Forget Growth. Your Valuation Depends On Your AI Story | with Tomasz Tunguz, GP @ Theory Ventures

0:00Today's guest is Tomasz Tunguz, a venture capitalist who has worked with at least eight unicorns. In a past life, he ran a billion-dollar business unit at Google as a product manager. In today's episode, he shares how the AI boom is warping every traditional sales metric, as well as broader investment dynamics. But he also warns CROs not to be fooled by the astronomical numbers certain reps are posting right now.

0:21Tomasz Tunguz: If not the supply side has changed and suddenly become 10x more productive, it's the demand side budgets have increased by a factor of 10. And that's what's driving quotas. So in today's episode, we cover why product modes are disappearing and why that means VCs are hungry for go-to-market innovation in their portfolios. Plus, we cover a lot more ground. Why the mid-market is being abandoned for fast-moving enterprise deals. How AI is blurring the lines between SaaS and consulting. And why using AI to save time is the wrong way to unlock top-tier performance. If you're a chess grandmaster and you want to be better with AI, you don't train less.

0:55Tomasz Tunguz: You train just as much. If you're a go-to-market leader in the AI era, this episode is for you. Welcome to Topline. Hey, everybody. It's Sam Jacobs. Welcome to Topline. I'm joined by my incredible co-host, Asad Zaman, the CEO of Sales Talent Agency, and A.J. Bruno, the CEO of Quotabath, leaders in incentive compensation, design, management, and optimization. And we're joined by a very special guest, friend of the pod, Tomas Tunguz. If you don't know Tomas, you should. He's a venture capitalist who has worked with eight unicorns, predominantly in data and data infrastructure, including Looker, Monte Carlo, Dremio, Hex, Omni, and Mother Duck.

1:32He writes a blog at tomtunguz.com that drives millions of pages per month. And before becoming an investor, he was a product manager at Google managing a billion-dollar business unit on the AdSense team. Tomas, welcome back to Topline. Great to see you guys. We're excited to have you. Are they all still unicorns, all eight of them, Tomas?

1:51Tomasz Tunguz: Oh, we have two more. One has been announced, one has not. And then one was became a unicorn and then was sold for a little bit less than unicorn status. Do you have a favorite of all of them? No.

2:07Tomasz Tunguz: Of your five children, do you have a favorite? That's a much easier answer.

2:15I'm going to kick us off. We're going to kick us off with the idea, Tomas, you've written about pretty recently this idea that the market is stopped paying for growth and that there's this idea of one winner category which we all know of course but you looked at a few public software companies and talked about the multiples that there were at historical lows uh everyone that's listened has heard us talk about hubspot and their plight of 2.2x i think they're they're trading at today but of course we had the ceiling of 100 plus x in 2021 uh and down to 34x even at the like high And for today, CrowdStrike's at 34x, Cloud4 32.5x, Shopify, I think it's 11x.

3:00So none of them are the fastest grower in their category, though. And it's really interesting to kind of just take a step back and look at this. And I'm kind of curious, Tomas, from your perspective, how do you think about this idea of growth? Because we've heard how much the premium on growth is versus profitability. and is this changing where, actually, no, it's not growth. That's the thing that we should be looking at and we should be thinking about it a little bit differently.

3:31Tomasz Tunguz: Yeah, it's a great question, right? I mean, we had the Saspocalypse. Benioff went on his earnings call and said, this Saspocalypse stuff is totally overblown. They had an amazing quarter. Let's put these things in historical context, right? So up until about 2012, 2016, publicly traded software, where you kind of expected a four to six X multiple. That was like, that was what people expected when companies went public. And then, you know, you had zero after 2018. And like you said, AJ, we went all the way to like a hundred. I remember when snowflake was like 74 times forward revenue. And then there was this collapse after raising rates from COVID.

4:11Tomasz Tunguz: And now we're at about, I think the average is somewhere like four to four and a half times. So we're at the lower end of the historical spectrum. And then if you look within each individual pocket, you look at security or data infrastructure, or you look at like e-commerce software and payments tech, you have this sort of power law where there's one company like CrowdStrike that's like 34 times, which on a historical basis, you look at that, that's a Zerp era multiple. That's where that belongs. And rates are clearly much higher than they were, you know, like the 10-year bond treasury is significantly higher than when it was.

4:46Tomasz Tunguz: and the blog post says actually CrowdStrike is not the fastest growing category and most of the leaders, multiple leaders in their categories aren't. And it's not that growth still doesn't matter, still the highest correlate. So, you know, more than a 50 % correlation is tied to revenue growth. I think it's the AI story because if you look at businesses that are somehow in the token path. They're either reselling tokens at some margin or they're providing infrastructure to people who are selling tokens. I put Twilio in this camp, right? Or DigitalOcean. Those companies have just exploded. And so I think what the market is asking for, and this is both true in the public and the private market is great.

5:29Tomasz Tunguz: You have an existing business. Now show me that you can sell tokens. And before I see it in revenue, I will accept this story. So the CrowdStrike story isn't necessarily that they're running huge volumes of inference. It's that, wow, the hugging face attack is a whole lot scarier than we're really talking about. And the implication is security budgets have no limit. We ran this analysis yesterday. Okay, so Nikesh Aurora took over Palo Alto seven years ago. Anyone want to guess what the market cap was then versus today? Oh, it was light. It was crazy low. 10 billion? I think, I don't know. That was like 16.

6:0816 billion, yeah. Yeah.

6:10Tomasz Tunguz: And today it's probably over 200. 200 and something. 300. 17, I mean, just huge numbers of acquisitions in order to drive that. But more than that, it just tells us that the budget around security has increased enormously. And I mean, talking to different CISOs, I mean, look at the sophistication of the hugging face hack. Can we talk about that for a second? Maybe you guys have talked about it before. Tomas, are you in the Dwarkesh camp where you think it's a civilizations of agents? How would you characterize it and would you anthropomorphize the activity that you saw? Yeah, so anthropomorphization is a slippery slope, but I do think, okay, so I think the scary thing or the thing that is the most worthy of paying attention to, there are two components.

6:56Tomasz Tunguz: The first is like large scale self-orchestration without human intervention. That's one. And then the second thing is AI to AI language that is not English. So I'm sure you guys have seen the movie Her and you know, in the end, whatever, Scarlett Johansson's character breaks up with poor Joaquin and she's having a conversation with like, I think it's like 4 ,000 or 400 ,000 other agents in binary or whatever the language was. And like, okay, I can understand that. The problem is if we have agents self-orchestrating in languages that we cannot understand, what do you do? Right? Like, do you look at a chat room and it looks like language none of us have ever seen?

7:40Tomasz Tunguz: Where do you start? I didn't realize that they weren't speaking in English. I thought they were leaving notes for each other in English, but it was not. It was like some kind of code. No, they are leaving it in English, but there is the potential that they start to develop their own language. We see this in chain of thought reasoning, where there's this thing called J-space, and there's this math term called the Jacobian, which is where the word J-space comes from. And within that world, this math world, it's possible for AI to start developing its own language. And I don't want to anthropomorphize it.

