2018, we came to this critical juncture at Twilio where we were saying, “Okay, do we just go more horizontally in terms of services for developers, or do we go more vertically in terms of communication?” We chose to go more vertically in terms of communication.

Part of the rationale was that I analyzed every developer thing that was out there, and I decided there were 3 categories of developer companies that actually got breakaway revenue. There were a lot of folks that were stuck at $10 million, $20 million, whatever, but there were only 3 that could actually break away into hundreds of millions or billions in revenue. Those 3 categories are, number 1, business development as a service. I like that Rory's taking notes.

I've written down “business development as a service,” question mark. Okay.

Business development as a service. If I'm a software developer at some company, I'm not allowed to go open a bank account on behalf of that company. I'm not allowed to strike a business development deal with AT&T on behalf of my company. I'm not allowed to stand up a new data center on behalf of my company.

Mm.

These are things I'm not allowed to do. But with Twilio, with Stripe, with AWS, you can now engage in these business relationships on behalf of the company that you weren't able to do previously.

This empowers you to go build the thing you need to build, and it turns out that when you just backdoor in that way, the developer has a lot of power. The thing everybody wants in the organization is a working product. When the developer says, “Here's our working product,” it just turns out that in order to have it, the leaders of the company have to go pay the bill to Twilio, Stripe, or AWS. It turns out people are willing to do that, right?

So, business development as a service. Second, CapEx. You got your pen, right? CapEx as a service.

Yeah.

It's similar and more relevant to the AWS story, which is—

Yeah.

—a developer's not empowered to go spend $10 million to build a data center, but can they put it on a credit card? Yes. So there you get your CapEx as a service. Developers are spending money of the company that they previously weren't allowed to spend, but this is a CapEx play. That's all of AWS and Google and everything else.

All right, the third—and this is exceedingly rare—is algorithm as a service.

Yes.

This has got to be an algorithm that is so complicated, so obviously beyond the reach of most developers, that you are willing to pay someone else to do it for you as opposed to doing it yourself.

The reason why it has to be so complicated is that a lot of folks think their thing is going to fit this bill. They're like, “Well, we're not business development, and we're not CapEx, but we're a thing that's really cool.” The problem is that developers take your really cool thing that you're trying to charge them for as a challenge.

Yes.

“Can I go make that myself?” It's like a challenge. You're challenging me. You're saying I'm a developer who can't build the thing you built? Screw you, I can do that.

That's especially true when it gets to meaningful revenue, because even if you do get your foot in the door and you get into a product and suddenly that company's paying you $5 million a year, the developers in that company are like, “Hey, I know, I can go save the company $5 million and be a hero. I just need to go recreate this thing.” And that's what happens.

When you're a CapEx as a service or business development as a service, you have backstopped that instinct with, well, you could go build your own Twilio messaging layer, but you still have to go integrate with hundreds of carriers around the world.

Totally.

Why would you? That didn't solve any problems, really. But the algorithm one has to be so hard that the developers say, “You know what? Frankly—

Nothing.

—I'm not smart enough to go figure that one out.”

In that category, I used to only put DynamoDB. The infinite-scaling database is such a hard problem to solve. Even if you have an open-source project, operating it yourself is so hard. You just pay Amazon, they take care of it for you, and you call it a day.

Now I'd put inference in that category, except that pretty quickly, inference became open source and people are running it themselves, too. So it's not even necessarily in that category anymore.

But you would put accessing the core Anthropic and OpenAI—that is their business, the enterprise—

Well, except for the fact that you've got open-source models, right? You've got Llama, so you can run Llama yourself and do inference yourself, right? I'm not saying that it is necessarily the right thing for folks to do, but people can do it.

Right.

And so that argument of, “No, I can do it myself,” actually is valid. They can.

I remember—look, honestly, back in 2016 and 2017, we looked at a bunch of these NLP APIs that were mid-level, trivial, hard. They got some developers, and you're exactly right: they got taken out.

Give Khosla Ventures credit. They did a bunch of those that only did okay, but then they very wisely did the one that did amazingly well—i.e., OpenAI.

And maybe what you're saying is that it may well be that the secret sauce is if you just continue to spend an order of magnitude more money every year making the algorithm better, then no one can catch up to you. And that obviously is the OpenAI/Anthropic play now, because, yes—

Yeah.

—you can get the—

Well, you're not—obviously, you're not selling inference; you're selling the model, right?

