Anthropic Files to Go Public | Cognition Raises $1BN at $26BN Valuation | The 996 Work Ethic

Yeah, I'm not interested if it can't be a billion dollar position anymore.

Losing money is like sex. You can talk about it all you like, but until you feel it, you don't know what it's like.

We are done with the 'ooh, I don't want to do the public markets, staying private is cool.' All these businesses have gone from capex-light cash flow machines to capex-heavy cash consumptive machines.

I would quit as a developer if you told me I could not use the model of my choice. I would quit. I really do think by the end of the year we're going to choose tokens over humans. Ready to go.

Listen, we don't need to talk about how the ARR increased 28% since the last show. I mean, it's pretty good, okay? It's the fastest growing enterprise software startup of all time, of all universe, throughout past Alpha Centauri. But now it's also going to be probably the fastest to IPO to anything near its scale. Right? SpaceX, it's going to IPO in five years, five years to a trillion. Okay. Cursor acquired for $60 billion in four years assuming the deal closes. Okay. Why would you bother with most of the companies in our portfolio? Why would you bother to even meet the founders? Why would you do anything as a VC? No, other than spend the next 24 months hunting these. And as an employee, here's the really tough question: why would you work for any of these companies? Like we have the CEO of Ironclad is now the head of legal at Anthropic, right? Or OpenAI. Sorry, I got it backwards. Jason Boehm, or he leaves, maybe he's still chairman of Ironclad, but why would you do anything when you can build a trillion when it is not impossible to build a trillion dollar startup in five years? Why would you rationally do anything else? Why would you even try to have a $400 million exit, $2 billion exit? Isn't it just a waste of our time? I mean, I know Rory will pick at this, don't get me wrong, but I think it will seep into our society. I think we will all start to feel this way when the bar, not just for valuations, but for time is reset. Why I'm going to quit and spend a year at whatever the... it's always made sense to join the hottest startups, but I think this is going to just make it an order of magnitude at least emotionally feel like they should just quit tomorrow and work for the hottest startups because it's two to three orders of magnitude larger outcomes.

Yeah, I am going to pick on it because in one sense what you're saying is true. Look, we're in the business of investing in the best startups. The best startup is now worth a trillion dollars and you didn't invest in it, right? What do you do with that information? It is the best startup in the last decade. You can fool yourself into thinking there's going to be another one just like it next year. That's one option, right? I think it's stupid because it is by definition a one in ten. Don't make your business plan on finding another trillion dollar, five-year outcomes in the next five years. I think that's just foolish for reasons we can talk about if anyone wants to argue it. Right? The second thing you can do is say, 'I'm psychologically so damaged by missing this that I need to go home and I can't play,' which is credible. There's going to be a lot of people who do that. Or the third is you can grow up and be a [__] adult and say, 'I wish I'd done that deal. I'd give my left arm to have done that deal, but I didn't. Now I got to go on and do perfectly good deals that will have great outcomes.' Because that's what normal balanced people with kind of psychosis that aren't damaged do, right? As I say, in almost every human endeavor like this, there's one person who gets the big prize and you just, and as humans you have to adapt and say even if you didn't get the big prize, it kind of sucks and I mourn it, but you then go on and live your life. Only one person gets to be president, not everyone gets out of politics. Only one person gets to be the richest person in the world, not everyone else can still kind of play in business. So in fact, I would argue one of the reasons business is more psychologically healthy than, for example, politics. I remember a dear friend of mine many years ago explaining he's interested in political careers. Problem is this in politics. The 700 most successful politician in Britain isn't even a backbench MP. The 700 most successful politician in the US isn't even a congressman or woman, right? The 700 most successful business person is probably worth plus or minus a billion dollars. It's an okay consolation prize. So the point is, but a genuine, because I do actually, what you're saying resonates. Look, I'll admit there are nights when I lie awake and say, 'What was I doing in early February, March of 2023 when the Series B went down?' You know, I'd been to some of the early stuff. I'd seen the thing. I can tell you what I'm doing every day on the calendar because sadly I've looked, right? And I'll tell you what I wasn't doing: meeting with Anthropic. So, you know, you can mourn that information, but you can only mourn for so long and then you get on with the rest of your life, right? So I do think I hear you and I also think, say something else: it will be [__] great when it goes public because then we can just move on. It's just, you know, it goes from being the singularity to a magnificent outcome forever. Money flows back to the system. Sucks if you want to buy a house in San Francisco, but it's great. The mystery goes out of it. It's just the 10th or 12th or maybe seventh or eighth depending on how it prices, largest public market cap company and we can all just get on with our lives. So yeah, I'm good with that. I think you're right. You can't necessarily kill yourself for not being in the Series B of Anthropic. I think...

Simplify. See, I should point out the B was the Sam Bankman-Fried round actually. You get to kill, you know, that's a poison chalice on every dimension. The Series C, where Spark brilliantly led the round and Menlo did it. Yeah, that's the round where you kind of go as a VC, that was the round you missed.

My version of it, thinking as a seed investor, is, and I don't mean this facetiously, I'm not exaggerating, is here's my, and this is different than the employee issues and the ecosystem issues. I'm not interested if it can't be a billion dollar position anymore. That's how it's changed my mind. I'm just not...

No position. I literally had this review with my fund management company today talking. They're asking me why I was doing things. I'm like, listen, I will make small investments with friends for sure. I will do things to be parts of journeys, but I'm going to pass on anything where I can't have a... at this point in my career. I'm not saying I would have done this in my first check to PipeDrive, which had a billion dollar exit. I mean, I just don't want, it's not worth the 20 years, hopefully it'll be five years, but I need a billion dollar position to get excited today. I need a billion dollar position. So if it's going to be worth a trillion, I can do a pretty low ownership, right? But you got to be worth north of $10 billion for it to even make sense to me given dilution.