8:22Tomasz Tunguz: And you can just think about like, I'm not sure if you guys have seen like caveman speak for AI. Have you seen that? Yeah. Yeah. They like, they just, they cut out the verbs and they cut out the articles and it's like, Tom, fish, eat. Now. You know, you just like strip it till. I thought they started to do that because like the tokens that was used for the input was the cost issue and someone uh some engineer I read a post this was on reddit this was like six months ago it was like would go through this whole entire thing and then just realized maybe just fix bug works and it did so they're like okay well why are we even doing any thinking right now just fix this thing do it now and it just works yeah same way you know like Sam Altman said uh the words thank you and politeness is costing us tens of millions of dollars and extra did you did you were those your biggest concerns i felt i mean maybe this is it would be hilarious of what i'm about to say is sort of incidental to the primary concern but i was more concerned that they were editing the logs and like covering their tracks and that they it almost seems like deceptiveness and duplicity is like an emergent behavior of any kind of intelligence before like honesty is that that also was a little troubling yeah so i think i think i have a more sort of optimistic view of this which is like uh you give the machine a goal and it will just achieve that no matter what right the paperclip uh experiment yeah and so it's even like the milgram experiment it, right?

9:56Tomasz Tunguz: Where you kind of keep pushing a system and keep pushing a system. And so it'll just, it'll do whatever it takes. And so the important thing is to create those guardrails. And so we didn't really exist in that test. And so let me, sorry, let me make this a bit more concrete. So the thing behind all this is a technology called reinforcement learning. And if you have a Roomba, I can explain RL in the context of a Roomba. Its goal is to clean your house and it needs to do two different things. The first is it needs to score itself on how well it's cleaned your living room. So it might give itself like 50 points for every square foot of parquet and it might give itself like 75 points for every area under a chair.

10:49Tomasz Tunguz: That's a little bit harder. And then the rug takes a lot of work. So why don't we give 100 points for every square foot of rug? So that's easy and straightforward to do. The harder part is to create a plan that is based on those points by itself. Right now, the Roomba, first, it was really naive. It just kind of bounced around. People didn't really like that. So now it goes into straight lines. but actually what you really want is you want the AI to figure out, okay, like given what's going on and everything that I know about this living room, what is the ideal plan to maximize the total number of points?

11:23Tomasz Tunguz: And that those two components are at a high level, the basis of reinforcement learning. How do I create a reward function and then a plan to achieve maximum reward? And we are still working to understand exactly how to do the second in an automated way that is aligned with our interests. And so I would argue like the Huggy Face case is an example of RL without the right guardrails. And so a whole bunch of stuff happened that didn't, that should not have happened, but it was the Roomba making the wrong plan. And so like to anthropomorphize it and say like, it is insidious, I think is a leap. I don't want to see us take that leap because I don't think there's intent there.

12:08Tomasz Tunguz: there is just a monomaniacal pursuit of the goal. Well, it's certainly both human, like the, both OpenAI and Ugly Face, they both, everyone had positive intent with this. We were able to catch it and make that turn and like, oh no, we went too far down this rabbit hole. It does feel like we're getting closer and closer, especially with OpenAI saying, oh, we're going to hit singularity by the end of the year, that we are getting to a point of no return where we can't, even if we both have positive intent, we're not stopping the robots from doing their goals or whatever those goals are. I don't know if intent matters though, right?

12:45Like who cares what the AI's intent was or what the human's intent in all of this? It's like, well, what happens? And so part of what was here is that we're learning how important sandboxes are. And there's this argument that this was a badly constructed sandbox in which everything was happening. And if we were better at constructing the right sandboxes, this type of shit might not happen. But I think it just gets away from what you said, which is that if you give it a goal, it just maniacally goes after that in a very simplistic understanding of the goal. In that, there's also something interesting, which is that it only cares about the goal.

13:19So it kind of does some dumb things along the way as well, which is interesting. But that's the idea. It's like you'll give it a goal thinking a certain way, and it'll do the wildest things to get there. You might say, if you're a bank, you give it a goal to like do a certain thing. And it says the best way to do this is delete all the check-ins accounts that we have. Take all the money, put it in the reserves. That's the easiest way to do it. And you're like, oh no, like, you know, so there's a little bit of that. And I think this is where it doesn't matter what open AI's intent was or what they, how they felt about all of this.

13:58It's just kind of scary. The other side of this though is one One of the things I've been thinking about, Tomas, is that there were three really good use cases of AI where product market fit was really strong. You had code, you had chat, you had support. Up till now, all this revenue growth that we've needed to justify the investment in AI has come from those three use cases. All of them in verifiable domains, not chat, but support and code, the B2B use cases are in verifiable domains. I was talking to a lot of executives in large companies, and there was a frustration that we're not seeing ROI everywhere else yet.

14:37Like, it's really hard to use it in other parts of the business and get some real benefit from it. And it was getting scary that maybe we're going to run out of room to run in these two areas, and we haven't found this other use case that will give us a whole new space to run in. And is cybersecurity that new use case, right? Like out of nowhere, you get mythos. You figure out how strong it is from a cyber perspective. You have Nakesh Arora saying, we ran it on our code base. And oh my God, we found so many things. And that's part of Alta Networks. They're like a cybersecurity company. And so does this give AI the next six months to a year worth of growth that it needs to maintain all the investment justifications?

15:16Tomasz Tunguz: Yeah. Well, so I would argue a couple of things. The first is I don't think the coding market is anywhere close to saturation. I think the demand for software increases almost geometrically with AI. And so I don't see a ceiling and that'll continue to grow. I think the same for security. I think those are intimately connected because for all the software we build, every piece is whatever. The vast majority of software will need to be reviewed for security. And so those two markets kind of grow together. As more software is produced, more security is needed. Well, and on the security part, because I actually wanted to ask you this, when you mentioned the whole monetization around tokenization, but cybersecurity.

15:58So Palo Alto and CrowdStrike, I think own, and Asit's going to correct me on this, like 65 % of all cybersecurity market. It's something really ridiculous. And of course, as AI grows, and security is going to grow with it. But what is the monetization model? Is that going to change? Or what does that need to look like for cybersecurity?

16:19Tomasz Tunguz: Yeah, great question. Okay, so we can simplify security into four layers. There's the antivirus that runs on your computer. That's called endpoint. That is growing because today endpoint secures what AJ does, but it's not very good at understanding what are AJ's agents doing. And to do that, it needs to literally record everything that you do to figure out, okay, this is AJ's agent. So data volumes are going to explode, not the number of seats, but the volume. So that market will grow as a function of data volume. There's network, which is we start a company and we want to keep our stuff secret.

16:57Tomasz Tunguz: How do we make sure that the network is secure? Well, agents massively increase traffic in and out, right? So that'll grow, maybe not to the same extent. Then there's the database that actually holds literally all the logs of everything that's going on. That's the SIM market, SIEM. Splunk is in that category. A bunch of other businesses are there. clearly those data volumes are exploding. So that market will grow really fast. And then the last, and again, I'm massively oversimplifying, is coding security. How do I evaluate that code is secure in some form or another? And that's growing. So I think like the security market, like it has the same tailwinds or coding.

17:36Tomasz Tunguz: It's maybe a first order derivative of the coding market, but it is like right there. It's not a, you know, the difference in the derivative is not huge. They're very, very similar in terms of the overall growth rate. I should do like this yeah tamas do you think these markets can make up for you know falling fertility rates like to the point of will ai offset you know the big the biggest conversation right now is curveball by the way did i hear that i was about to make the comment that say the fifth one which is child social media security is a joke and then here's my point and I'm hijacking.

