2026年预测：最佳买入与最大做空 | Salesforce为何可能胜出，以及NVIDIA面临的挑战

Guest

I think we see 4 IPOs backloaded next year. SpaceX goes out first. Canva comes out of nowhere second. Databricks does it in the back half of the year because it’s just time. Anthropic does it at the end of the year. OpenAI probably should have gone first, but it’s burning too much.

There’s no ceiling on venture, which changes all the math and calculations.

Rory O’Driscoll

We haven’t missed the boat, because what agents can do—we just started. If unemployment ratchets up even 2 or 3 points, regardless of the reason—it could just be the business cycle—you’ll see a techlash that makes what we’re dealing with now an understatement. Society will be terrified of AI.

Guest

If the robots are going to take over society, I want to be sure that I own the robots.

Ready to go, boys. It is the Big Fat Quiz of the Year. I don’t know if you have this in America. The Big Fat Quiz of the Year is like a comedy show with Jimmy Carr where they do a review of what’s happened, and then they do predictions for next year.

1. Founder of the Year 2025

I thought I could be the Jimmy Carr in this scenario, and we could do a review of what happened in the last 12 months and then look forward to 2026 with some predictions. So, we’re going to start with number 1, the most important part of all of our ecosystem: the founders. Founder of the year for 2025. Who would you have as your founder of the year?

Guest

I know what the product of the year is. The product of the year is Claude 3.5 or 3.7. Without this, we have no vibe coding. Without this, we have no Lovable, no v0, no Replit, no Cursor that really works, no Gamma—none of these products, many of which existed for years, worked until we could debate when it was.

They worked with Claude 3.5 at the end of last year, but they weren’t great until 3.7, and then 4 this year. And 4 changed our lives. You may not use it; you may use the outputs. You may use Agentforce, which isn’t all that. You may use Cursor, which is still 90% powered by Anthropic. You may use Gamma or Replit or, again, Lovable. For Harry, none of these products we use changed the way we deliver AI inside products.

It’s not just coders. It changed everything else. So, do I give Dario credit as CEO? I don’t know. I don’t love all those warnings about how 90% of the world is going to be unemployed in several weeks, but he delivered, right? I think, without question, that it is the software of the year. Do you give the CEO credit for the product of the year? Maybe not. I’m going to anyway, though, Rory.

Rory O’Driscoll

Damn. Because, you know, I hate agreeing with Jason, but I’ll just take a while to get there because, first of all, he didn’t answer the original question. The whole product of the year—focus on the founder of the year.

Anyway, look, I’m going to start with Harry. I’m going to be grumpy today because I hate these events. I hate these kinds of wrap-up things. I think they’re bullshit, so I will be grumpy the whole time, and I make no bones about that. Also, doing 2 of them in 1 week.

When we were waiting for you, I was like, “Rory,” and he goes, “Fuck off.”

Rory O’Driscoll

So, secondly, let’s talk about your founder of the year. The problem—one thing about that is what sounds like the Ballon d’Or, where they pick the best football player, as we call it, or soccer, as they do here in America. Every year, they try to come up with someone else, but we all know it’s Messi. There’s a 10- or 15-year period where it’s just silly, or maybe it’s Ronaldo.

So, I feel that if we really come to pick founder of the decade, founder of the blah, it’s going to be Elon. Let’s just leave that to one side and do this year, and exclude him from the year because otherwise it gets boring.

I do actually come out exactly the same place Jason did: Dario. I do think it’s not just product of the year, but also—what I like, I find the stuff about unemployment to be bizarre, and I think he’s wrong, and I’ll come back to that. But that’s the CEO as pontificator.

As the CEO—as a business executive making a difference—I think he played a very steady hand. By going for the sensible play and the sensible party—to make another Monty Python reference, because it’s been that kind of day—the sensible party versus, obviously, the other party, he’s cranked out, “I’m going to get profitable. I’m going to be sensible.”

You look at a growth rate faster than OpenAI, valuation convergence even after OpenAI’s potential new round at $800 billion. I think he’s played a great hand. So, if you owned those stocks at the start of the year, this is the stock you’d feel most excited about during the course of 2025, among the megacaps, the great big AI companies.

So, I give him founder of the year again, which is different, as I say, than agreeing. I do not think we’ll all be unemployed by Friday week. I think that’s all nonsense, but we can come to that another day.

I’m going to add 2. One is a cheat because she’s not quite the founder, but I think Gwyn Shotwell, running SpaceX.

Guest

I totally agree.

I think it’s going to be the biggest IPO of all time. Her navigating pretty challenging times geopolitically with Elon’s brand this year, and navigating that so well and elegantly as CEO, is exceptional.

Rory O’Driscoll

I’ll give you half my 4-out-of-10 chips back for that.

Yeah, that’s a good call. I’ll give you 4 chips, and then I’m going to surprise you with another.

Guest

Yeah.

Can I add one for fun where I’m not an expert? You guys could chime in. But I really think, to me, a space I don’t understand and thought was borderline ridiculous is Vlad at Robinhood.

Obviously, you’ve done this, because to dominate a market and then go multi-product—the way they’ve done it—and I know Rory makes the point that this is the whole point of fintech, right? It’s to build an Uber app that does everything. But the execution level here, from revenue to stock price up 220%, from just becoming almost a meme app for kids to becoming a trusted app for adults—I mean, JFC. I wish every founder, I wish half my founders, could execute half this playbook.

Guest

9 products do over $100 million in revenue.

Yeah, JFC. How many of our founders are talking about maybe getting another agent into the market next year? Vlad would be like, “I got 50 AI agents done over the holidays, guys, and they’re all doing $100 million.”

Rory O’Driscoll

Vlad for SaaS CEO. Vlad for SaaS—bring him into SaaS.

2. Fund of the Year

I like it as an addition. I can totally see that. Vlad is incredible.

Okay, we’re going to go to number 2: fund of the year. Which fund deserves the title as fund of the year? Again, with the continual caveat that I despise this process, I’m going to answer it specifically on performance for the year.

I think it’s Index Ventures: competent execution across multiple different exits, because exits are the coin of the realm. The only fair way to measure things is: did you end up owning a lot, having big ownership in big exits? Then you measure that for the year in which it happens. You don’t try to account for other years, and then it’s at least objective.

They had Wiz, which, for the record, hasn’t closed this year but was announced in March of this year, where they were an early investor. They had Figma, where they were the seed investor. It’s not an exit, but Revolut, where they just raised at $75 billion and they have an early position in it, and it was done by a number of different partners.

I just love that. There are a lot of people talking about new venture models, and we’ll talk about that in a second. Index just took that same old model, which I like because I’m a boring kind of guy, and executed to perfection.

So, I give him the nod for 2025.

Guest 3

I'll give you 2. Rory's right. At the end of the day, venture is ruthless capitalism. You have to judge folks on either present or coming returns. That's it. But multiples are another way to judge firms, right? Even if the absolute ones won't be high.

We're part of social media, and there's aesthetics. What is the most aesthetically successful CEO or fund? I'm going to vote for Neo. I'm giving Neo my vote.