Listen, I think it's a good question and I've made many, many, many, many mistakes. But I do think the inverse is if you see tangible reasons it can't create a billion dollar position. For example, the founders are very, very good but not great, okay. For example, the CTO, he's pretty good but not like jaw-dropping CTO, okay. Not going to launch 17 simultaneous products. If you see that, fine. The TAM is small but there just isn't a sense of how to grow it, okay. There isn't that drive. If there's complaints, I just haven't seen a lot of great outcomes from complainers, from A-minus CTOs, from small... It's okay to start with the smallest TAM, but I want to see that at least you're thinking, even if it's insane, about the large TAM. So it's more just to me, it's more just drawing a black marker through things where I might have taken a little bit of risk before. Just know I'm out. I'm just out because it's not that I can say for sure how big it will be, but these are blockers to a billion dollar position. Not a billion dollar outcome, a billion dollar position. I think you can have a billion dollar outcome if things are lucky and you can have some, you can have the pretty good CTO, the midsize TAM, as long as you get some tailwinds and a few things break your way. I still think you can be lucky enough to have a billion dollar outcome, but not a position.

My fund is actually not, our fund is $900 million, not a billion, just to be precise. And you want, look, obviously you'd love to make a billion dollars, but you also look, I think that...

I think you have to ask yourself how many of those as a base case, how many of those realistically exist? You see, as is often the case with Jason, I actually agree with him on kind of what I'd call the just the kind of things he's talking about in a founder. You know, you do want the drive things, but I will, you know, so yes, I think ambitious driven founders, upside, no complaining. So in practical terms, I agree with him. I don't think if I looked at the same deals he was looking at with excitement, I'd say I think that these are going to be a billion dollar fund individual position, implicitly a $10 billion total fund outcome because I'm just too aware of the base rates. I mean, when we looked at it, you know, five, six years ago, it was like my mental model, which isn't the case anymore in enterprise software, it was roughly, you know, you probably had 10 to 20 billion plus outcomes a year best case. You probably had two to three $10 billion plus outcomes a year best case. And then every decade, you know, you had one to three hundred to now I'll have to say a trillion dollar outcomes. Now you scale that up probably by 10, 20%. But really, there's not going to be more in a normal year. There's not going to be more than four or five $10 billion plus outcomes a year. So I just don't know if that's credible or reasonable. I think in that one to five billion range, if you own 10%, you're very happy you did it. You made $500 million, you're very happy. It's half of a fund and you're very glad, right? Especially if you put 20 or 30 million bucks in and it's a strong capital returner, right? I think as you go later and as we've discussed, as you go later it's much more about concentrated positions, but I don't think you can make 20 to 30 investments in a Series A fund or a seed fund and credibly really believe that each of them will be a billion dollar outcome to you personally or a $10 billion outcome totally. You can think, I mean, so therefore, hang on, let me finish. Therefore I think I tend to mentally have this following model: I want to underwrite to a realistic base case return, but I do agree...

Never do a deal with just capped return. If you can't articulate, the way we say it is you want to have your basics, but you want to articulate a credible upside story that can have that, you know, magic outcome. And that's how, as I say, I end up in the same place with Jason even though we don't agree on the math. You do want to have uncapped upside, but I don't think you go in saying I'll only do it if...

I think the big statement you said there is 'will be' rather than 'can be'. I'm so much more willing to go up the risk on doing things that I would never normally have done because if they do work, they're going to be so mega versus the 'will be' there like as like V1 SaaS companies where it's like I can see it much more realistically, but it's not that needle-moving to have it succeed. Do you know what I mean?

Everyone's always brave at the tail end of a 14-year equity boom, right? Again, my biggest disadvantage as an investor, right, was I was investing in 2001 and I watched the NASDAQ go down by 90% and most of our investments go bankrupt and half, you know, less than 40% of them survive. So yes, everyone's always 'I want more risk because the upside is there' when the risk hasn't come home and the upside is still there. So yes, I probably, I do think you have to at least have to be cognizant of the fact, you know, it's the old cliché. I've said it before in the show, so I'll apologize for repeating myself, but you know, losing money is like sex. You can talk about it all you like, but until you feel it, you don't know what it's like.

It's not that I'm not challenging your math. It's more that I think for founders...

It's for a lot of founders, it may after this IPO, after this Anthropic IPO, it may get even harder to get meetings. That's my point. I won't, there are meetings. Listen, I'm not so great. I'm not no Marc Andreessen with all the E's and S's, but there's meetings I won't take now that I would have taken in 2024 or 2023. I just won't take them. And it's not because they're not great human beings or building real companies. I'm just not seeing that the bar has gone up so much. I just won't take the meeting. And I'm not sure all founders get this. I do think that is real, you know, and I do get this question from founders, you know, and they're implicitly saying, where should I be in the grandiosity versus boring stakes? If I'm too grandiose, I might lose them. But if I'm too boring, I might also lose them because I'm not aspirational enough. And I do agree. What you are saying is correct is the base rate for aspirational has gone up. In other words, there's a level below which the level below which you're perceived as quote-unquote boring probably has increased significantly because people have seen what quantifiably amazing looks like. 10x growth for three years. You're right. There's no doubt it has an anchoring effect and will do for some time. So again, I think you're probably right on the, you know, on the how you think about deals and you just go that...

That's why I'm not sure the Anthropic IPO is all net positive. I think it will make things harder now. It's not just housing that's already happening, right? I think it will make everything harder when there is a general sense of not being good enough, right? That is reinforced across the ecosystem.