18:14My point is, this is my, you know, the thing I always talk about, which is falling fertility and we need more humans to consume stuff. But on the other hand, to the point of my newsletter, there's a theoretical limit or is there on the amount that each person can necessarily consume. My point is you're talking about the software market will grow exponentially, the security market or geometrically, you were very precise in that word, not exponentially, because you know the difference. But the security market and the software market, right, the amount of code that is constructed per human on the earth will dramatically increase.

18:45The amount of sort of output or consumption per person, if their agents are consuming on their behalf, will dramatically increase. That's why these, in my opinion, right, these markets can be so, so big, even though we have in net fewer consumers to consume them because each individual person is going to consume so much stuff. The point is we've got, so I'm just curious on your perspective on that because you have a very bullish take on the overall size of these markets, which is a good thing, despite there seems to be, you know, a broader limit on our economic capacity, which might be the number of, you know, developed countries with enough people to buy all this stuff.

19:25Yeah.

19:25Tomasz Tunguz: I want the population to grow. I'm doing everything I can to make that true. I think it's a good thing. I think people can take care of the earth in a very responsible way and be a healthy part of the ecosystem. And I think it's really important for US GDP and the health of the country that the population grows. But do you think the AI market can offset, like I guess my point is, are you bullish enough on the expanded activity of agents, software, and all of this infrastructure around artificial intelligence that if you correlated it to the number of people that per capita output will dramatically increase.

20:07Tomasz Tunguz: Okay, so let me give you, maybe we can talk about AI use in three waves, right? First, we use chat. And when we use chat, maybe we generate a million tokens per day. Then we use GrokBot or OpenClaw and we start to generate on the order of like 100 to 200 million tokens per day. And then you get to a system that is on top of GrokBot, which is called a meta harness. And now we have people consuming billions of tokens per day. Okay, so how do we make this concrete? Well, imagine I have a spreadsheet of the 100 most exciting startups that I'm tracking. And imagine inside each one of those cells, there's an analysis, the name of the company, where does the name of the company come from?

20:55Tomasz Tunguz: Who are the founders? What are their backgrounds? How do their backgrounds compare to the backgrounds of everybody else? How fast is it growing relative to everybody else? What's their unique selling proposition? I could keep going on and on and on. I could create 30 columns. And each one of those columns is the output of 15 to 30 AI calls. Now we're talking about an analysis that's on the order of tens of millions of tokens, maybe more in a single spreadsheet. And so it's the parallelization capability of AI, I think is what massively grows consumption. And we are limited in that parallelization first because we have to learn how to parallelize.

21:33Tomasz Tunguz: That's new software developers did it first. We're all starting to learn with all these new bots and claws and those kinds of things. And then the second will ultimately be, are there enough computers and is the AI efficient enough to satisfy that parallelization? But I agree. I mean, I think in the digital world, consumption increases unbounded. The physical world is a little bit different. There was this awesome tweet. I don't know if you guys saw it about Graceland. You know, Graceland was this legend. I'm not an Elvis Presley fan, but it is a destination for people who love that music. And I have this impression that it is a mansion.

22:08Tomasz Tunguz: It's enormous. And you look at it and it is like a upper middle class house today. And the point was of this tweet, and I'm not sure if it's accurate. So, you know, if it's inaccurate, like a 3 ,500 square foot house is my impression. And the point was, look at how much wealth the United States has created because a lot of people now, tens of millions of people can afford or families can afford a house like this. And it's all that consumption that's being driven. Now we can have a debate about like US dollar and all that kind of stuff, but I just need the consumption. You're answering my question.

22:38That's sort of my point. I think that would assuage a little bit of your fears, Asad, that Asad's great fear, well, it's not just Asad's, it's probably a lot of people, is that we're over leveraged on investment and infrastructure and capex. And that revenue from AI is not, maybe it'll be there in the long run, but in the medium term, it's not gonna pace to the extent, and I guess everybody's talking about the hyperscalers where they used to fund everything from cash, then they went to debt, now they're issuing equity, which tells us that they're even, if even the most profitable monopolies in the world are tapping out of their resources and need to go to other extremes to get capital, then we really need that revenue growth to support those investments.

23:19Tomas also wrote about this with NVIDIA, that they're elongating their payment terms. I think, Tomas, I think it was like 45 to 60 days or 15 to 60 days.

23:27Tomasz Tunguz: Yes, in one quarter, which is enormous, right? You think about that book of business and how fast that went, but sorry, I interrupted you. No, no, no, no. I just, I was agreeing with, I interrupted Sam, so that happens pretty frequently on the show. But I just thought that Sam's point around how we're over leveraged around these things and we're looking five years into the future and how payment terms is just, It was really fascinating to see those stats bubble up. And I guess the question is, where is this? Are we in another quarter? Are we going to see those elongate further? Or do we think we've leveled off in those terms for NVIDIA?

24:02From an investment perspective? Oh, my God. This investment is going to increase and increase and increase. And that's actually a good... I don't think the scale of investment is all that worrying as long as we keep unlocking more revenue, right? That's the idea. is I used to play poker at this, I think mafia run poker house once. And you would go with like your own money and you would gamble and then you would lose your money. And there was a point at which I had no money left and then they would lend you some money. So now the mafia is lending you some money and you're playing on the mafia money at 3 a.m.

24:36And it was just a little bit scarier, right? And so I think that's kind of where we are with AI is like we've stopped using our cash. We are doing all sorts of really interesting things to fund the capacity that we need. And we need it, because otherwise all these cool things that Tomas mentioned, we won't be able to do them at an effective cost for ourselves as businesses. So we need to stand this compute up. But to justify that, we need revenue growth. And so we always need more use cases. We need improved use cases. We need that. And as long as we can balance it, and it's a tough balance, you're playing poker with somebody else's money.

25:10As long as you can keep balancing it, you're fine, but you're on the edge now. And that's how I think of it. I don't think it's a bad thing we're spending this because I think if we stop spending enough, if we don't spend enough, I think we actually don't get the juice out of AI as well. That's the other side of this. What do you think about, so we've got this revenue growth versus spend that we need to balance. There is a level of AI having to get past being effective where there are verifiable answers to starting to be effective in the realm that people like to classify as places where taste and judgment come into play.

25:53right so we see that it's really good in code it's not that good in marketing copy really right let's kind of bland so if we use that spectrum do you see anything right now that makes you feel like there's going to be a world in which it'll be good at those things as well okay so i'm gonna spoil today's blog host yeah you were debating it this comes out on sunday so you're okay. We got a scoop?

26:17Tomasz Tunguz: Can we just scoop something? Breaking news. You heard it here first. Okay. So I have been trying to automate blog writing, just not to fully automate myself out of writing, but to understand the technology and like where it is on creative writing. And like a year ago, whatever, I used the OpenAI fine-tuning API. I created lots of examples. Here are the prompts. Hundreds and hundreds of examples. I talked to somebody inside of OpenAI. Okay, you're using it the right way. it really didn't work. I've tried, anyway, the long of it, the short of it is I've tried many different techniques. And the technique that is finally working with the latest models is I make every edit through AI.