You're selling the model itself.

That's what you're doing on a drip basis.

You're selling the model itself. Yeah, that's what I was trying to pick up early on when you were saying inference: you're selling having trained the model and doing the inference. You're selling 2 things together.

Yeah. And if the model—let's say we hit a plateau and all the models are basically the same, including Llama as open source—then you just have a question of, okay, is inference a product that I will pay someone else to execute for me, or is it more cost-effective for me to stand up my own?

And it's really a question more about tunings and things like that than it is about actually running inference yourself. But I believe inference itself is not such a hard algorithmic solve that you need to pay someone else to do it for you. Clearly, training a model is.

Yes. You have this secret recipe that cost you $5 billion, and what you're saying is, if people had the recipe, actually doing the inference—even though it's a lot of the revenue you're generating—isn't that hard; you're just amortizing your model. Interesting. That was super helpful, actually.

We've looked at a lot of these developer businesses over the years. That's actually a very helpful framework.

I was going to share it too, actually, Rory. That was mine—me and Jeff were chatting before the show about it, and I kind of gave it to him, but I wanted to give the founder the chance. I'm a team player.

Well, you've been calling it out at me.

Everything I know about developers I learned from Harry.

Okay, we're at the comedy section of the event.

You have to understand, Jeff, my team is going to clip that, okay? Everything—and everything Harry learned about awesome hair, he learned from me.

Nice. And both those statements are equally true. Okay. Clip it. Please do. Please do.

All right, we're going to do a quick fire. So what price will Figma be at in 365 days? It's at 52 today, which is a $25 billion market cap. Give me some numbers, team.

I'm going to give you a number that's going to say it's about the mid-40s, and I'll tell you why I give it. It'll prove all these silly people wrong. I'm going to give credit to the bankers. They priced it at 35. They get a roughly 10%–15% pop and 1 year's compounding. The price that it should be if the bankers were roughly correct would be around the mid-40s.

So I'm just going to assume that they're more correct than all the idiots who priced it at 110 and moved around and talked about it. I hope it ends up at that price, and it'll allow the bankers to say, “We told you we got it right; it just looked wrong for a while.”

I'll take the interest-rate bet and say it'll be 75.

All right, lower rates. ZIRP. We're zerping again, people. We're zerping. Jason?

I'll bet 60. Will Canva go out in Q4? Yes or no? No, no chance. 0%. Have they even talked about it? I feel like they're pretty happy where they are.

Well, Cliff came on last week and he was pretty open about it. I just don't think—I mean, Jeff's been through it. I don't have the benefit of being on last week's show, so I'm working—it just takes time. And they could have already—I mean, I don't think they can confidentially file, but it gets tough to get it done in Q4. It's already September 9.

I don't think Cliff would have come in last week and talked about not doing a direct listing if they were about to file. I think the lawyers would have shut him down, even though he's a—

Agreed.

I think the first half is the right question, so I'm going no on this one. Hold on. Let me call Cliff and I'll get you an answer. Yeah, call him. Get an answer. What time is it for them right now? I don't even know. Am I waking these people up?

Cliff will pick up.

This should have been the show, Harry. Jeff Calls. That just should have been the show. Jeff just—

We should re—

—billionaire software executive friends and just ask them random questions about business, but sort of uncomfortable questions out of the blue. This will be his next appearance. Jeff Calls. That is what podcasting is, just without planning.

Yes. There's no planning here, I assure you. Okay?

Rory's on a tear against me today. He's just not—

I'm sorry. It really just boils down to the 2 coffees. I've had too much coffee. It's not your fault.

Don't take it personally, trust me. It's okay.

My final one is what Jeff mentioned: IRL CEO arrested for fraud. What happens here? Jason, this is a topic you're passionate about. You can kick it off.

Well, I'm not a criminal litigator or lawyer. Rory's got one in the family. But, yeah, hopefully he goes to jail for stealing millions from the company.

I genuinely believe our ecosystem is so turbocharged right now. In the Bay Area, venture rounds are all getting done on Saturdays. Forget about no diligence being done 2 years ago. Now diligence isn't even being attempted.

I think the best control today would be if more founders who committed fraud went to jail. I just think if every month someone went to jail who completely lied in a round, sent financials where they aggregated all their year's revenue in 1 month, or pretended unpaid pilots were pilots, it would have the proper chilling effect and mitigate the rampant fraud we're seeing today.