Neo, if you look at aesthetics, let's look at it: first money into Cursor and [unclear company reference]. Pretty hard to do for someone who, 24 months ago, seemed to be having weird arguments with Gary Tan on Twitter. Then he takes that and relatively hustles—literally hustles—his way into Cursor, hanging out at MIT and giving programming tests himself. I think he was actually the second investor in Cursor, but maybe the first pseudo-institutional investor, putting his own $5 million into Cursor out of his own pocket.

Maybe his absolute returns won't equal Index, or won't equal the average GP sitting at that—where's that fancy office, Harry, next to Burlington Arcade where I saw Index when I was there? What's that?

Whatever it was, it was very nice where Index is, right? Maybe Neo isn't going to—maybe all that hard work won't equal one nice nine-figure managing general partner check at Index. But from the aesthetics of venture, I give Neo a vote.

I also give it a vote because, sometime in AI time—12 months ago—it seemed like Y Combinator had won the accelerator race. Since then, ironically, the Bay Area has pulled away even further. But Neo, South Park Commons, and so many others are exciting: HF0 and the rebirth of the accelerator. The fact that Neo got so many hits so quickly is impressive.

I would say I'm impressed with Index. I'm a little jealous of Neo. Creandum is my second vote, notwithstanding the European bias here. Those are my top 2 votes for the year.

And Creandum—why? Expand.

Guest 3

Trade Republic just did a deal at $15 billion. There was that little one. I'm tired of talking about Lovable. Is there an E in that one or not? I can't remember. Klarna wasn't bad. It might not be a Revolut, but that was a pretty good one this year.

What I mean is their ability to—I mean, obviously, it's a global fund, right?—but their ability to do pretty darn well in Sweden and project power. Harry, you're the European, the head of everything European, but for me, as a Yank with some European investment, I don't think it's a quiet giant, but I'm pretty impressed. Creandum have created a phenomenal, premier, dominant brand that is rivaling Index and Accel in a way that very few people have done in the last few years.

That's why they got number 2 on my list.

Guest 3

I'm really worried that Harry's going to start on his “Europe is Europe” or “Alias” chant now that we have 2 of the top names being from Europe originally. I just don't want to listen.

I'm actually going to break it down. Seed, you've got to put Hummingbird up there.

Guest 3

That was third on my list. I just thought it was an obvious choice, Harry, so I didn't include it. But that was actually third on my list, right after Neo and Creandum. I just felt it was obvious.

A $100 million fund returning $800 million. They were the first-ever investor in Lovable. Kraken continues to be a great company, where they were a very early investor, and then they did have a $3 billion exit this year.

Guest 3

Wait, in all seriousness, $100 million to $800 million—is that what you said?

Yeah. It's a $100 million fund, and they made $800 million on it.

Guest 3

Yeah.

Okay, wait—the fund did 8x, or the partners made $800 million?

Guest 3

No, the fund did 8x.

Okay. Honestly, I don't want this show to go forever. I don't feel like I would get full kudos from my LPs for that performance today.

Guest 3

Give me a break.

You're laughing? You're laughing? Honestly, my anchor is the same as Neo, the same as Initialized, the same as South Park Commons. I'm not in the top, the very, very top group of my peers. I would have to do 8x even to get to speak at their AGM. Seriously, this is funny.

If this was the first podcast of this series, I wouldn't have said it, okay? There's no effing way. But in the age of AI, if you're a seed fund and you bought 10% of Cursor—and Neo didn't get 10%—if you got 10%, 8x would seem like a rounding error. You would invite those guys to dinner.

Guest 2

Jason, I get you. I'm just like, I'm the one getting in trouble for irrelevance, and he's saying one deal returning 8x a fund isn't enough to be called to speak at an AGM.

My seed is Hummingbird. My Series A is Benchmark. I'm sorry, but likely Manus, Sierra, likely Fireworks, likely Lora, likely Cerebras, and likely Lang Chain—my word, that is one killer fund.

Guest 3

Yes, I agree. I think what's interesting about them as an early-stage fund is that a lot of folks have wrestled with how to win in this market and have internalized that one of the “easiest” ways to win is to have super-big funds and just deploy lots of capital at the late stage and make the math work.

What I find super impressive about that fund—I think it's the 2021 fund—is that they stuck to their knitting and still have compellingly good overall returns by just doing the thing that they do. In a way, it's a little like my comment on Index. Because that's the milieu I live in, I'm always impressed with people who just stick to what they're doing and do it well enough to make it work, and don't get distracted by a different strategy.

The horrible thing is getting distracted by a different strategy, trying to do it, and failing when the thing you did worked. I think it points to another bias I have, and all these comments make it clear: in venture, there are a lot of strategies that can work if done well. I think the mistake is often to say, “X strategy works and Y strategy doesn't.” The truth is, X and Y strategies both work if done well, and neither works if done badly.

You have Benchmark, who've killed it just doing Series A, and seed and A, exactly what they said they would do 20 or 30 years ago. You've got firms like Andreessen who've killed it by getting huge and making 15x on a late-stage fund. It turns out there are lots of ways to make money. You just have to be good at doing it, right?

As you say, Hummingbird is doing it by, according to their own statements, picking psychologically damaged human beings and backing them to the hilt. It's a wonderful, wacky world.

It should be for you to raise.

Guest 2

This is true.

Guest 3

Speaking of discipline to stage, the growth fund of the year for me—we can't not—is Josh Kushner and Thrive. The expansion of fund size, the expansion of vehicles, the holding vehicle, and now the roll-up play. OpenAI, a central position in Cursor, Databricks, Carvana, Revolut—I mean, what a year.

The other one is Founders Fund deploying $3 billion in under 12 months, but deploying it into Stripe, SpaceX, and likely Anduril. You're like, as an LP, “What the fuck? My deployment period has gone way down.” And then you're like, “Ah, it's SpaceX, Anduril, and Stripe. Fine. Keep going. Keep going. Great. Good work, guys.”

Guest 2

Agree.

My odd runner-up for the fund is Google, because they did their big deal.

It may not be at a fund structure, so I'm having a little bit of fun with Google owning almost 10% of SpaceX and perhaps 14% of Anthropic, depending on dilution right now. Obviously, now this is the greatest circular investment of all: owning the vast majority of [likely Waymo], which is going to blow up.

Those 3—SpaceX, Waymo, and Anthropic—to have double-digit ownership, even if it's done in a circular way in 1 of them, I give that corporate investor of the year.

Guest 3

I think it's a good nomination.

Very good addition. The other one I had, just as a notable mention—we had this at school when you did sports day—was Menlo. To be fair to them, they had amazing returns with Chime and then did Anthropic at $4 billion.

Guest 2

It was on my list too. I've actually talked to Venky Ganesan at Menlo. I think they did a brilliant job. That was a case where one train had left the station, and there was a new train pulling out. You jumped on that carriage and grabbed on firmly. I give them huge credit for that.

Guest 3

Yeah, totally agree.

All righty. If we dig a layer deeper, though, beneath firm investor of the year, I'm asking for 1 individual.

Guest 3

Just to name someone we haven't named, you've got to hand it to Elad Gil. Being good enough to raise a $3 billion solo fund—I mean, you 2 guys have solo funds—and being good enough in terms of your returns to raise that size of fund as a solo GP, you just have to say, “Wow.”

I would also say, going back to the word that we clearly loved in 2025, having a perception of taste and style such that you become a preferred investor—you've got to hand it to the guy.