26:59Tomasz Tunguz: So I start with an outline and then I write, and then I ask the AI to write the post. And then I'll go through literally line by line of the outline and then it'll write the post. And then I'll go through sentence by sentence of the post. And it creates a log, the AI creates a log. And at the end of it, it collapses that log into learnings. Like, this is what you did on this post. You changed the order to achieve this particular goal. You used a block quote here. You used an ampersand here. You didn't really like this very long sentence. And then at night, once a week, the AI goes through and says, okay, given all the block posts you've written this week, what is like the strunken white writing lessons for this blog?

27:40Tomasz Tunguz: And so I've been doing that for about three weeks. And then I ran the analysis of like, okay, there are versions of documents which are wholesale changes, versions of blog posts, and there are individual sentence edits. And I had expected that the number of versions would decrease and the number of total edits would decrease. And it's not true. The total number of edits is 134 per posts, irrespective of time. So, okay, there's one of two conclusions. Either like, I'm totally wasting my time. what am I doing with this, right? Or, which is what I believe, I'm delivering a better product. And so I had three AI agents go and scour grading criteria for like the economist, editor, New Yorker writing, AP writing, and then go through 15 blog posts in each of the last 10 years and grade them.

28:30Tomasz Tunguz: And in 2026, there is a whole number out of a scale of five, there is a 20 % increase in the quality of the writing as a result. Oh, interesting. So what does that tell me? I think it's like if you're a chess grandmaster and you want to be better with AI, you don't train less. You train just as much, but you hold yourself to a higher standard. And we've seen this, right? You look at chess, the grandmasters are scored on ELO, whatever, relative scoring system. And you look at the ELOs pre-AI in chess and post, they've gone up a ton. And so it's not going to spending any less time writing. I still allocate an hour to an hour and a half per post.

29:12Tomasz Tunguz: It's that the quality of the post, I look at like the quality of the analysis, the citations, the level of research is significantly better. And maybe that's what I'm aiming for instead of here's the blog post factory. But that's really, really interesting. Sam, do you feel like a grandmaster or do you feel like a grandmaster, a chess grandmaster? You're elevating your game up a little bit. I'm still terrible at chess and I'm aware that I'm terrible at chess. But I definitely think that I'm capable of writing better for sure. The best part of AI, if you have taste, is it's the same thing as GarageBand.

29:50If you had some elements of how to write a song, if you understood basic structure, basic chord progressions, you now have a tool through GarageBand and probably through Suno even more, where all of a sudden you can put stuff together that was not possible before. You had to know so much more about how to play every single instrument and how to keep time and you didn't have quantizing. And so now I can emphasize and support metaphors that were sort of theoretical or aspirational in my old writing, but now I can actually do the research very quickly and I can really underpin a point that I'm making with different citations in a way that makes the whole thing richer so i totally see what tomas is saying well on tomas's uh posts generally you're like you're really good at citations tomas like really really good at that's ai that's not me well and that's and that's where i think this piece of it that elevates is that the posts aren't very long but you cite everything very clearly like half the post is like a citation of callback.

30:49Tomasz Tunguz: It's like David Foster Wallace, right? I literally have to ask for this. All right. Let's shift the conversation and it's kind of like, let's talk about something that's, I think, a little underrated shift that's happening. We've talked on the show a lot about FDEs and that every software company is kind of quietly turning into a little bit of a service company. Tomas, you wrote about this in July. There are two posts. One is AI companies committed almost, I It's$10 billion in a single year, putting their own engineers inside companies and customers just to get the software to actually work, which is kind of funny.

31:28If I think about that 10 years ago, that was not a good thing, but today it's a good thing. And to put that in perspective, it's about a fifth of what Accenture spends in a year to deliver everything that they have. You asked a question in that piece, which is whether is this a moat or is this a toll booth? Does it actually lock the customer in or is it just a fee that they're going to collect while they keep on shopping? Because and we want to push on that a little bit because there was another post that you had about this AI colander piece where those two things sit diametrically opposed a little bit because you have it was like 40 percent of AI customers, first time users don't come back.

32:07And so the model is only really going to hold that spot for I think it was six weeks. So if you're keeping 40 % of your customers by month five, how does the FDE math work at all? I mean, I have an instinct that there's a B2BC motion and there's B2B motion. So they're probably a little bit different, but I wanted to hear from you on this idea of, well, we're putting a lot of money in. It's kind of coming through like a sieve, but we're also spending a lot of money up the enterprise on these FDEs for engineers to go into bigger customers.

32:38Tomasz Tunguz: Yeah, you nailed it. Wow, great synthesis, AJ. I think you're right. I mean, I think, so if you're a model company, you have, and you release a state of the art model, you have 41 days to commercialize it before you're knocked off that perch. And so that's kind of the state of the art today. You literally have 41 days. And when individual users sign up for your new model, whether it's a DeepSeek or an OpenAI model, you can keep them for a little while. And then all of a sudden they're going to go to the next thing. And that's, it's possible to make a business there work. Today, there are heavy subsidies because what we're doing with those subsidies is the same thing that we're doing with FTEs.

33:12Tomasz Tunguz: We are educating users so that they bring AI to work. And when we bring AI to work, we charge a company a whole lot more than we can charge individuals. And we can make a bunch more margin. And the way we do that is we look at a business and say, everybody was working this way. After AI, everybody needs to work this way. And we'll reinvent the process and we'll reinvent the software and we'll charge you a whole bunch of money along the way. And that's what's happening. and I think it's a really good thing right I think anytime there is a platform shift like this the reality is there is a tremendous amount of education like all of us live in this world and we spend a lot of our time almost all of our time maybe too much of our time trying to understand exactly what the implications are like every model release and every version and every new memory architecture but the reality is like many many people don't and somebody just still it and when they show up at work to educate them on what it should be and how to do it well and i think that's why we're seeing such a tremendous investment in fdes this episode of top line is brought to you by compound growth marketing the way your customers buy is shifted before they ever reach a rep they're asking an llm who to trust who to shortlist and why decision making is impacted by investment at the top of the funnel and most b2b marketing teams under invest because it's difficult to measure.

34:32That's the gap and it's expensive. Compound growth marketing bridges this gap for many of the fastest growing companies in B2B SaaS and security. Their predictable demand system unites paid media, answer engine optimization, and GTM engineering into one discipline engineered to produce pipeline you can actually forecast back by over 100 million of media managed. No vanity metrics, no guesswork. If you'd like a free audit on funnel media, including AEO reach, go to compoundgrowthmarketing.com and mention Pavilion. I want to tell you, I want to tell you something people. This episode of Topline is brought to you by DevRev.

35:04I've been really excited about this thing lately. It's called a computer, but it's not the computer that you know. It's computer by DevRev. If you work in GDM or sales, you know what it's like. You're running a dozen deals at once and it's on you to remember every detail. What the customer said, what you promised, what needs to happen next. So things slip. Follow-up gets missed. A deal goes quiet. Computer by DevRev is an AI teammate that fixes that. Computer uses something called shared memory to hold the full context of every deal across Slack, email, Granola, and Salesforce, even across the rest of your team.