I understand what you're saying, but I think we all have plenty of insecurity already based on the private. I actually think to some extent, again, I'm not going to continue the prior analogy though it is tempting, but we are a PG program, not X-rated. So I do think that the mystery when the financials are revealed, are stripped naked, some of the mystery tends to go out of the deal. So I'm actually just looking for, I mean, say it was at SpaceX, it was, we can talk about what's going on in the wider, it was just great to see the numbers, deal with the facts and go, got it, that's what I thought it was. I can differ on how they're valuing it. But that all makes sense now. There wasn't any magic pixie dust. It was, it was an amazing technical launch business, a wildly exciting Starlink business. And oh wow, it'll be the same thing with Anthropic. Oh, those are numbers. Good to see. That makes sense. And you'll just, you know, just get the mystery out.

Well, it's not clear, but I mean going to talk about consuming. I mean, and Anthropic said they filed in, they made an announcement in the last day or two, so June 1st. I think OpenAI had made a statement that they were filing in May, about May 22nd. And I meant to go back and look at the statement. And Jason is better than me at looking things in real time. Did they say they were filing or had filed? Right. I think it might have been filed in which case they're roughly on the same track. Right. And I think I was saying to you, Harry, before the meeting started, the most noticeable thing here is that, and I've quoted the, you know, the famous quote from, I think in the show, in Reminiscences of a Stock Operator, that book from the 20s I often cite, what you're seeing now is everyone gradually jumping forward their cash raise in the public markets instead of 'we are done with the ooh, I don't want to do the public markets, staying private is cool.' We are [__] done with that, right? You know, Elon had 20 years of private, now it's go, go, go, 1.7. Now OpenAI, Anthropic, and if you look at their statements over the last 12 months, it was 'we may go public next year or two, we may go public in 27, oh, we're going public now,' right? Everyone's coming in, right, to grab the capital. And the other thing just to chart, right, today as there was a ton of announcement days, Google announced an $80 billion capital equity raise, right? So the most profitable company on the planet with the exception of Nvidia said, hm, I'm going to need more capital. Better go get it. Right. So I think what you're seeing here is even though Anthropic just pulled off a wildly oversubscribed private raise, I think smart people on those boards are all recognizing the scale of the capital required means we should all jostle to the front of the queue. It's a little like, you know, one of those airline flights in countries where they just don't queue. You know, when they open the door and it's just a mad rush to get on the plane, right? It feels like that here. It's like, you know, Google, SpaceX was going first. Google just got ahead of it. Google just grabbed the first $80 billion. That's a, you know, some of it's done now, all of it, $40 billion if it's over time. SpaceX has just formalized its price at $1.75 trillion for early June. And, you know, Anthropic and OpenAI both said we're probably going to do roughly the same in October. So you're probably looking at across those three names, four names including Google, $300 to $400 billion of equity issuance, all of which is really AI-related given the SpaceX S-1. So grabbing big piles of cash while they can.

Yes, forced to. I mean, you know, in theory, they could have borrowed more. There are lots of things. I think they're smart. I think stock's high. I think equity is cheap. I think it's, there's not a ton of downside to taking a wee bit of dilution at an all-time premium. Getting the world's best investor. Well, you could argue Berkshire without Warren is really, but you're getting a reputable investor on your cap table and you're getting another $80 billion, which maybe you can lever up with debt. I mean, so, you know, they, so I think it's, stepping back, what it's really indicative of is all these businesses have gone from capex-light cash flow machines to capex-heavy cash consumptive machines, right? Generally that's never good for stock prices over the medium term, just across history, things that have high cash flow spinning out are really good investments and things that eat money tend to be bad investments and Google is right at riding a tail... Yeah, on a trailing edge basis it looks amazing. I think it's damn smart to issue money, to issue, to raise equity.

Listen, I don't claim to be, I've only been in the conversations a few times. I don't claim to be an expert. Of course, it's smart. The dilution is, as viewed from like a venture perspective or startup perspective, is unknowable. I mean, it's tiny. It's unseeable. But...

Well, first of all, it may be that they don't want to spend $80 billion. They might want to spend $200 billion and they might lever the same thing. They might say $80 billion of debt, another $80 billion of equity, which will say, and you'll take on $120 billion of debt without engendering a credit rating. So, first of all, they could do both, right? That's one comment, right? Yeah. And then, you know, it's not clear the payback. I mean, again, to give it is not clear. I mean, we can talk at some point about, yeah, is the payback there in the end at all? But even when it is, the payback on these AI data centers is normally two to three years. Now Elon has massively outperformed that with his storage deal with Anthropic where he's getting all his money back if the deal lasts a year and a bit, but normally the payback is not, it's not nothing and three or four years is a fair amount of time. So I hear, look, this is, Google is the second most profitable cap company on the planet. They could borrow all they wanted to within reason. I mean, they have some debt. I think about $70 billion of debt. So they could have borrowed all they want. I just think it's smart to have a strong balance sheet if you find that you might want to spend, you know, $300 billion a year for the next four years before it comes back.

Well, certainly it insulates you from any colds or flu you get in the debt market on any given week or month. At least you don't have to worry about the vagaries of the debt markets which do have micro panics. They do have micro panics.

