20VC：每日交易：CoreWeave IPO｜Scale以年末20亿美元收入达到250亿美元估值｜Sequoia在Wiz上获得25倍回报｜关税冲击下科技股暴跌｜Cursor：护城河可守，还是科技业护城河消失的危险案例

Jason, I am so excited for this. The first daily deal that we've done. Thank you so much for agreeing to do it with me. I remember cold-DMing you when I was at university, so it was a long time ago.

I want to start, though, with what we're looking at today. Tech stocks are getting hammered in late trading in response to Trump's administration imposing tariffs of between 10% and 49% on imported goods. Apple shares are falling more than 6%. Can you help me understand how we should be looking at and analyzing this today, given the really sizable drops?

I think you've got to go a little bit long, and you just have to ignore it, as VCs do, and push on. Check in six months. That's the best answer I have: check your stocks in 6 months.

Did you see Thoma Bravo's analysis that said we're going to see tax spend go from 2% to 4% of GDP? I'm intrigued to hear what you thought of that.

1. The Golden Age Of Software

I think we are now entering the golden age of software and technology. It's the golden age. This is the golden age. I think we delusionally thought 2021 was the golden age.

Harry, let's look back. 2021 was small compared with today, but it was also crazy. The average public B2B stock was growing 70% in 2021—27 public companies. I'm not talking about Cursor. I'm talking about moldy oldies. The moldy oldies of SaaS were growing 70% in 2021.

We'll never see that again. That was a pandemic bubble where we all thought we were geniuses, but the revenue was real. No question, the revenue was real in 2021. The oldest companies in the world were reborn. GoToMeeting was on fire. Webex was on fire. We hadn't heard from these guys since the '60s, and they were on fire in 2021.

Now, this Thoma Bravo thing is what we're feeling. I don't know. I'm not sure that every knowledge worker is going to be replaced with software, but I do believe that if this trend of software going from 2% to 4% of GDP continues, that is going to birth 100 decacorns—100 or more.

It is the golden age of software. There's just going to be a lot of stress in the golden age. A lot. It's not all a freebie.

What do you mean when you say “a lot of stress,” and where does that show itself most?

2. Revenue Durability Breaks

Look, I'm not the smartest guy with a crystal ball. But what we're seeing today—and you tweet a lot about it, or X a lot about it. Is it X or tweet? I don't know what it is anymore.

I still say tweet.

Do you still say tweet?

Still say tweet.

Tweet.

There is—and we were briefly chatting about this before we started—what I think is really the stress today. The stress will change next year and the year after. The stress today is that a lot of revenue is not as durable as we thought.

You tweeted about how every VC has 5 or 10 unicorns that aren't unicorns anymore, that are growing in the single digits or the teens now. I heard that when we were in London last year for SaaS Europe. A bunch of the VCs in the green room were talking about the same thing, but it's everywhere.

We could talk about the root cause. Some of them didn't go multiproduct. Some of them were too slow for AI. Those are 2 of the big causes. But it's endemic, and what it really means is that in the old days—and I'm talking through 2022—if you got to scale, especially in B2B, if you got to $50 million or $100 million, as long as your team was decent, you were set.

You were set because revenue was so durable. Would you grow from $1 million to $4 million in a year like Zoom did? Maybe not. But Zoom is growing 0% now, right? All of our assumptions about durability were wrong, and that is just super stressful.

So even if software overall, because of AI, goes from 2% to 4% of GDP and births thousands of new winners, we're also constantly looking over our shoulder now because we don't get to relax. There's no durability in our revenue anymore.

I think this is what's concerning. There is this whole generation of companies that's growing in the mid-teens, barely profitable or just profitable, and there's a whole load of people waiting on them to go out. They're not good enough to go out. They're not good enough for private equity, bluntly. They don't have the growth rates, and they don't have the profitability.

I look at that cohort and I think, “I'm not sure what happens to that.” Is it consolidation? That's a big task to do. I don't know what happens to that generation or where that liquidity comes from.

Stepping back for a minute, I think it is a huge issue, which is fun to talk about, but I think it's going to get worse because of this lack of durability.

You can take a stalled unicorn in your portfolio. We don't need to name names. Everyone has 1 or 2. You can say, “Oh, it's because Jason wasn't a good enough CEO,” or, “Jason was too slow,” or, “Jason worked from home.” That will probably be true for that company.

I don't think in 2026 we're going to be saying that. In 2026, we're going to be saying, “Look, we have to reset our expectations of revenue durability,” and that the idea that companies might radically slow down at later stages is going to become commonplace. That hasn't been commonplace—not just in venture, but for employees joining, too.

It was a safe bet to join these late-stage companies because they were set. Asana was set, and Twilio was set, and all of the others. But this assumption is going to fade away by next year. We will no longer think any revenue is durable.

I actually think it will impact the value of these failed unicorns because they'll be less attractive to buy.

When we think about the durability of these unicorns, with respect, the ones that we consistently see are Windsurf and Cursor.

Yeah.

When we look at the durability of revenue in that case, Cursor has obviously just raised at, I think, a $9.6 billion valuation. How do we think about durability of revenue in that case? It obviously hasn't impacted value when you look at that.