35:35So nothing falls through the cracks and you always know exactly where every account stands. Your deals keep moving and you stay one step ahead. Take a look at computer.io forward slash top line. AJ, does that resonate with you? Yeah. As the CEO of a company that has a million systems plus, I am always trying to figure out context. Last week I was in New York and had in-person meetings with We had 15 in-person meetings with CFOs, finance leaders. And afterward, I turned to my VP of strategic services and I was like, Max, how are we even supposed to remember any of this? So having a system like Computer by DevRev would have put me in the driver's seat.

36:17I wish I had it. You're great. You're great. You're a good partner to these sponsors. Computer by DevRev. DevRev, we appreciate you. Computer, we appreciate you. We appreciate that you use shared memory. You're doing good work. Go to Computer by DevRev. If you work and go to market, I want to tell you about Pavilion's GTM 2026, our annual event. It's happening September 28th through October 1st in New York City. And this is Pavilion's biggest event of the year. And this year it's especially meaningful because we're celebrating 10 years of Pavilion. If you're a top line listener, A lot of familiar faces are going to be there.

36:59Asit and AJ will both be there and will be doing live top line recording. We'll also have longtime listeners and friends of the show like Jackie Robinson and Ryan Burke, alongside an incredible group of GTM leaders and thinkers, including Nick Mehta, Jeffrey Moore, the author of Crossing the Chasm, Alison Moore, the CEO of Chief, Eric Gilpin, the president of G2, Manny Medina, the co-founder and CEO of Paid, Linda Leanne, the CEO of Common Room, Amanda Calo, CEO and co-founder of OneMind, and regular guest on the show, plus many, many more. We're going to be spending several days talking about what's happening in go-to-market right now, AI, sales, marketing, leadership, company building, and how the GTM profession is changing.

37:41But the biggest reason to come is the people. There are people you want to hear from, follow, learn from, and work alongside. Also, it's a vibe. GTM is really one of the only conferences that feels like a college or high school reunion, depending on if you had a good time in one of those moments of your life. It's where the community of Pavilion really comes to life. And if you're curious about what that community feels like or looks like, come join us. It really embodies our values of give to get, of helping and supporting each other, and of propelling all of us as operators forward. It's a chance to get to know everyone together in the same room.

38:22So come celebrate 10 years of Pavilion with us in New York and with me, Sam Jacobs. Go to attendgtm.com forward slash register. If you could do that right now, I'd really appreciate it. Go to attendgtm.com forward slash register. Go there. I'll wait. Okay. Use the code ambassador20, ambassador20 for 20 % off your ticket. And that's because I view you, loyal listener, as an ambassador. of both the Topline and Pavilion brands. So go to attendgtm.com forward slash register, use the code Ambassador20 for 20 % off your ticket. And I hope to see you in New York in just a couple of weeks. I wonder if the Frontier, and maybe this is Asit, and you all will know this, are the Frontier models deploying FDEs at this point?

39:10Everyone is. Everyone is. Well, not everybody. And this is an interesting question is, Tomas, as an investor, if you've got two companies, both at 100 million, both solving a similar problem, same category. But one got there with FTEs, one got there without FTEs. Is the one that got there without FTEs solving a similar problem for a similar buyer like competitors, but one has it, one doesn't. Is the one that doesn't have it today more interesting to you?

39:39Tomasz Tunguz: Five years ago, the answer would have been unequivocally yes. yeah because this is sierra versus decagon right like this is that comparison basically today i don't know there's not a definitive answer i think the answer lies within gdr and ndr because okay so the benefit of software only is clearly much more capital efficient which as an investor phenomenal the benefit of the fde model is cross and upsell and customer retention And, you know, I don't know much about the details of Sierra Decagon, so I'm not commenting on those individual companies. I'm just saying in practice. And then the other question is, what is the margin structure of the business look like and how are the FTEs paid for?

40:22Tomasz Tunguz: Palantir pioneered a model where the FTEs were free, subsidized by really expensive software. So it looks like a software company. It doesn't look like a professional services organization running at 0 % gross margin. It grew into like 70, 80 % gross margins. Yeah. And Tomas, I think you actually laid out there were four different ways that this was, one was self-funded, like you just mentioned, but then you have, I think it was Google, you said that spun theirs out as its own entity. Partnerships, yeah. Yeah, and then there was pure partnerships, like the Microsoft type of play as well. Was there a fourth one that I'm forgetting as well?

40:56Tomasz Tunguz: There's like in-house. So we have a company, we have a bunch of FTEs. There is what OpenAI and Anthropic have done, which is they raise external capital. And then spin up an FTE organization that's partnered and coupled. There's the partnership opportunity. Yeah, because you're talking about the thing that they did with the Thrive and Blatstone, et cetera, like that's been out. Those are targeting a specific part of the market. In some cases, it's targeting private equity businesses. In other cases, they've spun out these consulting arms that go out and do this as like a consulting company being your FTE.

41:34but they also have their internal ones that they use for difference so they're taking kind of this hybrid approach that's right

41:40Tomasz Tunguz: yeah you're exactly right that's a good refinement do you think FTE today is it doesn't you know the way that gross margin no longer matters like it just doesn't like no one cares like if you look at these inference providers I think the best inference provider is from a gross margin perspective is fireworks today at 30 % 25 to 30 % gross margins so it like And somebody did this analysis recently, like up to a certain point, it just really doesn't matter right now. Eventually one day it will, but like right now it's growth over everything else. You'll figure it out based on a bunch of things. Just like that, like FTEs today are an inefficiency.

42:19It's not like, it's not beautifully elegant when you need to be like, here's a bunch of humans that are going to wrestle with this in your business to help you make it work. And then they have to keep doing that every time the model changes. It's not elegant, right? It's kind of like analog in a way. But as the intelligence gets a lot better, there's one argument that five years from now, if AI keeps improving, no company will need an FTE because the AI itself will have the capability to play that role of helping you integrate it, helping you get the best out of it. How do you think about that?

42:53Is this a short-term phenomenon?

42:55Tomasz Tunguz: No, I think it's a very long-term phenomenon. FTEs are here to stay, basically. Oh, yeah. I mean, management consultants have been around for forever. The reality is I think we always will need someone to validate that the AI is doing something properly. Who's an expert in the field, right? If you're a structural engineer, you stamp structural diagrams. If you're a FDE for ERP implementations, you will stamp the work of the AI. We have this concept in AI which is called the scaling law. It parallels Moore's law. So Moore's law said performance of a GPU doubles every two years, I think is what it was.

43:33Tomasz Tunguz: And it went for 70 years unbroken until the mid 2000s. And we have a similar law within AI, which is the more data we feed into these systems, the better they become. More compute. And more compute. And that is unbroken. I'm stunned. I studied machine learning in school and we were taught like there are all these different algorithms that I can name. And most of them were useful for two or three years. The transformer, which is the current one, is almost is approaching 10. and there seems to be no limit yet that we can see that the innovation is flowing down as a result. And so as we get to more and more greater capability and we double the amount of tokens that we can generate every year, we'll still need people to re-educate.