I agree. I think Elad's fantastic. One thing I didn't realize with Elad, though—I was chatting to him the other day—was that he does have 18 people.

Guest 3

Of course, that's even better, where it's all you, but then you have the people who do the work.

And Harry, who are you to judge about that? How many people do you have, Mr. Stebbings?

None. It's all me. It's just me working away. Yeah, absolutely.

Guest 3

Yeah, sure. You know, built by the sweat of your brow. Don't make me laugh, but yeah. No, I think, again, it points to a part of the market where the importance of brand is easier to build as an individual in the short term.

I think being able to break through the noise as a perceived high-value, high-thoughtful investor, where there is an endorsement on the cap table, speaks to the market we're in now. I think he's done a great job.

I'm going to throw in a name, which is Lee Marie at Kleiner Perkins. Two deals returned a shitload of money, one being Windsurf and the other being Chronosphere: $2.5 billion and $3 billion. Importantly, both were liquid this year and returned a lot of money back to KP. Amazing work.

Guest 3

Good for her, because liquidity is the—I will say, I did contradict myself on that name. I picked Elad because I think, just in terms of the raise, but you're right.

One of the exercises we do, actually in the first week in January—we haven't done it yet—is we just rank the actual bona fide exits for the year. We count nothing else. We call it the Great Exit Database. We've done it for 20 years, and we just rank them by size and then look at who did them.

By the third week in January, I'll know exactly, and that is, in the end, what counts. It's a raw capital-dollar business, and I think if she had those two, that's a great win. I know Greylock had the A in Chronosphere again. It's a great outcome. It's cold, hard cash, and come early 2026, an excellent outcome for all involved.

3. Breakout Companies of 2025: Who Made the Biggest Impact?

All righty, back to the company side. What is the breakout company of 2025? Whose year was made in 2025?

There's lots. First of all, I'm interpreting this, just to be clear, as a year where, at the start of 2025, no one had heard of these things, and then, oh my God, in 2025, they kind of exploded. So it's not OpenAI as a topic; it's the companies that broke out.

I think there's a ton. It's really interesting, just to pick a few. I think OpenEvidence in the AI medical space really nailed it.

What we saw when we looked at a lot of these academic research search engines was that, if you look at people looking up papers, about 60% of that traffic was doctors. Doctors are the biggest single demographic of people who want to look up a recent academic paper and find out a result.

I remember we were looking at some of the broad horizontal science search engines and thinking, "Hmm, that's interesting." OpenEvidence didn't just think it was interesting. Very wisely, it focused on that space and said, "Look, doctors—even ordinary GPs—sometimes have a patient with a special, obscure question. They want to know what the latest research is on XYZ."

This perfectly lends itself to an LLM-based application. They built that product totally suited for their target audience, and they went from nothing to 500,000 doctors out of about 1 million in the US in the space of 1 year. Obviously, the revenues followed.

Why does the revenue follow? Because you have doctors typing into a search engine, "My patient has this obscure disease called blah, and I need help to find out what's going on." Who wants that real estate? The drug company that makes the $40,000 orphan drug that only has 10,000 customers in the US with that disease, and now you've put up your hand and said, "My patient is 1 of them."

It's a perfectly targeted business. I mean, it's a great business. They're expanding and doing a lot of other things now on top of that, but it's just come from nowhere because it hit the need perfectly: the need on the specialist search side, which is the doctor, and then totally the need on the advertiser side, which is the drug company. Perfect product.

My patient is Irish. How do we solve this problem?

Rory O’Driscoll

Well, you put us in prison, conquer our country, or give a fuck for 800 years. That's how you solve it. But it didn't work. As I told you, I'm grumpy today. Sorry. Sorry.

Jason, what do you think?

Guest 3

Okay, I'm going to give you an answer that, at first, you're going to say doesn't qualify, but I think when I fully answer, you'll agree with me. Mine is Databricks, and I will tell you why.

Of course, we talked about Databricks. I went back—we did a SaaStr Annual during the pandemic in 2021. We were the only event in the entire California Bay Area that I know of that did it. It was pretty fun. It was all outdoors, and Ali came. I didn't even know much about Databricks, and he did this deep dive on everything.

I went back and looked at it. Nothing was about AI; it was all about managing cloud compute and data. The job of every founder that's listening to this, and every VC, was to ride this torrent of AI. If you're not an LLM, your job this year was to do more with LLMs. If you're an LLM, your job was to get GPT-5 out so you could change the world.

If you were a B2B company, however you broadly describe it, your job was to ride the greatest wave of our lifetimes. Snowflake—it's all definitional, right?—but Snowflake's AI revenue is pretty small. Its agentic strategy is just starting.

Databricks rode this, and now they're both at $5 billion. Databricks is growing and accelerating at 55% growth. If we had a company of the year for who rode the AI wave and utterly changed the trajectory of the company, it would be Databricks, even though obviously they were successful before. But JFC, for this one.

That gets my vote, because this was all of our jobs, right? This is the job of every founder and every VC, too. If you didn't ride that wave this year as a founder, you get a D-minus.

I think that's a good answer. Mine would be ElevenLabs. When you look at what ElevenLabs has done this year, scaling to $400 million in ARR and fending off competition from OpenAI and any of the other incumbents, it's a crazy good one. It deserves to be above the fold.

Guest 3

There's a 2-by-2 of best-of-breed, which is hard to win at in AI, right? Because it had to be best-of-breed to win, right? There may be cases where this isn't true, but for us, it's utterly insane value.

This is what sometimes people miss. How did ElevenLabs go to $400 million in a year? When we were at SaaStr London a couple of weeks ago, we needed to do all the voice-of-God work for 200 speakers or whatever we had. That used to take 2 weeks, and you'd have to pay someone maybe a couple grand. They wouldn't do some of them, and they'd mispronounce Stebbings. It was terrible.

Amelia literally fired up ElevenLabs the morning of the first day, when you have other stuff to do, and just had ElevenLabs do it all. She trained her voice—I think she's used ElevenLabs before, but she couldn't find it. She retrained her voice in 2 minutes, and it did all of them perfectly.

That was $30 instead of $2,000, and it was done in 10 minutes instead of 2 weeks. That is why these apps are blowing up, because it's not the little step function we got before AI. It's $30 in 10 minutes versus $2,000 in 2 weeks. I don't even care exactly what it costs at that level, right? So, yeah, that one counts.

One of the things I love about a lot of these companies is a go-to-market that empowers individual users to grab the technology and, just as you say, get shit done. You had the boring job of recording 200 names, and now you just type it in and away it goes.

I think that's the case with many of these. Actually, I have a bunch of other breakout products, and they're all the same. At the app level, which is where I think most of us focus, it's really just about getting into the hands of individual users and having it explode.

I'm amazed you guys didn't throw out your favorite. The fact that you're going to give breakout product of the year and not mention Lovable and Replit—I'm kind of disappointed in you guys, right? Come on.

Guest 3

I think it's—listen, for me personally, it was the breakout tool of the year. It was where I spent 200 hours in terms of impacting the world.

I think we're going to see it next year. I don't think either of these prosumer vibe-coding apps had a material impact on the world. It's just because it's early. No one built a game-changing B2B app with 9 figures of revenue on Lovable or Replit.