What's happened here is you've had, it's interesting, it's been a massive round trip, right? I look at just, you know, WCLD, which is the ETF that is the SaaS industry, the cloud industry, versus, you know, the QQQ or SPY. What you basically have is 30% down kind of a month ago, and then we talked about this, and remember we talked about, you were giving me [__], 'we have to buy some stock,' so I bought WCLD, right? I couldn't figure out which ones, I had enough time to think, so I bought WCLD and it's up 25, 30%, right? But we're just back flat on the year. That's the fun fact, right? I mean, the SaaS is, and by the way, I would also caveat literally that was true as of yesterday night. For the last, I think just over a month, SaaS had outperformed semis. Fun fact, but for the year, semis have murdered SaaS to the upside because SaaS has just gotten back to flat. Right? And now, interestingly, even today, you've seen a significant bounce down, right? To me, what, as I say, I'll defer to Jason on the individual ones, but overall, what it says is the narrative got way overdone, and then people looked up and said, 'These things aren't going to zero. If they're not going to zero, they have cash flow value. So they probably are worth more than that.' It doesn't mean that they have kind of long-term momentum. I'm not looking at WCLD and saying, 'I made 30% this year, this month, I'm going to make another 30% next year.' I'm probably, thank you very much, that was great. Right. And then on the individual, I'd kind of pass it over to you. Get your thoughts on some of the individual companies.

Well, look, yeah, for sure. I mean, you know, the media, I mean, if you just read in the last 30 days, it's great, right? Depending on what basket you use. I have a slightly more optimistic basket, but my basket of cloud stocks, software stocks is up 5% this year after today when we record this, another great day, right, in this run, but NASDAQ's up 21% and I think semis are in the triple digits or close. So great that the overcorrection is over, but the fundamental concerns are all there, right? I think our only learning is that there was a, I never understood the total panic of the SaaS apocalypse. It couldn't all be vibe coding, right? Notwithstanding Harry's show where they all want to vibe code their all SaaS. It didn't make sense. But the meta issues of seat contractions, of the fact that AI software spend according to Gartner will be up 60% this year. So that means it's got to be cut somewhere else, right? Those issues haven't changed. And I would say we're no longer, what's great is we're not in freefall because freefall leads to panic everything. Panic, panic buying, panic everything. But even with the reacceleration, you know, Atlassian still at 4x ARR, you know, HubSpot 3.8x, Salesforce 4x. So I know Rory really disagrees with me here, but I just don't think investing really works without 10x or higher outcomes. And I know they're public and I know they're mature, but I think the SaaS, the panic part of the SaaS apocalypse is over. Like we overcorrected as we always do and we may be overindexing on semiconductor stocks today, that will be seasonal as well, but the issues haven't gone away. It's just we overpanicked on the timing of them.

Actually we are in agreement, you know, on the, yeah, because I think we're saying the same thing is that this was a rare occasion where literally an entire sector was discounted to the point where it made no sense. A month ago I was able to identify a bunch of stocks where I could say, hey, they're so cheap that it's silly, right? In SaaS land, now you actually have to say which of these stocks, given that we've repriced, on which of these stocks has a genuine reacceleration or AI catch story. That's a harder message. I mean, I think some of the ones who killed it do, but overall it's still pretty tough. So I think we're in sync, Jason.

Yeah. I think that the interesting thing is there was, look, there was a modest reacceleration of multiples for almost everybody, right? Even the hardest hit Mondays and Atlassians, they all, their multiples all bounced off the hard deck and just went back to crappy from worthless. But the real, the only real learning of this, of the year to date is that it's been long enough now that the public companies, let's just call them mature, the mature public companies that are benefiting from AI are seeing the boost. So yes, semiconductors are up, but you know, when Jeff Lawson was on this show, he said, 'I haven't run Twilio in a while, but I'm pretty sure we're going to benefit from AI because agents and AI just need to use more of our voice and other agents.' And it took a quarter or two, but he's right. Stock's up 57% this year. It's gone from I think 4 or 5% growth to 20. Okta, you know, which was your dad's enterprise auth system, up 57% this year, 56%, right? Datadog, which everyone uses, right, every AI leader uses, up 100% this year. So it's just our mistake and the Captain Obvious learning is, hey look, we gave it six months and every software leader where agentic products and agents need more of it is up, and no matter what they say at X, the ones that only humans use is kind of down even if they bounced off the hard deck, right? There's just not an appetite for more human per-seat licenses. And even Salesforce, which reaccelerated growth, it was all through hard work. It was through Agentforce. It was through organic, inorganic purchases. It was through everything. And Mark and the team came on the call and they said, and they split the business up into two verticals for the first time, I think a gen, for what they're called, Agentforce and the others, and they said the software business will be perpetually in single-digit growth. Like that's as good as it's going to get here, but we're double digits in the rest and that's growing 12 or 13%. So let's not, we bifurcated, we saw what made sense, which is AI-fueled agentic-focused products, but I want, you know, the classic human per-seat software really is dying. It's not dead, but no one wants to buy this crap. In fact, they're cutting it because they got to come up with money for all these goddamn tokens. Everyone wants these damn tokens. Roaring Harry. We just don't need another human seat for folks that don't do any work on our project management software. It just, that has to be the math, right? You can't grow 60% AI software without cutting some material amount of the rest. So we are seeing it, but we should have known this. We should have all, forget about the way I made my bets, just who fell the least. That was at the nadir, right? I bet on who fell the least. We should have all made the bet, the Jeff Lawson bet, which is, okay, who's going to honestly benefit from agents and agentic? And it's not that complicated, right? Of course it's going to be Twilio, Datadog. I never would have thought Okta, but if I'd been a little smarter with Claude, we would have figured it out because it is an obvious one. If everyone doing auth, that's a private company's blowing up, they should blow up too, right?