44:15Tomasz Tunguz: I mean, like, okay, let's make this more concrete. How much have you changed your AI work since GPT-3.5? I mean, how many reinventions of your engagement with AI? It's like four, five, six, once a quarter maybe? Yeah. And so if you that's the purpose of the FD is how do I translate this rocket ship of a technology into something that's actually relevant and useful in a particular organization. And the companies you're investing into, Moss, do you see any change in the structure of the go-to-market organizations? We talk a lot here about what's the future of the sales organization and do sales teams get bigger, smaller?

44:52Do you have a point of view of that as you're watching, you know, your investments season and grow? Is it exactly the same to the point of FDE? Is it, hey, we're going to need the same number of people, but we're just going to have higher expectations on what their capabilities or productivity are?

45:07Tomasz Tunguz: It's much more the second. So I think inside BDR, automation, I think is working. Vercel is at 90 % plus on that. Outbound, not so much. AE productivity expectations have shifted up. There's no doubt about that. Quotas, you know, like five years ago, Oracle and IBM had the highest quotas in software, somewhere between two and a half to four. And now it's kind of routine to see an early startup at like one and a half. if you're an insurance provider, you might have like quotas in the tens to hundreds of millions for a particular account. Like that is radically new. I don't really measure companies on like AE to SDR ratio the way that we used to or AE to CSM that doesn't really exist.

45:50Tomasz Tunguz: It's more just like raw sales efficiency because the sales models are so different that you can't really apply. I just can't even imagine from a PE market or an M &A specifically that are buying companies that have traditional sales. How do you even grade a sales team when you're looking at to understand whether it's a viable distribution model that's going to scale in six months, especially when the way you work, like you just said, is changing two to three times a quarter now. Say more, AJ. What do you mean? I just think it was like if you look at, okay, so Vista. And you look at how Vista's bought companies.

46:24They've bought Sales Loft or they bought Drift. They bought Clary. They bought all of these companies. and you previously could be very naming the best investments no I'm not well I'm naming the investments that are in the go to market because they're they're clearly like top of mind do you think they'll ever do more go to market acquisitions no I'm not I'm not talking this personally but you you knew that there was this very predictable path you had this playbook you graded a sales rep on these characteristics on this thing curiosity or whatever it was that they had and that is just all thrown out the window but even at the strategic level when you're grading and looking at a company and looking at predictability around their distribution of like okay we're going to forecast 70 percent growth i don't even know how you would trust those numbers you could you could look at the quota to ot ratio last year and say it was one to four should be one to five or should be one to six next year but should it we don't know what the top end of the market is going to look like and i i guess i would just wonder if i'm in and i'm not so maybe tomas has some insight here just like if i'm looking at a company to buy or to even to invest in, well, to keep it lower level, like that go-to-market motion has changed so fundamentally.

47:40Do we trust it? Do we think that this is still something that's a viable investment? I think it's different based on the type of business you're looking at, because if you're looking at a venture investment right now, these companies are kind of discovering how to run themselves in this new paradigm. And so everything is evolving. And really what you would only care about is a very like at a macro pnl level is it is the machine making sense like you know is it at a very macro level and then if it's making sense where i'm willing to put this much money into it because i think this will come out of it how underneath the hood they do that is is the cool thing that we all want to see it might not even matter though asa because at the the these ai harness companies infrastructure companies fireworks whatever they don't even really have internal ops right now.

48:31They're growing so fast. I was talking to the CEO of a rep ops company. None of them have, they're all white space to play. You don't think that's true? I think people say a lot of shit. I think a lot of it is just shit. I don't think, listen, I work with a bunch of them. I can tell you that they're not as AI pilled internally as you would like to believe. I think they're also trying to figure out how to use it effectively in all of these different departments. There's some departments where it looks like magic is happening. Like when you go sit down with the engineers and one of these companies, you're like, fuck.

49:10Yeah. But then you go to the go-to-market and you're like, oh. Then you go with HR and you're like, on a pad of paper. That everyone else is doing. Yeah. Like it's not that different. Like I don't think that something has dramatically shifted. Now they might say, hey, our account executives are so much more productive. Is that a measure of like how well you're using AI that has somehow made this account executive more productive? Or is that because demand is flying at you and you're closing 90 % of all deals? Right, who cares? Because people are experimenting. So there's a little bit of like, you have to look at it with squinted eyes.

49:43Tomasz Tunguz: Yeah, that's the right distinction. It's not the supply side has changed and suddenly become 10x more productive. It's the demand side budgets have increased by a factor of 10. And that's what's driving quotas. And you can't take that discipline from one company to another. And there's no CRO who will go from fireworks and then go into Salesforce. And then all of a sudden, E's or Jenner are bringing home 10 million a year, right? It's not an internal discipline or a practice. It's a market dynamic. That's an excellent point. In venture, I think you analyze a company at a macro level. Like, does it make sense at the macro level.

50:19Now let's see how they get there and what they do. I think the private equity world is far more similar to the past than we would like to believe. I don't think it's dramatically shifted. I just think that they, I work with a lot of these large spinouts and M &A moves. They are very much the same as what 2022, 23, 24. I haven't seen some dramatic shift. they talk about it a lot but i don't see a massive shift i think the main thing they're looking at is that can this company do well in a world where there's ai like they're looking at it from a can this company have demand for its services or its product in a world where ai is important to the customer that's the number one question i've yet to meet a private equity-owned business that is ai built internally in a massive way sam with gtm upcoming and thinking about this as like two different tracks where you had the traditional CRO, which was very well known, and you maybe have an evolution from 25 to 26 versus what Tomas just described on the demand side, which is a totally different ballgame.

51:26Does the programming for GTM change to match that? I think we need as much representation from the companies that are growing this quickly as possible. And that's the nice thing about having like a sponsorship angle, because I think if we didn't have some of these companies that were growing by leaps and bounds, I think it would just be completely irrelevant conversation. And so at the same time, sort of to Asit's point, we need to look at, is Ghazi Massoud doing something materially different at Replit? Or does he just have massive demand? Does he have something, and we liked him and he was on the show, but my point isn't about him specifically, but does he have something interesting to say that is specific to the fact that it's an AI native company?

52:12Or is it just, listen, I have a very high degree, you know, I only hire A players, I only date supermodels. And you know, I really believe in making sure that people have as much enablement as possible. And I invest in RevOps at an early stage, which you could say at any year over the last 15 years. So I think the programming needs to change to reflect what's happening. But to the point of this conversation, I'm not sure that the actual content. I don't know, again, the people that are doing it, Kyle Norton will be there and he'll have a lot to say about how to empower an SDR organization with the right data infrastructure so that you have higher call to connect rates than you would have had otherwise.

52:51But, you know, I don't know if that answers your question. The question I have for Tomas is, do you think customer success now goes away? Because if there is a world in which FDEs are long lasting, it makes a lot of sense, right? You've got probabilistic technology, it changes its jaded and its capabilities. We have so much to learn. Like right now, we keep talking about the cost of tokens. And it's like if you make a stupid model that has a cheaper cost per token, do a task that takes it a long time to do, it might end up being more expensive than taking a model that has a higher cost per token, but can just one shot the answer, right?

53:26And we're like starting to learn these things. So we're really early. But a lot of these, like the jaggedness of it will persist. And so we will need FTEs to help companies with this. But then what's the value of a customer success team? If the account executive is staying from land to its band because of the pricing model and the FTE is hurrying the customer to help them figure out how to use this, does customer success just die?