I do believe it will come. I was beyond skeptical of vibe coding when I started the journey. Now it's clear it will. But if we're talking about human impact on the world, those guys have just started. They're just at the very beginning of the journey.

So I'm not giving them that. They're not going to make it above the fold on my list for that.

4. Biggest Surprises of 2025

I found this next one a hard one. What was the biggest surprise of 2025? What did you look at in the news and go, "Oh my gosh, I can't believe that"?

Guest 3

That's easy. I didn't expect the talent wars. The fact that everyone would tear up everything to get talent—in retrospect, it's obvious, and I'll talk about that in a second, but wow.

Meta is willing to just give a dude $100 million to show up. Meta is willing to buy a company for $14 billion and say, "Give me the people. You guys keep the company; see if I care. You can give that $14 billion back to yourselves, and then let the company be a husk."

Rory O’Driscoll

See if I care again, right? Nat and Daniel have a venture fund. No, we should just buy that too, right? They were the biggest, but they weren’t the only people doing this. I mean, you had the whole Google-Windsurf saga, right? The entire convention around why people buy companies, how older employees get treated, and how much a human can be paid to do a job got thrown out of the window in the space of 6 months. It blew me away, right?

As I say, in retrospect, the logic that someone has unveiled is pretty compelling, which is: if you’re spending $73 billion on capex, spending $5 billion to make sure that the people using the capex know what they’re doing probably makes sense. So you look back and go, it shouldn’t have been surprising, but at the time you’re like, wow. I mean, remember, it was incredulity about the money and also incredulity about the damage to the, quote-unquote, social conventions. That, to me, I wasn’t ready for this year.

Guest 3

One is just that there are no limits to the exit value of tech startups. It just changes the whole industry. We’ve talked about it; it has changed so much since we started this podcast. But the idea that we will have a trillion-dollar IPO next year—maybe 2—we might have 2 trillion-dollar IPOs. Getting in at Anthropic at $100 million or $80 million, or whatever the deal was, was the deal of the century from a risk-reward outcome. The fact that, who knows where the future goes, so many of these companies—Lovable could go from $2 billion to $8 billion or $6 billion in weeks, and so can Harvey. These weren’t one-off step-ups in valuation. That’s the biggest one.

It was unexpected. For the moment, there’s no ceiling on venture, which changes all the math and calculations. When I go back—I was thinking about the earlier days when I started investing—I remember Byron Deeter came up to the old SaaStr office. He ran the cloud practice at Bessemer, and he said, “I’m here to mark up your investments, Jason. I’m here to mark up your investments. I want you to give me your top 1 or 2, and I will mark them all up.” I thought that was the greatest deal in the world. I was going to invest at $10 million or $12 million, and Byron was going to invest in them at $50 million, and I would be a genius because I got in first.

I thought that was all the game was in venture, because I would walk in at the end of the year and have a 3x and a great IRR, and Byron was the sucker because he had to enter at $50 million and $70 million. Maybe that was true at a brief moment in time, but not really. Right now, the sucker is whoever is wasting all the energy to have smaller ownership anywhere on the journey when you could have put in either 9 figures or significant ownership later. It just changes the whole calculus.

And literally, what is it? Fuse Energy in Europe. That one just did it at $5 billion. Here’s the classic one: you know who that was led by—[unclear: “lowercase”]. So maybe the greatest seed investor of several generations ago, Chris Saka, right? He put a lot of money into Twitter and others, but he sniped deals at $3 million, $4 million, $5 million pre, and is now leading deals at $5 billion. Those are different times, aren’t they? So that I didn’t foresee.

The other one is just, as a vibe coder, it’s autonomy plus reasoning. I started this year as an AI want-to-be believer but skeptic because I couldn’t get anything to work at the start of the year. I launched 1 agent that was a digital version of me that people loved, but all it did was parrot me back with AI. People loved it; it was used 100,000 times in about 30 days. But JFC, what agents can do now—we haven’t missed the boat as investors, because what agents can do, we just started. Outside of parts of coding, we just started. Those are my 2.

Mine’s super simple. The 3 days when Windsurf was being bought by Cognition and I interviewed Varun the day before. I remember thinking, what the hell is happening? Are they being bought? Are they not being bought? OpenAI were buying them, and then they’re not, and then what’s left? That was a bizarre moment.

And then NVIDIA investing $100 billion into OpenAI and the start of the circular deals was the most bizarre.

Rory O’Driscoll

Overall, just not all of us giving up on caring about things like circular deals.

Guest 3

We’ve all stopped caring. We don’t care. If it drives the stock price up, great. If it marks up my fund, great.

It’s just the way it is now.

Rory O’Driscoll

You won’t care until you do, and you’ll look back and go, “The first ones were sensible.” It’s like everything is fine. It kind of goes back to the other thing you mentioned, Jason, which is the surprise of the extent of the upside—how big these things can be, right?

I agree with you. I think that is probably the biggest change in venture, zooming out over 20 years. Both the marks and the markets themselves became bigger, and the end results—the outcomes—became bigger. On top of that, because they were held for longer as private companies, more of that bigness went to the venture sector versus the public markets. That is the overwhelming economic story.

Technology as a percentage of GDP shot up, and the percentage of that technology that the venture business grabbed shot up. The converse comment—I’ll talk about this later—is that we still don’t know whether the calibration is quite correct. Until we see some of these $500 billion-plus deals get done in the public markets and find that equilibrium price point in a very different, more liquid market, we won’t know whether all these marks are correct.

5. Predictions for 2026: Top Performing Tech Stocks of the Year

Typically, what happens with exciting financial innovation is that it works amazingly and it overshoots. One of the interesting things, on both some of these valuations and frankly on some of these round-trip deals, will be: have you pushed a good thing too far, and does it blow up in your face? We’ll talk about that later.

Well, we’re going to move to predictions for 2026 then, Rory. One of my predictions is that you’re going to continue to think that every question I ask is shit, and all I do is hope to one day meet your bar of question-asking. But that will be our year in 2026, Rory. I promise. Okay, I’m going to go on.

Rory O’Driscoll

You’re going to start as you mean to go on, then, because I’m not a fan of the 2026 questions. So go on.

Oh, good. Well, there we go. Jason, he’s in a really good mood. Now, I hated 2025 questions, and 2026 are worse. You said you didn’t want to be negative, so I’m starting on a positive. Number 1: what’s going to be the best-performing tech stock in 2026?

Guest 3

How the hell—I mean, the best? Okay, let me ask. I’m going to pick on you. What was the best-performing tech stock in 2025?

If you don’t know the past, how the freak are you going to predict the future?

Guest 3

Do you know the past?

I do, because I checked this morning.

Rory O’Driscoll

So literally, if you rank every stock in the US and define “tech” vaguely, and don’t put in a market-cap filter—you know, a very small one, cutting out anything below $200 million—Planet Labs, Bloom Energy, Opendoor, Oklo, and Seagate are the top 5-performing stocks this year.

That’s why your funny little non-recognition face, Harry, shows the thing: what you were really asking in that is, of the big caps, obviously. Of the big caps, NVIDIA and Google nailed it this year, and the others were roughly flat. So that’s a more boring answer, but it’s always—if you’re going to ask what the best-performing stock is, that’s why it’s such a shit question.