I just think it would be helpful for the ecosystem if that were penalized. I know Rory thinks it's a cost of venture, but I think it's reaching an all-time high, and in the end, it's a net negative if trust breaks down in investing.

If trust leaves the system, it's just so much harder. There's so much trust in this system, and there's not enough time to earn it in investing. Sometimes there isn't enough time to earn it. You can't get to know someone for 4 months anymore, even 4 weeks. You might have 4 minutes.

So I wish a few more people went to jail. There are no consequences to standing up at a top accelerator and saying, “We have millions of revenue,” and the next week, the revenue isn't there. There's just no consequence, and maybe that's funny to some people. I don't think it's funny.

I think, like Jeff, the founders used to have—almost every founder used to have—this ethical standard a few years back that I think has dissipated in today's world. It's just rampant greed. I like the greed because it'll make us money, but there's too much of it.

I disagree that it was good and now it's not. I think dishonesty ebbs and flows with greed. You tend to see peak dishonesty at a time of peak greed, so you see more of it now. But I don't think human beings have changed. I don't think we're more moral than people 30 years younger than us.

The truth is, you look at 1929, you look at the boom in the '80s—when there's lots of money at stake, you tend to see more fraud. No surprise what's happening right now.

I do think people who absolutely lie should suffer severe consequences, up to and including prison. That's a general concept. I will say all these cases tend to depend on the facts and circumstances, ranging from—you gave an example—we had contracts and they had opt-outs. Is that fraud? Sure, in 1 way it is; maybe it isn't. All the way to forging documents, which is clearly illegal, right?

I think the truth is, and I'm not going to comment on the specifics because I don't know, it'll range from you absolutely lied to you told things in a way that was very poor, and they should have asked the right questions and you didn't.

You're right, I have a lawyer in the family—a criminal defense lawyer way back in the day, long since retired. But the average federal conviction rate for most crimes is 70%–80% or more. It's slightly lower, I think, for white-collar crime, only because you get lost in the details and the noise of what exactly is intent.

So I don't think it's a layup, but I do believe, as you say, Jason, some actions are just so blatantly out there. “I just flat-out lied. I forged invoices.” If you do that, you should go to prison because there's a lot at stake. If you can't rely on that kind of stuff, there's a lot of diligence you have to do.

I'm not calling for vengeance and death, but I would be careful.

That felt like a bit of a law-and-order lurch at the end there. Come on, say something nice, Jeff.

Well, between Jason, who wants some sort of venture capital ICE regime, I actually think that if VCs are so eager to get a deal done because they're like, “I don't have time to even check any of this stuff,” then it does feel like their greed in that scenario gets rewarded with some amount of fraud. You're like, “Well, that seems about right,” and there's a German word for that, I'm sure.

Here's the pattern that I see every time I read these. I saw this—the CEO of IRL got indicted. I've never heard of IRL, actually. Whenever I read these stories about a CEO or founder who committed fraud or whatever, and they've taken millions in venture capital, the pattern I see is that I've never heard of any of the companies.

Maybe that's just me—I'm an old man and I'm not keeping up with all the cool things—but there's this sort of thing where, well, if I've never heard of any of these companies, maybe there wasn't a lot of real business behind them and they just look good on paper. Because as a real human being operating in the world, if I've never even read a story about these companies, let alone been an active user of them, something seems a bit odd.

It's an interesting comment, Jeff. To your point, the commission of the crime is on the 22-year-old who lies, and they pay the consequence. But, yeah, there was a little bit of me saying the 40-year-old running a lot of money, who's sophisticated, who's running a big firm, kind of owes the system a duty of care to check some of this shit and not be carried away.

SBF went to prison, but he was, relatively speaking, a young man with a lot of hubris, and we've all been there. I know I was when I was that age. It would be better if the people who are paid for their judgment exercised that judgment and, a few times, said, “Slow down here. Maybe we should have an audit,” before we manage $50 billion of other people's money. The consequences may fall on the guilty, but I'm not sure the moral blame should be allocated the same way.

Yeah.

Jeff, I have to say, dude, you have been a fantastic guest.

Yes. Loved it.

I have loved having you on. I know Jason and Rory have as well. Seriously, awesome.

Lovely.

Thank you so much.

Thank you for having me on. It's a pleasure.

I'm actually stealing the little fucking three-way list.