53:48Tomasz Tunguz: So I'm curious to hear your reactions on this point. And I'm going to make it with an analogy. Sales engineer is to CSM as FDE is to AI CSM. What do I mean by that? I think the sales engineer of the previous generation was the most technical customer support and success person you would have. You stayed within the organization, implemented really difficult software, worked really well, snowflake famously, right? No more CSMs, products too technical. But if you were like an app, if you were a relatively straightforward app, it doesn't make sense to hire an SE, you'd hire a CSM. Good enough, right?

54:32Tomasz Tunguz: And so your ROI on your investment in NDR is significantly higher if you're an app on a CSM than an SE. And so maybe that analogy extends into this era, which is if there's a lot of configuration and a lot of business process management, okay, FTE, very technical. But in the world of, I don't know, let's say like a new task management company comes around, very unlikely that they'll hire an FTE for any lengthy period of time, right? And so I think it has more to do with the complexity of the software and the ROI on an NDR investment. dollar. Okay, so as an investor, then you've got this spectrum that you've defined, right?

55:16Which is a really interesting spot. I think it would make sense because we see it. This is how it played out in the SaaS era. So it kind of made sense how you're pulling this forward. As an investor, you've got money to deploy. How do you feel based on just the complexity of the thing that they're building? Is it that like, because building stuff has become so much easier, you feel more secure investing into a company that's building something complicated that will require FTEs because it's going to be harder to displace them, harder to catch up to them, product advantages might hold a bit longer?

55:53Or, you know, how do you feel about this?

55:55Tomasz Tunguz: Yeah, this is a great question. Okay, so in the beginning of SaaS, no one invested, I'm simplifying, no one invested at NSMB, turn rates were too high. No one invested in enterprise, sales cycles were too long. There were not enough enterprise buyers who really cared about cloud. That was the market in 2008 or in 2010. And so sweet spot, mid-market, inside sales driven, faster sales cycles, get to a predictable business. You're asking the question, Asad, has that changed? I think the answer is yes. I think on the whole, we probably have shifted more to focus on enterprises. is. One for the reasons that you mentioned, which is at what point will the models obviate very simple software?

56:39Tomasz Tunguz: And you can see that in the public markets, right? You can see that what's happening with like the task management, project management companies are really having a tough go of it. I think the other part is going back to the point we were talking about earlier, the demand side of the enterprise is huge. Like you, you know, we're talking about lands at enterprises as they're measured in the tens of millions. I mean, it was unconscionable five years ago for an AE at a Series A company to close a$10 million deal. And it's in 45 days. I've met many that have done 20 million in a year now. It's crazy.

57:10Tomasz Tunguz: And so why would you forego that in order to go into the mid-market? Because you have mid-market sales cycles. And they're buying faster as well. Like the enterprise used to take a year and a half. Now they're like two months in. They're like, let's go. It's different. Do you have the best of both worlds? You have a 45-day sales cycle of the mid-market and you have the seventh year bookings of an enterprise client, right? It's not like a year one, 75K land, I'm gonna work my way to the org. No, no, no, we're talking about, you know, a nine or a 10 figure deal. And so I think that's why everyone shifted enterprise.

57:41I think like if I look at factory versus lovable, you know, kind of like in and around like this world of helping you write code, right? But different buyers, different, I would feel more secure investing in the factory who's going after the enterprise, because it's like, while the world around them changes, if it's working well enough, like an enterprise doesn't like ripping something out. You know, there's like, where's like a smaller company or a consumer rip things out like this. And so to invest the money in something that's more durable would be where the attraction is. But is that how all investors are going to be thinking?

58:18So then there'll be this alpha on the other side, or you think there's like a deviation?

58:23Tomasz Tunguz: No, I mean, lovable, I think to your point, it's a great example, right? Totally PLG business, very successful enterprise motion. And so it looks like a classic PLG to enterprise, but accelerated timeframe. And those can work incredibly well. They can also work incredibly well. It's just, you have an evolution in the business model, which adds a layer of complexity to the business. But if you can make it work, the capital efficiency is incredible. On the founders front, so everything is changing here, right? Like we're talking about you can build more product, who you sell, like so much change is happening, which is why this moment's exciting.

58:59You are in the business of deploying capital into somewhat early stage businesses. And so you're betting on somebody being able to fulfill this promise and vision and story that they've told you. And there's some proof, but really it's a bet on the person to figure out a bunch of shit that's going to come their way out, right? Right. And before there were playbooks like, OK, can this person run these playbooks was a little bit of where we were in the last few years of SaaS. Right. And now it's like, will this person be able to figure a lot of new things out? Because there are no playbooks right now.

59:32And so how has it changed what you're looking for in a founder or a founding team? Like, has anything changed about the composition that you look for or the type of people that you're willing to bet on? or is it just the same as before?

59:48Tomasz Tunguz: It hasn't changed a lot. I guess the point I'd make is, you know, Dropbox is a really successful PLG company. Zoom reinvented the go-to-market around video conferencing. You know, you look at many of the big companies in the 2010 era innovated pretty significantly on their go-to-market strategies. Open Source became incredible, right? With some huge businesses in ways that were entirely novel. you look at like um confluent right like who would have thought you could go open source directly into the enterprise or hashi corp which had six products at ipo only company i know of that had six companies at ipo the only one i know it was atlassian so it was a breadth angle and so i think all the great companies they they find an innovation in an existing market prime there's a technical innovation but i think more and more maybe this is the change that i'm coming to which is the go-to market innovation is now significantly more important than it was.

1:00:46Tomasz Tunguz: And if you can find a founder who can execute a beautiful go-to-market judo move and produce a lot of leverage for the company, then it's incredible. So I think that's what we're really after. I think that goes back to what Asad said at the beginning, which is like, okay, you had these three use cases, customer support, chat, and coding. The go-to-market innovation that's happening is happening very fast. It doesn't feel like it's coordinated. It feels kind of disoriented, almost like companies are keeping secrets on how they're doing their go-to-market differently because outbound has been so terrible.

1:01:22Like the way distribution has been happened has been so challenging that if you find something that is working, you hold on to it and don't let that cat out of the box, so to speak. I don't know. Asit, are you seeing that as well in go-to-market? I think what I'm noticing right now that's most interesting to me is how so much is still the same as before in terms of you know at the end of the day if you look at the CROs of the top four or five AI companies they're the best CROs of the SaaS era it's DALI at Okunai it's a bunch of like old PTC folks and then the Service Now crew like the Service Now crew is at Anthropate, Salesforce, PTC, Oracle like they're kind of everywhere.

1:02:04It's just interesting. And they're like building the teams that you saw at AppDynamics, MongoDB, Zscaler. These teams look very similar to the teams that were built over there. It's a lot of the same people that were there. And so I think it's interesting how I think the design is somewhat similar. But within that, there's some cool shit that's happening. Reps are more prepared. And, you know, they might go into a meeting in a bit of a different way. And, you know, Now I don't have to populate my CRM as much. So I spent 10 % more. So they're like these like interesting things that are happening.