The best-performing tech stock typically is a smaller stock. All of those stocks are more than a 3x year on year. I know some of them. I mean, look, Planet Labs—you’ve got to love that company, especially as space starts to take off. They do satellites in space for imaging. It’s kind of popped to about $5 billion this year.

Bloom Energy, I know, but not as well. Opendoor, we had them on. It’s been TBD on how it goes, but the guy can make the stock jump. And then Oklo, you’ve got fission. And then Seagate, which is the 5th-best-performing stock.

So you look at those 5 names: utterly idiosyncratic, fairly unthematic, right? You’re not going to get there with some kind of top-down analysis. So if you can barely nail the past, the best-performing stock in 2026 is a bridge too far for me. So no, I don’t like your question, Harry.

I’ll tell you what I think. I will bet that the top 6 public B2B stocks—at least 3 of them—are in the top 6 next year. If we go back to 2024, 2 of the top-performing were AppLovin and Palantir. AppLovin was up 700% in 2024. We all thought it was crazy, right? And then this year, it’s up 124%. Still—actually better than my fund. My fund didn’t go up 124% this year.

6. B2B Stocks to Watch

Palantir was the number 1 performer of this public B2B group in 2024, up 340%. My point is, I’m not an expert. Maybe I’ve got the math wrong. It’s possible I have the math wrong. But my point is, I don’t think the macro is going to change much next year. I’m going to bet—not being a public-company guy, not being a hedge-fund guy, not being Steve Cohen—that the trends aren’t going to change much next year, like the mobile-ad trends fueling AppLovin.

Guest 3

So, I'm going to bet on Palantir, Cloudflare, [bleeped company], Shopify, CrowdStrike, and Snowflake. Those are the big boys—not the hyperscalers. Those are the big boys in B2B. There are going to be some shifts. Shopify is getting some interesting headwinds, but it's slow to AI. MongoDB had a really rough patch this year, right? And then blew up in the back half of this year.

So I think this top six is going to open down. But let's call them the PCMSCS—what did we used to call them? FANGs, and we stopped caring, right? My PCMSCS: Palantir, Cloudflare, [bleeped company], Shopify, CrowdStrike, and Snowflake. I'll bet you $10,000 that 3 of them are in the top 6 of this cohort at the end of next year.

Jason, which 3 would you put in there?

Guest 3

Well, listen, I'll answer it for fun. If you think about what's interesting about these top 6—Palantir, Cloudflare, MongoDB, Shopify, CrowdStrike, and Snowflake—they're all leaders in slightly different elements of this AI phase, right?

Palantir's got AI for defense and government. Cloudflare owns how we deploy the internet. [bleeped database company] looked like it was going to be a database loser in AI, right, with Supabase, and everyone came roaring back. Shopify is e-commerce, which is different.

CrowdStrike is security, right, and Snowflake is data and data lakes. So now you're making me bet on a macro trend. How could I bet against any of them? The reason they're in the top 6, I think, is that they're all leaders. I'm not going to bet against e-commerce. E-commerce had a record year. We don't even talk about it much on this show. E-commerce is on fire.

You could do an exclusion. You could go, actually, Snowflake is going to be continuously cannibalized by Databricks.

Guest 3

But Snowflake has reaccelerated. Here's the thing: Snowflake is not going to catch Databricks, I don't think. But it reaccelerated in the back half of this year. Even Snowflake reaccelerated. [bleeped database company] reaccelerated.

This reacceleration, and also this M&A thing we talked about on the prior show, is super important going into the new year. I think the folks that are reaccelerating at the end of this year, that have figured out their AI tailwinds, I'm sure not going to bet against them in 2026, because I think these tailwinds are going to accelerate for at least 2, 3, 4 quarters.

7. Why Salesforce Could Be the Buy of 2026

In fact, I'll go the other way. The ones that underperformed this year—we go from the bottom up—I think Salesforce might be the biggest beneficiary next year, and I'll tell you why. Half of their customer base wants to buy an agentic product that works tomorrow. They may not want to buy another cloud. They may not want to buy your Data Cloud. They may not want to buy Informatica. Okay, I don't know if they want to buy Informatica, but when their rep calls them up and says, “Listen, Agentforce is crushing it now. We can get rid of 200 people in your sales team, 100 of your marketers, and 100 in field support with AI, and it works,” you're going to see attach rates like we've seen in coding. Everybody's going to want to buy that Agentforce.

So if these are bets, now we've got to squint at the bottom. Who's doing that, or who's quietly giving up? Maybe Asana is. I don't know.

I think those are such different bets. The first list is your momentum bet. They're all killing it in terms of momentum, and what you're saying is that the momentum continues. That would be true until it isn't, because what you're not taking—

Guest 3

I think the AI is going to—I don't mean to interrupt, but I think it's not just momentum. It's this AI.

That's the AI momentum. What you're basically saying is, if I look at factor investing—if I talk about the first list—you're saying all these are going to do relatively well up until the moment that the AI lift starts to recede, or even the AI trade starts to decline a little bit, and then they'll all wind down because they're way over any kind of meaningful long-term valuation multiple.

As long as it's all working, Palantir will keep compounding. At some point, someone's going to look up and say, “Oh my God, 70x revenue. I should panic.” It's pennies in front of steamrollers. The Salesforce bet was a super-interesting one, and you could apply it to a number of companies because it's totally different.

Guest 2

They're trading at all-time lows on a revenue multiple—5, 5.5 times revenues. I totally agree, and it's why we're talking about it at the start. All they have to do to lift that puppy 20–30% is monetize this recent acquisition they did, co-attach 20% of their customers to the product, grind out a year of getting that co-attach up, and suddenly you buy that stock.

You're less likely to get a 3x, but you're more likely to get a 20% lift, and you're not taking the valuation risk. Those are very different bets.

Guest 3

Yeah, in a world where the juice goes out of the market and the high-flyers collapse, at 5.5 times run-rate revenues with 30% cash flows, Salesforce could go down 20%, but it's not going to go down 60%, right? So those are just very different bets.

I mean, it's fun too, because you're right: one of the disappointing investments in the public markets this year has been traditional SaaS. I own a lot of Salesforce; it's down 30% on the year. HubSpot, where we were lucky enough to be early investors—the stock's down on the year, across the board. All of those guys are now at rates and valuations where all you have to do is get a bit of a lift and you can outperform. It's almost the antithesis.

That's the question. So I look at the bottom 6. The bottom 6 I have are GitLab, Atlassian, Adobe, monday.com, BILL, and HubSpot. You kind of have to look at those 6 and say: are they talking AI, or have they proven it? Are the tailwinds real?

I worry about most of those 6. They talk a lot on social media and Twitter, but it's not enough to have an AI product. You have to have the co-attach that Rory talks about. You have to have something where someone's going to go in and pay as much or more for your core product for this agent, right?

I think people will pay $100,000 a year for a Salesforce GTM agent. Whether Atlassian can pull this off, or Adobe, or folks I love—monday.com, BILL, and HubSpot—I don't think they've proven it yet. So I don't know. I'm not being critical, but they haven't seen an AI lift, right? An attach lift. That's the one I worry about for next year.

Guest 3

I think in different ways they can. I was on the board of BILL for many years. I think René is a really talented entrepreneur.

As good as they get.

Guest 3

I think there's a lot they can do, and they'll be doing it. I think it's—

They've done some of it. They've started to do it on the—I think, actually, interestingly enough, yes, they've done it on invoice recognition and kind of the back office.