I think the real challenge is more a question of what does critical mass look like to go public, to Jason's point, what does liquidity look like, all that. I mean, you know, you guys often talk about Replit. There's an example of a company that was pre-AI that brilliantly attached to the trend and just got a ton of lift, right? Is that there's stuff you can do to get lift and if you get the, it's Jason, there's only one test. Are you growing quicker? And if you're growing quicker, you've gotten lift. If you've gotten lift, you're fine. And you can imagine lots of parts of the new LLM-first AI harness software stack. Like isn't it superb is the Postgres database that's just brilliantly co-attached to everything. It's grown quickly too. So there will be instances of that, but it requires deft product management and making sure you're attached to the future. If not, then as if you're not going away quickly, but you just have a question of how do you create value and how do you realize value, which will segue to a few other discussions.

I wish I was more of an expert on Cognition today. When we started this pod, I did say that probably the two absolute highest IQ CTOs in my portfolio were using Devin in the early days when things were still pretty crappy before anything. And the idea I think is still super compelling, super compelling, and now we have some metrics. The idea at least is, hey, it's great that your engineers are 10 or 100 or 1,000x more productive. What's far more interesting is if you can have an autonomous AI engineer. That's much more interesting. And I remember that probably the smartest CTO in my portfolio just in the early days ran Devin in Slack and he would just tell Devin to go do these things and come back and make the commit on their own. And I'm sure it was pretty mediocre in like, you know, a year ago or something like that just because of the nature of the underlying models. But the vision to me is actually more compelling than all this Claude Code crap. It's much more compelling to just tell the agent what to build or not tell the agent, have these autonomous engineers that do it, right? So, on the other hand, you know, there's just maybe it's not that impressive because there's just so much money in the space. Maybe if it works reasonably well, right? And it's just a way to repackage the models for one workflow that's more interesting. Maybe it's not as impressive as it looks. But the vision, I think, is the most interesting vision in the space. Why do we want to empower mediocre sales reps? No, we want to automate them with AI. Same with mediocre engineers. Let's get rid of them, man. Let's have the best ones, but all the rest, let's have Devin. Devin doesn't argue. Devin doesn't only want to work on interesting problems like most of your best engineers. 'It's not interesting. I'm not going to join.' I was talking to someone that turned down an offer at Anthropic. Just said it wasn't interesting enough, but they wanted to work with him. He just didn't want to work on like little edges of basic application. It was boring, right? So, well, let's have Devin do it, man. So I hope it's the biggest winner of all. This is very much a market where leads change hands at a furious pace. So every one of the companies that's worth a trillion bucks, and Google probably has a, every one of the companies worth a trillion bucks is going to want to eat your lunch. So it's a high risk, huge market return and good luck to them. I love it. It's what it should be.

I don't think we're overestimating market size. So, but let's pray. It's funny. We talked about this last week and I was going to tweet this because literally on Tuesday when we record, I was like, it may seem heretical to say that maybe this stuff isn't got an ROI. By the time the thing came out on Thursday, there had been an explosion of these 'oh my god, ROI' articles. So it was right on the cusp of the zeitgeist where people finally woke up. And I was thinking about it last night. It all makes sense. Sometime in November, December, Claude produced the magical version of the model of Claude core that just works. In early 2026, they kind of changed the pricing model so you have to pay as you go. And it's like literally everyone cranked in Q1 and I can almost imagine in every CFO's office in the land someone was doing accrual accounting and like about by midday they suddenly realized, oh my god, you know, we used to estimate our bill based on this and suddenly we're 10x wrong on our accrual and literally the penny dropped simultaneously across the entire corporate US, right? We told these guys to crank in Q1 and [__] me, they cranked, right? And it looks like we spent our entire, our budget, right? And literally it was universal, right? And it's happened. So it all makes sense. This is about the time I've discovered it. Now, to...

your point, I was raising the issue when people weren't. But now I'm going to take the positive side, provided the code that's generated is good, right, and useful, right? I can imagine a scenario where people put a pause on. And I just saw, I think Uber announced today they're just going to give everyone 1,500 bucks a month, which is about the right amount on what we're seeing in terms of averages. I won't say the right amount. That's a normative statement. It's about a little above what we saw the average spend was. So they're basically going to cap everyone, right? And try and get control of the spend that way, right? And the thing you have in your favor over the medium term is the frontier models aren't getting cheaper. I want to be clear on that. They're getting actually slightly more expensive. But the model that is frontier today will in a year be 5 to 10x cheaper because it won't be the frontier model anymore. So if you're getting value from this egregious spend today, you might slow down for the next 12 months, but as long as you stay on that pricing curve, what will then be not a frontier model, but an older model will be available at a cheaper price and you will be able to continue to get value from it. That's a long-winded way of saying it. I don't think you wake up and go, 'Oh my god, we're not going to spend any money on this.' I think the pace of adoption might slow markedly as people realize quite how much they spent.