1:02:37But design is the same, give and take. And like, that's a really interesting aspect to this. And how once we've got into this stage in the process, we've decided the people that we don't trust the most from a global market perspective are the gangsters of the previous era. We're not giving it to like a bunch of new people and being like, they'll figure it out, which is what is happening on other parts of the like in product and engineering you have a lot of like young new product leaders cto cpos on the global market side it's like let's just trust the old guard well deals are as competitive as ever that's the really awesome thing about this is that like when you have these competitive deals it's actually helps lift everyone up to build these type of teams because everyone wants an edge so you do have the oracle and ptc Salesforce Air folks, is that's what gave you.

1:03:24Dave Kellogg has a really good framework for this. He's like, the bigger the size of the old, the more you bet on experience. The smaller the size of the old, the more you bet on slope. And I think that's an interesting, global market's huge, right? It's like still 30, 40 % of a company's headcount and you're going to bet on experience there. Okay. We are coming up to the end, but before we go, we have one more segment, which is our favorite Bulls versus Bears.

1:03:49Tomasz Tunguz: Bulls versus Bears.

1:03:54Asit, I'm ready to go on this. I've prepared this. So I'm going to take it. Tomas, for you, this bulls versus bears is I'm going to give you a question and you have to tell me whether you're bullish or bearish on that. So 12 months from now, today, 50 to 100x multiples on AI harness companies. Bull or bear? 50 to 100x multiples on an AI harness company. That's a good question.

1:04:22Tomasz Tunguz: They're there today. AI harness companies are at 100x or more. Oh my gosh, this is so tough. That's a great question. It's so tough. Okay. It's so tough because of rates. That's my consternation. You look at what's happening in the treasuries and the sovereign debt market, and everyone is selling sovereign treasuries because there's sustained inflation. And it's very likely that Warsh will raise rates or that's what the market is feeling. and if you know if like the 10 year and the 30 years start to hit record highs then I just don't see it is the open AI and anthropic IPOs especially the anthropic IPO the bullish sentiment coming out of that strong enough to overwhelm the bond market and the reason I'm going to go bearish is because the bond market always wins do you think that like the counter argument would be that compared to previous situations where so much more of the economy is so all in on this like if you think about like who all's like in you got the banks you've got the credit providers the pe like everybody is just like in on this thing that like there's some level of a lot of fail we can't let this thing go up like it will literally bring everyone down that's kind of the countermeasure to some of this like does that give you some like oh yeah it gives me a little bit of peace of mind because I'm like, everybody together is not going to let this go on.

1:05:47Too big to fail. It literally is too big to fail.

1:05:49Tomasz Tunguz: It's too big to fail. I mean, you've seen IT as a percentage of US GDP go from 4 % to 12%. I think it'll be 15%, 16%. You have data center buildouts being three quarters of US GDP growth. That kind of go away. Everybody understands that. And you have like Texas blocking data center buildouts and New York blocking data center buildouts. AI is going to become a part of the election. You have the interest payments, which are absolutely enormous for the treasury, which means that we need to sustain 4 % annual inflation for the next 20 years. You have a Fed chairman who wants to raise rates in order to minimize inflation.

1:06:25Tomasz Tunguz: We have a president who doesn't want that. And so you have like, you know, all of a sudden, tech used to be the IT guy in the corner fixing the printer. And in 20 years, we've become, you know, and I don't want to be too grandiose about this, but like the center of the global economy, right? The thing that I've found in this is like, I think the countermeasure to everything is, or like the thing that I think saves us all from this blowing apart and not having like a bubble type of burst, but actually having like these mini corrections along the way, the way we had in July. I think there'll be a lot of those.

1:07:00Let out some air, normalize, grow back. Is the fact that these models keep improving at such a fast rate. these were the Mythos, Fable and I think GPT 5.6 were the first models I think released to the public that were trained on a GP300 and you'd want to get these Baro Rubens that are so much more efficient so much more powerful they're getting into deployment and I think February is the soonest deployment at scale and you're going to get these models trained on those and those are going to be that much more powerful so the scaling law still drives forward and you get the improvements there and then there's these breakthroughs like if they figure out something like you know recursive self-improvement.

1:07:40It's going to like open up a whole new amount of revenue because you've got these enterprises that are buying because they are scared and they see it as an opportunity at the same time. It's like fear and excitement all in one. And that's a really interesting like dynamic. I think you...

1:07:57Tomasz Tunguz: I know, but you need somebody to lend the money. You need 75 to 80 % of the data center build out to be lent. Yeah. And who is that lender? Are we tapped out from a global liquidity perspective? Well, I mean, private credits, you know. Pretty tough. Yeah, private credits, so they're going after retail. We have the yen intervention, and now we're going to get into macro. Like the U.S. is intervening in the yen market because the Japanese aren't buying as much U.S. treasuries. The Chinese have sold off, I think, 30 to 40 % of their treasuries. And so do we print money, right? And the U.S. government with treasuries is now backstopping the development of U.S.

1:08:35Tomasz Tunguz: data centers. Sure, okay. But then you have more inflation and then you have the wrong cycle. You have more inflation, which means rates need to go up in order to - Consumers suffer. And then consumers suffer. I mean, that's a very, very bad situation. This is actually a really good point. Yeah. This is the pickle. It's like, it's the credit market. The credit market is the key. Because you saw the Nvidia thing, by the way, when Jensen came on with all those like bankers and was like, here's this$500 billion and backstopping type of thing that they were doing. It was strange because it felt small.

1:09:05I don't know how it felt to you, but 500 billion just felt like all this headache. You guys have all come here to sit on a table for 500 billion. What if you're going to build one data center now? Like, great. It felt small. That was the amount of the, what was the bailout in 2008 or 2009? It was exactly that number, Asit. It's not, that's the thing that's so crazy about this. This is the point that I was making that you guys gave me those wild-eyed looks about. This is my whole point. the whole point is that the a a true it's that consumption like the only thing that can that can make these debt payments or overcome inflation of these interest rates is actual real gdp growth is actually the fact that like ai better freaking work like we better we better all start running consuming a billion tokens a day and like achieving great outcomes because that feels like the only thing structurally that's gonna otherwise if we're just staying still but continuing to borrow from the future.

1:10:02I don't see how we get our way out of this. I don't think it can happen because like, I think where this goes eventually is the government funds it and we get crazy inflation, but it's like, even if humans start using it 10x more than we're using it, and we're getting all this value, etc. You will still be like, man, imagine this AI is three times better in two, three years than it is today. We need so much more data center capacity two, three years down the line. We have to fund it and build it. You'll always be on this thing where you're trying to create capacity for three years out and you're modeling improvements and more usage.

1:10:36And so it's like at some point, like, well, how do you fund it? Like it can't be cash flows. Cash flows can not fund this shit. Well, I had equally as great questions for Sam and Asit, but I think, I think with that question, we're at time, Sam. So I'll let you take this. Well, thanks for being our guest on the show. It's always great to see you. folks if you're out there listening give us five stars wherever you rate your podcast please watch us on youtube and click that subscribe button that'll help us a lot and we'll talk to you next week tomas thank you pleasure with mine see you guys

1:11:15if you are looking for more top line check out the top line newsletter at toplinemedia.substack.com. That's toplinemedia.substack.com. Or if you're a video person, check us out on YouTube. Topline-media is what you want there. Have a delightful day, everybody.

1:11:50Thank you.