I also think there's a lot you can do, frankly, on broadly financial management and the kind of stuff that Ramp does, where they're coming at it from less on the card side, more on the ACH side. So I think there's a ton they can do there that's AI-first, but not just AI; there's other uplift they have.

Guest 3

I've been a BILL customer since day 1, but the question is: can they dramatically lift ACV and pricing for these deals? That's where I worry about BILL. As a happy, 100-NPS BILL customer and superfan of René since we started together as founders, I don't know.

But I'm worried about some of the ones that didn't grow this year. It's not that they won't be amazing AI products next year. It's that they can't charge more for them, right? And that's why Notion, to me, was not on this list. Notion was the biggest surprise of this upper group for me, because I didn't think they could charge this much for their agent.

I sort of get it now, but I didn't think people would pay twice as much for Notion to talk to their Notion. But I get it now.

I'd love to know. One of the things that I talked about with another big executive of a large, multibillion-dollar-revenue SaaS company—and I'd love to know the data from Salesforce, because it wasn't a Salesforce executive, to be clear—is that you're seeing this AI revenue, but what's happening is it's being bundled.

You're attributing value to the AI product, but separating out what is your core product and what is your AI product is tricky, right? Some folks are getting a little bit of revenue lift that looks like it's coming from assigning dollars to AI, but really it's the core product.

And then the scary thing about that, just to be clear, is if you break apart your product and price the AI product separately, and then a year from now the customer says, “Oh, I'm not going to keep that AI product, but I still love your core product,” but you've gone down 10% in pricing to add 20% uplift from the AI product, you could have a nasty renewal cycle.

So, you've got to not just sell the product; you've got to deliver the value. I think all these things are going to come down, in the end, to delivering the value. I think the opportunity is there for most of these companies, or at least the ones that I know well.

I was going to say, on the to-do and personal-planning side, the thesis is not as clear, but actually, as I think about it, there is a ton you can do in terms of managing personal productivity with AI. So, the opportunity is there for all of them. I think that, as we said last week, the difference will be the executives who can see a way to getting it done.

Guest 3

I mean, to me, the worst one, Harry—I know you want to stay on track, but I'll just have 2 comments—is Adobe. Adobe announced they had $5 billion of “AI-influenced revenue” this year. Listen, let's be kind: I'm sure it's true. I'm sure they're using Firefly and generating images with AI, but to me, it doesn't count unless it's net-new bookings.

If you're going to hide in your tower or at your end-of-year planning meeting and say, “Hey, we grew 8% this year,” that's the bad news at scale. The good news is half of our revenue was AI-influenced. I mean, that's table stakes, right? So, to me, Adobe gets the worst score.

To me, if it's private—and I think I have the numbers right—Notion hit 50% growth at $600 million this year. I think the only thing that makes any sense to me for the reacceleration is that they got people to pay $20 a month per seat for their AI offering instead of $10.

One of my portfolio companies shared a Notion board update with me the other day, and it didn't have the AI on it. I couldn't talk to it. I'm like, “This sucks. I can't use Notion without AI.” As a 10-year customer, whatever I am, it is worth the 2x. That's the one that so many B2B companies want to do. They want to double their pricing for AI, but Notion might be one of the very few that earned it.

It's very hard for HubSpot or Atlassian or others to just double their pricing to use their AI. We just expect it to be free, like Adobe. We expect it, but somehow Notion pulled it off. I might be wrong, but I just don't see any other way they did this. I don't see any other way they did this other than getting people to go from the $10-a-month plan to the $20-a-month plan. It's the only way I could see the math pencil out for AI.

Yeah. You've got to be hard-nosed about 2 things. One, delivering value. You can't bullshit yourself that you're not delivering value. Then you've got to be hard-nosed with the customer, making them pay for it.

You're exactly right, Jason. I'm thinking about this now in the context of another one of my companies—a mid-stage, $100 million-plus-or-minus company unveiling AI products. It's very easy to slip into, “Oh, this is good. We'll give it to you in the base package,” and that's just not going to move the needle.

You're exactly right: you've got to say, “No, this is really good. Here are the different quantum of value we're delivering to you, and as a result of that, you've got to pay more.” That dynamic—which I'm thinking of in relation to a specific private company I'm working with at the moment—is the dynamic that everyone from Salesforce on down has to be dealing with. Not just AI-influenced bookings, but actually, “Here's the extra.”

I totally agree, because if all that happens is you take your same revenue line, bundle in a whole bunch of AI, and don't expand your revenue, congratulations: you've reduced your operating margins by 10%. My suspicion is Adobe is one of the worst offenders there.

It feels like it should be doable. I think Notion is a great example of doing it. But if someone like Salesforce or ServiceNow can figure that out at scale, you just get such operating leverage. It's wildly hard, but I agree with you. Almost every Salesforce customer is saying, “Oh, please, God, integrate this bullshit so I don't have to deal with it.” And damn you, you haven't.

Guest 3

Yeah. I think what's going to happen—for here's the thing that I know is true, right? 2025 was the year of coding, but it was also the year of support. Support went from not working to working once Claude 3.5 and 3.7 came out. Just like vibe coding worked, so did AI support. It just didn't work before; it wasn't good enough. It was the same issue. It got so good, and that benefited a few leaders a little bit.

When it works, Salesforce is like a 7-cloud company. But when it works in 2026 for GTM, every Salesforce customer is going to want it. Marc came on this pod with us earlier this year, right? He talked about how they had, what, a billion—maybe 100 million—leads at Salesforce. I didn't talk about the only problem, which is that Agentforce takes time to deploy and is just getting going.

But when they can turn that on, don't you think, as we said on the pod, every single Salesforce customer is going to want that product when it can be delivered quickly? “Would you like this? This is another $100,000 a year, but our agent will automatically go after all the customers your team didn't follow up with.” Sign me the fuck up. That is a bet I will make for next year.

The flip side is, if you ask me the biggest failing of this year—and I want to stay in 2025, going to Rory's point—it was Copilot. Copilot was the worst of all worlds. “We're not going to put it in our base product. We're not going to make the core product better, but we're not going to provide so much value, like Notion, that you want to pay for it.”

That was the cynical whiteboard of January 2025 that failed up and down the B2B world: the expensive Copilot.

I think that's a great point. I mean, starting, let's be honest, with Microsoft—

Guest 3

Yeah, there wasn't enough value.

It wasn't enough value. I think that's going to be the biggest risk. It's interesting: we're kind of dissing on Adobe, we're kind of dissing on Microsoft, and we're dissing on Salesforce. I think what they all have in common is they're just so big.

The bigger you are, the easier it is for you, as the supreme leader, to just get told that it's good 3, 4, or 5 levels down. Unless you're hands-on with the product and really brutally honest, you're just going to fall for that. Maybe it'll happen, and you're going to end up not delivering value. I think being really on top of that is key.

8. Google, Meta, Apple, NVIDIA, Microsoft: Buy One, Short One

Google, Meta, Amazon, Microsoft, Apple, NVIDIA: buy 1, short 1 for 2026.

Guest 2

Google, NVIDIA.

Just to be clear: you buy Google, you'd short NVIDIA.