You know, I had two of my fastest growing portfolio companies say they already blew through their budget this year. So, it's not just the big guys, right? Two of my fastest growing companies. None of my slowest growing companies have said that. Not a single one of my slowest growing portfolio companies have said, 'Guys, we burnt through all the tokens.' May, I don't know if it's causation or correlation, but it was interesting that two of them said that. I think that at a practical level, of course, there has to be cost containment, right? We're not all startups that just raised 50 million with six people where it doesn't matter, right? There has to be cost containment and folks are massively wasting tokens. They're massively wasting them. So it has to come that people have to be more thoughtful with what they build. The vast majority of tokens that are used in coding are in QA anyway. It's not in production. We may have to be more thoughtful about how we do that or not. But I think it's okay if we don't shoot from the hip as often. I think it's okay if we slow down the number of features that we build. I don't think it's the end of the world, right? Having said all that, like more and more folks are using multiple models at the same time, right? For example, not to talk about Replit too much. Replit, it does it automatically now. Replit, if you have a complex feature, Replit builds it in Claude Sonnet, not OpenAI, it builds it in Sonnet to save money and then it has Codex come in and check the work. It has both of them and you don't even know this if you don't even check it because it's mostly for non-technical users. One interesting learning, if you want to learn about cost sensitivity, actually study Replit and Lovable because they're under huge pressures right now from their customers, massive pressures. So if you want to see the future, we can look at what the Uber dude said in one of my portfolios, but it's much more interesting to watch how, to me, how Replit and Lovable are evolving their platforms radically because a couple extra dollars there leads to churn at the bottom of their customer base. So they're radically focused on cost containment and yet they still run, I don't know about Lovable, but still runs two models, not for everything you do, but for anything complicated, the architect agent now is Codex that it brings in to check Sonnet and it works really well. Like it's incredibly powerful to check your work in Sonnet, probably Opus with Codex. It's incredibly useful. It's every single time it finds issues. Every single time it finds issues. It's so powerful. So that's going the other way, right? Is that as we go more and more multi-agent and we ask them to do more, of course we're going to want to. So we have to have budgets, but the tension is only going to grow. And I don't think we've yet to Rory's point, you just asked my opinion, right? I think in the course of this show what we've learned, and it has changed since the beginning of the show, is it really doesn't matter, and this could change in 90 days, but so far it really doesn't matter if older platforms are cheaper and overall models are cheaper because we don't want them. There are use cases where we want them. There are, but overall as an ecosystem, we are all in on the best. We are all in on Opus. That's what we want. So we're not benefiting as much from the deflationary benefits of AI and we're paying into the inflationary side.

But we're pro again, I don't want to sound the negative because I'm actually a net positive. So I want to come back to that. But first of all, I think the zoom out comment is this at some enormously high level point is this is validatingly good news for the model providers. What's happened is somewhere around 3-5% of tech spend, everyone noticed, 'Oh my god, we're spending this,' and no one said we're going to cut back to zero. Right? What this means is you've just established a category that probably has a market size of half a trillion to a trillion dollars and everyone is now going through the corporate process of saying, how do I find that money? Where do I find it elsewhere? How do I manage it? How do I cap it? But what they're not saying is stop it. Right? So you've built a category that's huge. So if I'm Anthropic, if I'm, I mean it's obvious to us because we're in the valley, but you know the doomers who say it's all just going to go away because it's silly. No. Corporate America has said, 'We're spending this kind of money. We don't like it. We're spending too much, but we're going to have to have a plan to spend it.' It's enormously validating, right? I think that, I don't know if it would be interesting to see, Jason, as part of managing that spend, will it still be true that we all want the most expensive model for everything? I know I see some of my app companies to find a way to use multiple models, use open source for the cheaper stuff. I'm not sure if that's for their application, but for their development.

I was going to say, is it a question that there are things where you want one but not two foundational models, but there are other things that you offload, right? Because I don't think, look, again going back to what I said, the cost of off-frontier models keep going down but the cost of frontier models keep going up. And while I just made a positive statement, I don't think corporate America is saying, 'Yay, the cost of the frontier model is going to keep going up and we're good with that.' So I think we've now been found by finance. They're looking to the CEO and saying, 'Dude, on January 1st, he said, "Use all you like. It's going to be amazing." And now it's May 15th and we got a problem. Let's figure it out.' I don't think they're going to say, 'Keep using the frontier model for everything.' You know, where not on a price per token basis, but on a price per pass, on a price per run basis, it's been going up. I think we've now going to have to discover costing and marginal costing.

Maybe. I just think we're confusing these narratives of two things. And I wish I had the exact number, but we're confusing the models used by applications, right, versus models used for software development. And for applications, everyone, I mean this is OpenRouter blowing up. This is everything, everyone's optimizing. Even if you're not optimizing, you're optimizing, right? Because if I had a simple, you know, everything high-end you do in Opus is like a buck, okay? Or could be more. It could be dollars and at the low end it's 50 cents. You can't do 50 cents for a chat, for every single chat, or a dollar. So this could change and there are certainly workflows where you need massive amounts of inference and thought, but I think we're confused into developers are under pressure. These poor guys, as Harry has on the show, they got to work 996. These poor guys, they're not going to use a crappy model on Saturday. Like I quit. I can't. No, literally, I would quit as a developer if you told me I could not use the model of my choice. I would quit. It's not worth my time. Like, you're literally telling me I've got to use... So I just, I'm not saying it won't happen in some use cases, but I would quit. So I think what the more interesting thing to me from all of us, here's the thing that's interesting to me to tie it all together, then maybe we can move on. I think now that we're hitting budget discussions, just discussions, right? And now that CIOs are more involved, and now it's, I think it's not just Uber, even though I think the Uber story was a little blown up and a little apocryphal, but it's going to happen everywhere over the course of this year, right? The budget has to come from somewhere. I really do think by the end of the year, we're going to choose tokens over humans for engineering and product. We're at the margin. You're going to go in and you're going to say, because this is certainly the way it worked for me in the old days when I was at a big tech company, your budget this year is 50 million. Instead of your budget being 400 heads, guys, or 200 heads, your budget this year is 100 million or 200 million or 400 million for EPD, engineering product development. And when I worked at Adobe, that was all humans. Like we ignored all the other costs because they didn't matter. EPD, your budget was just headcount, right? Everyone cost the same $300,000 a year. Didn't matter if they were an office manager or your top, because just to keep it simple. Okay, now we're going to have much more sophisticated budget discussions going into 2027, which is your budget is this much. You decide where you want to spend it, leaders, and I'm going to be sitting around. I'm going to say, do I want to have another 20 mediocre engineers on my team or do I want to give my best guys unlimited tokens? And that may fuel the real, whether they're AI layoffs or just AI backfills. It may fuel another wave of this which is very distinct from the ClickUp whatever excuse for getting fit. Right? This may be a rational choice at the end of the year. There is only so much money and I'll take tokens. I'll take tokens over a B. I would take, we've already made that choice at SaaStr. We got rid of all our Bs. We'd much rather have tokens.