Guest 2

Yeah, because at some point, if the capex cycle continues and we're all still spending, NVIDIA does fine. But at this point, do they need continued acceleration of the capex cycle? Google has upside if the capex tide goes out. NVIDIA is going to be very heavily priced, and Google still has its business and can gradually roll out AI in a more measured fashion.

Oddly enough, Microsoft—and even more so Apple, and definitely Amazon—are kind of meh in the middle on the AI cycle. They're not getting any lift from it. Oddly enough, even Microsoft, obviously with their OpenAI investment, have wisely, in my view, pulled back from doing some of the marginal data-center stuff. So, they're meh in the middle.

Apple's meh in the middle to the downside. Amazon's meh in the middle to the downside. So, you're kind of taking a—if you want to play the “How do you think about AI capex in the next couple of years?”—those are the 2 that are interesting.

Surely Apple would be the big AI failure. I mean, Apple has failed across all elements in terms of integrating AI.

Guest 2

It has, and as yet the stock's held up. Look, I own a lot, and I sold some this year, right? To me, that's a different one. To me, the question there is: how long can you continue to trade at a premium multiple when your growth rate is less than 10%, albeit your EPS is growing more?

You're trading at an all-time-high revenue multiple or P/E multiple, and there's no catalyst. But I don't think it's an AI—I don't think they'll sell a whole ton more if they had some great AI. I think they'll do marginally better. So, yeah, I think there's risk in that stock.

If I was just answering on a standalone basis, looking objectively at what I did, I sold some of my Apple and I didn't sell any of my Google.

That's Rory speaking with his facts.

Guest 3

You're a fool to invest against NVIDIA for 2026. I think you are arguably a fool to invest in Amazon in 2026 for a simple reason—and I didn't say 2027 or 2028. It is not enough time for all this competition to come online.

It is not enough time for everyone. Everyone's trying to do a TPU now, right? Their own version. Amazon's got theirs. Google and OpenAI have their own teams. Everyone's trying to get out from under the massive NVIDIA tax they think they're paying, as Rory keeps pointing out.

It's just natural to design out your most expensive component. It's just natural. It ain't going to happen in 2026. There is not enough time to produce enough GPUs and TPUs to get them into data centers that don't exist. I'm not a public-company expert, but I don't believe the public companies look as far into the future and as thoughtfully as we think.

Guest 2

So I don't think we're going to have any down quarters unless AI stumbles. And I also don't think all the great things Amazon is going to do—Amazon stock is only up 2% this year—and I think they'll be the underperformer next year for the same reason: there's just not enough time. But 2027, I don't know the pace of it.

You just can't make semiconductors from scratch that compete with NVIDIA in weeks. This is not software; you can't launch Base44. There's no Base44 of GPUs. It just doesn't exist.

What do you think the biggest risk is?

Guest 2

I think I was very thoughtful about it. I didn't like the question when Harry asked it. Look, it's always the truth about shorting. That's why I don't do it, right? You can be right conceptually and just wrong in timing, and then you're definitionally wrong. Whereas if you're long, you just hold, and in the end you can be right.

It's why shorting is such an incredibly hard way to make money. Buying puts, which is basically the coward's way of shorting—and which is what I would do—is an even shittier way to make money because you also pay more for the privilege. You bet right?

So again, Harry, back to your comment: I hate your questions, but to get a useful one out of it, Jason, what do you think is the bigger risk to NVIDIA? Is it continued massive demand for compute, but other people introduce competing TPU products—or GPU products, in the case of AMD—you know, market-share erosion? Or is it erosion of demand, such that the continued explosion in demand tapers off because the investments overall get cut back? Which of the 2 is the biggest risk?

Guest 3

Listen, I believe there's no risk other than some existential power risk, because the only real risk here that I can see—I'm only so smart—is that OpenAI can't give them $100 billion to buy their chips. But the good news is NVIDIA is going to give them the $100 billion. So that risk is off the table, right?

OpenAI has taken the position that growth in compute, growth in GPUs, growth in just power, is 1:1 correlated with revenue growth. They put this out. It's 1:1. So they're like, “Yes, I need $100 billion, but I can prove to you this is—whether it's correlation or causation will be the debate—I can prove to you there's a 1:1 correlation that historically, if you give me $100 billion to buy compute, I will generate $100 billion in revenue.” Right?

The market is going to solve for that problem in 2026. So I think everyone should feel pretty good about their 401(k)s.

Guest 2

I think you could be right. I mean, look, the breaking news in the last 24 hours was another $100 billion for OpenAI, which obviously—I mean, Tomas had mentioned keeping an eye on the Oracle CDS swaps. My guess is a lot of the risk goes out of that deal if OpenAI gets another $100 billion. And you're right: as long as the protagonists who believe it should keep going keep getting money, it will keep going, right?

And you're right, OpenAI is the prime protagonist, the prime believer in more capex equals more intelligence. If they got another $100 billion, you're right, bets are on for another 12 months, which would take you back to Harry's comment. I don't think you try and make a clever market-timing position yet. The tell's not there.

Guest 3

Yeah. And we can make fun of these circular deals, and they should be made fun of. But NVIDIA is going to take its almost $100 billion of free cash flow a year and reinvest it aggressively, thoughtfully, strategically. It wants other partners. It can't do it all on its own, but it is going to lubricate all of this business model in 2026 as much as it can.

9. IPO Speculations: Who Will Go Public in 2026

It is aggressively reinvesting that cash. Leaving that cash on the balance sheet does it no benefit.

Alrighty. So Jason, I can assure you one thing with my next question: Rory's going to hate it. He's going to hate it, so I assure you—

Guest 2

He's going to like this one. He's going to like this one. That's my bet.

Which companies of the mega companies speculated will go public in 2026?

Guest 3

This is an interesting—okay, I don't hate this question because [laughter]—

See? I win.

Guest 3

You know, I don't hate it because it's a legitimate question and an area we think about. I mean, you have the 2 AI companies, OpenAI and Anthropic. You have, just taking the huge ones, Databricks, Stripe, and then the big one of all, SpaceX. I think Stripe is least likely; take it off the table. My gut would be Anthropic gets it done. It's a more manageable, bite-size opportunity, right?

I think one interesting question on both SpaceX and OpenAI—and actually all these in general—we kind of hit on it last week, and then I just saw a really good The Information article—is the bankers are finally getting to grips with the problem of how do you take a company public at plus or minus $1 trillion? You know, raise $50 billion, 5%, and what do you do with the other $950 billion in terms of lockup, in terms of generating demand?

And I think they're beginning to internalize: look, in a hyped market, all 3 can get done at the high-water marks or beyond, but that's a lot of demand that you have to create, and there's a lot of shares to follow it, right?

It's not like some of these other high-priced story stocks—the story stocks that are worth, yeah, 70 times revenues or 20 times revenues: the Palantirs, the Teslas. The way that happened is retail bought in early, made out like bandits, and they have real believers at the table who are in the money, right? They've got a lot of people saying, “I'll never sell my Tesla.”

It's going to be tricky to take a new company out starting at $1 trillion. No public investors made money on it. A lot of the institutional money is already in it. And institutions are ruthless. When they hit their price target, they'll be sellers.