So therefore the next question is you have to come to some kind of opinion what the percentage mix is. People are going to be pushed into some kind of hard choices and then it will force a belief. Do you really believe that these lift, they give you a 20% lift in which case, you know, assuming no net change in, you know, demand, you need, you know, you have 25% lift, you spend $80 on people and 20 bucks on tokens. Do you feel it gives you a 50% lift? I mean, sorry, 100% lift at 50/50. I mean, it will force quantification, right? Once you have a dollar budget and a set of deliverables, which I think is the next shoe to drop, you will.

But even more than that, at more competitive companies, the best people won't tolerate not getting what they want. So you're going to look and so what's going to, I think what will happen going into 2027 is in engineering, QA departments will get destroyed because you'll be like, I got six QA engineers. I got 10 and they're great, but do I really, like I'd rather go to two with tokens and whatever's left in my customer success department that I didn't put into FDS, I'm just going to get rid of them for tokens to manage customer onboard and customer support. I'm going to get rid of all the marginal roles, the ones at the bottom of that list. I'm going to sacrifice for tokens. It's so easy. You don't, it's really simple. The ones on the bubble that weren't already cut in the first wave will be cut for tokens.

It would be interesting to see if you're right, or rather you'd have to have, I mean, you know, you're still, are you going to waive testing? Are you going to waive code with, you know, like you have to think.

Shrink it to whatever the minimum you can do and have the models do the rest. You will just inherently make that choice rather than someone that's just okay, right? You're just going to make the, and I will tell you we had this CCO CS summit at our SaaStr annual event. It was a couple hundred leaders. Almost everyone was talking about how they're getting rid of Gainsight and all the rest and all their teams. That was the topic this year. We're getting rid of all the people and all the software because it's more efficient to just handle this at the agentic level. This was everyone old and new.

Customer success. Yeah. I'm just saying these are all roles on different lines that will just get cut at the end of the year. Whatever is left in these departments, I'm not going to cut my best engineers. I'm not going to cut my actually smaller sales team than it used to be, right? I'm not going to cut my best folks, but just I'd rather have tokens. I'd rather have tokens do the inbound call. I'd rather have tokens handle the 3K deals. I'd rather have tokens do QA or CS. Right.

Jason? With all due respect, right, I think I liked what you said earlier, which is a nice way of saying I disagree with this, right? At the app level, like for customer support, I think the cost of tokens is so low relative to the total cost. It's in the noise. I mean, I don't think token intensity for something like customer support will be a significant factor. You won't look back and say that's a huge one of tokens. I'm just trying to understand the end-to-end product and development life cycle across initial engineering, code review, test, all that, right? What do you think the split will be between dollars paid to engineers in total and dollars paid to tokens just in that area leaving out?

I don't know for sure. I'm just saying listen, let me go back in time. Let's imagine I was back at Adobe. I had 400 people in my little BU. Okay. And at the end of the year, I had it was a fixed budget and we got around the room and we decided we wanted to go into next year with 300 people and the equivalent of 100 humans of tokens. Okay, another hundred of tokens. I wanted it, right? This is very, this is what I would do today if I was fixed. Okay, so now I go, now I'm going to get rid of a hundred people. So I'm immediately, whoever's left in support that isn't great, gone. Okay, whoever's, I would get rid of my entire CS team except my head of CS. I would get rid of this whole, most of the, all of my functional QA team and I would just leave the smartest guys, like I would just get rid of all of those people because I need the tokens, man.

Again, I'm just going to be fact-based, right?

I think it will happen at the end of the year because people will make that choice.

What you're saying, but listen to the math to that 400, what you're basically saying is every, the remaining 300 people, you took 100 heads and replaced them with tokens which implies roughly a 33%, every engineer who's getting 200 grand or 300 grand has roughly 100 grand in tokens. That's what the math would be, right?

Sure. But sales people will have sales applications that aren't cheap. Everyone will have agents.

That's why I don't want, I mean I would, but you keep refusing to, I'd like to just keep it to engineering because I think that the math will be different in sales and customer support because I think it'll be more app with less token intensity but maybe third party apps. I think engineering is where it's more interesting. I can't remember, was it Uber someone announced today they're going to keep it to, as I say, 1,500 per engineer per month which is 18,000 a year, call it roughly 10%, right? And that's probably a first pass swag that says on a $200,000 engineer they're getting 10% token budget, right? You imply on a $200,000 engineer you're getting 33% token budget which would be what's that, 66,000, right? Yeah, and I don't think Uber is like one of the greatest software shops out there. I think products, it might even be Uber said. My point is this. This is the question that's going to get litigated this year. And the amazing thing is you can get to the Anthropic and OpenAI trajectory on even 10%. Right. This is why I think it's.

I think you're having two different discussions. You're absolutely right. I don't think it will necessarily, even just 10% is enough to fuel their growth, right? Yeah. I think it'll be, it may be higher than 10%. It may be higher. And if it's 33% then two things are true. A, buy at any price in the IPO, just any price. And then B, it's going to be pretty tough because you're going to see one in three, one in four engineers, and per your construct, it wasn't just across the entire engineering product development stack get replaced plus or minus any growth that comes from that. So yeah, I mean I'm not convinced it is as high as that just to be clear. I always just for my conversations, but I also admit I'm not an expert and I don't know what absolute state-of-the-art token efficiency in engineering looks like. But I do know one thing. This is the number that I most want to understand over the next, and it's the first question I usually ask all my VPs of engineering. How are you thinking about it? What's working? What's not? Because this is the number that will determine is $1 trillion a fully priced company that could slow down a little bit for a year while digest, or is it oh my god no one's even going to pause for breath. We're just going to keep rolling this out. It's going to eat one-third of engineering salaries and that's going to get you to 4 trillion two years from now.

I just think that this idea of capping tokens like this Uber thing, I just think it's a transitory thing. It's not a utility. It's not just electricity or our density for desks in our office. I think that's a great thing to do now by the end. But for 2027, 2028, you should give department leaders a choice and they're going to choose tokens in good companies. They're going to choose tokens over the people.

Agreed. And I think what's going to happen, you're going to give them a choice. But then practically, I mean, the thing that is so insidiously clever about the AI products that the CFOs are going to be tearing their hair out, right, is it's a product that allows you as an individual worker to be and look way more efficient and have it take away a whole bunch of your grunt work, right? And you know, the cost isn't borne by you, right? Is that which would you prefer? Crank for the next two hours on a PowerPoint or you know to get it just right or type it into you know Claude and say make me a PowerPoint that does this, this and this, right? Yeah, and especially if there's no trade-off cost, no one's going to want the restrictions but every CFO is going to want the restrictions and the dynamic around that is going to be huge. I mean, you know, I think Benedict Evans does the example of, you know, does it end up like cell phone minutes where you give people big buckets? I don't know because the problem here is you can only talk for so long on the cell phone. I think the analogy doesn't work because here as you say as an engineer you can spin up agent after agent, right? But I don't think you can have at the level of the VP of engineering you can have a budget and a tradeoff. At the level of the individual engineer you're going to have to figure out how to empower your best engineers without letting them bankrupt the company and it's going to involve something and it's going to be a moving dialogue, right? If I were building a harness, I mean I don't like to call the cognitions the curses of this, well, the harness company, but having something that was great for the engineers but gave some kind of peace of mind on the budget side would be interesting. I imagine there'll be some movement to that. So that.

All I can say on this, if I go back, if I put myself back in time when I was a VP at Adobe, if you came in to me and I said I could have a choice. I keep 400 people or I go to 300 and my EPD team would commit to tripling our productivity this year. It was a no-brainer.

Not only that, I can instantly think of the people I get rid of. It takes me like, and but pre I would have kept them because I needed someone to pick up the phone, Harry. I needed someone to go to Meta and keep the customer. But right now, if my team made this commit to quadrupling output, I could instantly think of 20 to 40 people just goodbye. Like goodbye. And it wouldn't even take me an hour. It would take me about 10 minutes to get rid of because if I had to make the choice, I know which I'd go.

In reality, I've never let anyone go almost ever because you always needed people. I've always been lean. I was profitable at 6 million in revenue in a B2B company. I ran lean. Right.

But and I enjoy it. But at the Adobe scale, I would know how to get rid of like 50 or 60 of them. It would take five minutes, right? And a lot of them I inherited. But.

And let me ask a question then. But you're the VP of engineering. Let's just play that out, right? Let's make it real. You're the VP of engineering. You had 400 people in your engineering department, and you said, I tell you what, guys. I'm going to drop down to 300. I'm going to take 100 salaries, turn them into tokens, and I promise to deliver not even 3x, but one and a half times what I've delivered before. Right? Do you know VPs of engineering who will say hand on heart today they know they can do that end to end, not just lines of code, not just pull requests, but shipped product with the features the rest want with that level of cut? Just curious.

Yes. Yeah.

Okay.

The faster the startup is growing, the more it's true. The slower it's growing, the more they tell you it can't be done.

Great. I think you're totally, it's totally true for startups. I mean I have them come and say oh my god I can't believe we can do this with five people. It will be interesting to see at fill-in-the-blank large company can they do that because remember from a budget perspective that's where the money is. I agree our smart startups with I mean 10, 20, you know, 25 year olds are cranking and doing more than you could do with 40 people, right? The interesting thing is can Uber take one-third, can Microsoft take one-third of their engineering and do that? This is, you're right, we'll see, it will be fun. There will be VPs of engineering on both sides of that trade who lose their job. It's just the emails are going to go out, the 'I'm sorry, you've been laid off for tokens.'

This is the next, you think of this as the next, all the stuff we've talked about the last x months. It's not even very, it's not even interesting. It's theater. It's theater to get efficient. It's theater to free up racks that they December 31st. People are going to get these cruel emails. It's not you, but we needed the tokens. It's not you, but we needed.

Agree with that. And the question will be for those VPs of engineering, are you getting the value from those tokens?

We all know in our portfolio, there are plenty of old school folks who still don't think this stuff works. They don't think it's worth it. There's plenty of folks. It's not all age because there's young curmudgeons and there's old curmudgeons and some of the earliest adopters are the most experienced engineers because they're kids in a candy store. They love it the most, right? But there are resistors to this day. And these are the products like Marketo that haven't added a feature in 11 years. Like good luck to them. I just released a show with Brandon from Miro and he said they now spend more on tokens than they do engineering salaries.

That's a very good question. I don't know. 40.

I think more like 80. But yeah.

But that's the conceit in those stories, right? Hooray. How many engineers do you have? 1,200? No, 80. Okay. Well, then you know what?