So they're beginning to wrestle, I think quietly, with the problem of: what do I do when I take my company public at $1 trillion? I raise $50 billion. There's a $50 billion float. It bounces around like crazy for 6 months, and then $950 billion of stock is available to trade, many of whom have been in the stock for 10 or 15 years. Hmm. Could be tricky.

Rory O’Driscoll

Probably 1 of them gets done because there's going to be a lot of brainy people going at it. But it's an interesting banking problem that we've never seen before. I mean, it's one thing to take Aramco public, where it's priced not at an excessive price; it's priced at a boring P/E multiple.

When you're taking—if you really are taking something public at 30–40 times run-rate revenues [snorts], you have the growth rate, in the case of OpenAI, to justify it, but you have the burn. You don't quite have the growth rate in the case of SpaceX, but you do have the profits and a unique market position. Neither is quote-unquote crazy, but you're going to have to generate a huge amount of demand for a highly priced, risky stock that hasn't traded.

So I don't know if all 3 get done. I would be surprised. I don't know which. I think it'll be idiosyncratic. And if OpenAI gets to $100 billion in the next few weeks, they can exhale and take a whole year off from capital raising.

Come on, Rory. Put a bet down.

Rory O’Driscoll

I'm going to put a bet down that Anthropic will go out and SpaceX will go out, but Databricks won't, OpenAI won't, and Stripe won't.

That's probably where I'd go too. Though the most cogent argument against going public—I was rereading the Ashley Vance biography of Elon Musk from 2015. It's a really good read now because it's 10 years on, and he actually calls some things very interestingly.

But there's an appendix to it where Elon does this really good post about why we're not going to go public at SpaceX right now, why being public is a pain in the ass, and why it's just really hard. It was a really good post, and I read it and thought, you know, one of the smartest first-principles thinkers on the planet—God, he really had a good argument against being public. So if he's going out, it must be purely because of cost of capital.

Rory O’Driscoll

June next year, SpaceX goes out. Money on.

Okay. Do you think it goes out at $1.5 trillion? I'm not going to bet.

Rory O’Driscoll

No, I don't think it does at all. I think it goes out at $1.12 trillion.

We'll see.

Guest 3

The question is: who's going to go out? I think we see 4 IPOs backloaded next year. I think this is the order.

SpaceX goes out first, probably as early as makes sense with the summer dynamics. Canva comes out of nowhere second because it's not that AI-first. So now it's time. The AI story is not perfect, but the numbers are there. It's got the numbers. Do it now. Otherwise, you risk looking obsolete. Canva's number 2.

Databricks does it in the back half of the year because it's just time. It's just another financing; it's the Series M. Databricks is just the Series M. They do it.

Anthropic does it at the end of the year to go public because it's the simpler way to solve their capital needs and because there's no downside other than the headaches of being public. OpenAI probably should have gone first, but it's burning too much. So this is sometime mid-2027. That one's just below the line.

But I bet SpaceX, Canva, Databricks, and Anthropic—all next year, in that order, more or less.

Rory O’Driscoll

It’s roughly consistent with what they’ve said publicly—the limits that they’ve said. I’ve proven with Harry that I’m always too optimistic by a quarter or 2, or sometimes 2 years. But I think this makes sense.

10. The Impact of AI on Employment

I love it. Is there one more question, boys, that you want to choose for 2026 predictions? Or do you want to annihilate me again, Rory, on my choice?

Rory O’Driscoll

No.

[Speaker?]

I’ve got one.

Yeah.

By the end of 2026, does real AI-driven unemployment show up in the numbers? The federal numbers, the real numbers tracked—not just by what tech people are shouting on X. Can we really track an impact on unemployment by the end of next year? Because this will be a massive change to society.

Rory O’Driscoll

I think there are 2 separate questions in that, because you made a statement.

[Speaker?]

Reduction in employment—help me. I got the wrong version.

You made a statement that—let me tell you, I disagree. This is important. You made a statement that if it can actually be tracked, then more heat will come on the issue. I disagree.

I think that the mere fact that unemployment ticked up—first comment, separate comment—all the AI executives talk about how there will be an impact on unemployment. It doesn’t matter if it’s true or not. It’s interesting from an economics perspective, and I’d love to have a discussion, but I think politics doesn’t work that way.

Politics isn’t sitting there. Humans aren’t sitting there going, “Well, let me see the long-term correlations and look at the BLS data by subgroup to figure out, is this…” All they’ll know is 2 things: unemployment’s ticked up; it ticked up this month. B: the guys who are building AI are saying that they’re causing it. They’re admitting the crime. They’re like, “We did it. We put you all out of a job.”

So I predict that if unemployment goes up for any reason—tariffs, random bank events, whatever, a big company goes bust—the drumbeat against AI, because you’re the guy, will go up because they’ve kind of confessed to the crime. They’re like, “We did it. We put you all out of a job.” So, almost regardless of whether it’s true or not, unemployment going up will be a backlash event for AI.

[Speaker?]

For sure. I didn’t see that clearly until I started to speak. But now that I think about it, yeah, you’ve put yourself out there. You said this is going to change everything, and you said, “Oh, government, you should deal with this.”

If unemployment ticks up—I mean, remember how scared we all were in ’09 when unemployment was ratcheting up, or 2020 when COVID unemployment was ratcheting up. If unemployment ratchets up even 2 or 3 points, regardless of the reason—it could just be the business cycle—I think you’ll see a techlash that makes what we’re dealing with now an understatement.

I’m with you. I would just add to it. I think your point that they’ve already admitted to the crime is very important, right? I also think there’s this: when the CEO of Walmart is saying every job will change, that’s also signaling there’s a lot of things going on. When the CEO of Walmart says that, there’s a lot to unpack.

And so I agree with everything you said, but I think there’s also an intellectual argument. We’re also waiting to see—I want to see the numbers. I want to see if this is really true. It doesn’t really matter if Cursor got to $27 billion with 300 employees; it doesn’t really matter. It’s not going to impact McDonald’s and the service economy.

I think when we can really say, “Oh, my God, we lost 500,000 jobs last month, in November 2026, due to AI,” I do think it will change our society even more. Even though they’ve admitted to the crime, I think this will accelerate it.

When there are hard numbers at the top of The New York Times and The Wall Street Journal—assuming we still read anything at the end of next year—I think it will become every dinner-table conversation, for real. Every single dinner conversation.

Interesting. Society will be terrified of AI. I think the good news would be, as investors, at least the deals were working, because my fear is that—

We’ll all be on Elysium together. The 3 of us will be on Elysium. Perry will have the biggest house on Elysium with his funds. Rory will be living good. I’ll be a couple of smaller houses away from the park, but we’ll be up in space.

I remember someone succinctly describing their objective as follows. They said to me, “If the robots are going to take over society, I want to be sure that I own the robots.” Right. It’s a good point.

Elon even said that literally.

Oh, did he?

Yeah. He said, “I wish I could slow down robots and AI, but because I can’t, I’m all in.”

Rory O’Driscoll

Yeah. I actually don’t know what the impact will be. We could end up in the weird quadrant where it’s not as impactful in the short term as we think, but because we’ve confessed to the crime, we’re going to take the heat when it becomes generalized, because every CEO, as you pointed out earlier, can blame it. So I think it could be interesting politics there.

Okay. But we don’t do politics. Okay, boys. Happy Christmas. What a joy. I think we put a pin in it there before Rory tells me off. This has been fantastic.