Well, look, my view today is—and I’m looking at my own portfolio—I think VCs, at least, have given up caring about durability. They only care about growth. We only cared about growth in 2021.

You can say Cursor and Windsurf and others, or even OpenAI, “How could they be worth this much?” You could even criticize Scale’s margins as a software company. But actually, if you step back and think of them as a multiple of next year’s revenue, they’re often not that expensive, right?

What is Cursor predicting? This isn’t made up. This is a slope, right? You’re predicting—you’re just taking a Google Sheet and dragging out 99% monthly growth over the next couple of years. I think a lot of these companies anchor at around 20X revenues, maybe 20X 2026 revenues or 20X 2025 ending revenues.

I think, on paper, VCs—outside of froth—are making these 20X ARR bets. Froth is a huge issue. It’s back, right? It’s just: of what? Is the revenue really ARR? Is it really recurring, and is it durable?

In some ways, these crazy AI deals, once they take off, are the cheapest from a revenue-multiple perspective compared to seed and A deals. Those are insane revenue multiples, right?

You said it was the best place to play in venture in a tweet recently. Why do you think it’s the best place to play in venture in these kinds of growth AI rounds, moving large amounts of money?

3. Momentum Investing Returns

Well, first of all, Harry, honestly, it’s what a lot of folks we’ve known for years are doing. I know several leading VCs who, about a year ago, said they were only doing AI investing. They weren’t even going to do any software investing.

There’s a high overlap, right? I’m not saying it’s completely separate. But I didn’t get it at first. Is SaaS dead and all of that?

Times are good like today. Times are weird, with the stock market down today and public multiples being crappy, but they’re also really good. If you want to make easy money, have the best beach home and the biggest plane, and not have to fly first class but fly private, you probably want to be a momentum investor.

It was such a good play in 2021, and it’s such a good play right now. Everyone that can throws some chips into the deal. If you have $100 million of chips, why do a seed deal? Throw it into any hot AI deal and watch it grow 5X, right?

We could debate whether you’ll get that money back and what will happen at the IPO. But VCs, no matter what anyone says on Twitter, are still graded on markups. LPs are fine with it, no matter what they say on 20VC, because LPs are graded on markups—at least a subset of them.

We’re in a markup world, and when markups are crazy, hysteria sets into venture. Hysteria sets into venture when markups are easy.

I totally agree with you. It also does—

Yeah.

—it also helps with the problem of duration. When you think about the duration problem for pre-seed or seed investors, it’s 15-year hold periods, often now. If you’re coming in 3, 5, 7 years in, or at $200 million ARR, a lot of that has been de-risked or removed in terms of timeline, and so your time to liquidity is just much less.

If someone had told me what time to liquidity really meant when I started investing, I don’t know that I would’ve done it.

What do you wish you’d known about time to liquidity that you know now?

I mean, I’m being a little facetious, but when I started, outcomes were smaller. Shopify and HubSpot both IPO’d at $1 billion, right? So you could get there faster because the outcomes were smaller.

The reality is that to get to a $10 billion or $20 billion valuation, for most companies it’s going to take 5 more years to get to this phase. So it’s really 20 years. It’s not just 15. It’s 15, and then you have to IPO, and then it’s 3 years to distribute after 6 months.

We’re talking about 20 years to real liquidity, to a real fund. There’s nothing wrong with it, because, again, if you’ve done well, you can ski and hang out at the beach house and go to events and stuff like that. But 20 years is so far away. I don’t know, as a human being, how you process getting all your money back in 20 years, right? I don’t know.

That’s depressing.

I don’t know.

It’s depressing to think I’ll be 48.

Yeah. I mean, in 2021, Bill Gurley made the point—or 2020, maybe it was after—but looking back, he said, “You have to take advantage of these windows,” right? As an investor, all my liquidity was basically in 2021.

There’s a brilliant LP that we both know who has analyzed different return profiles and liquidity windows across the last few decades and realized that venture is a very poor asset class unless you take advantage of these very narrow liquidity windows, where there are 6 to 12 months in which you actively are a very strong seller.

A strong seller. And I do think, as part of this golden age, it’s going to—we are seeing more deals. ServiceNow, which has been conservative in M&A—we’ll talk about it—just did its biggest deal ever for Moveworks, at $2.8 billion, and then bought Logic.io today for probably 5, 6, $700 billion.

So there’s going to be a wave of more M&A, especially if—there are pros and cons when looking at the stock market—but if Trump remains pro-M&A, as long as your pesky Europe doesn’t get in the way, this is going to be a golden age of M&A. Just a golden age of M&A.

We were talking about this generation of companies that are much more mature but have low growth. If we look at a generation of companies that have insane growth, from $1 million to $100 million, you know, Macaw just joined the 100 millionaire club. I think it was announced today.

You’ve got several others who are moving at unbelievable speed. Lovable and Bolt are obviously two very well-known ones. Is triple-triple, double-double misleading for founders in a world of AI?

Well, look, it’s a great niche question. What I can tell you is just this: for B2B investors that you and I know well—I mean, you know everybody, Harry—I would say 80% of them won’t touch a triple-triple, double-double deal today. Eighty percent.

This is the message to founders. Sequoia sends these memos every once in a while. There was the old doom memo. I sent my first email ever to the founders I’ve ever invested in.

What did it say?

It said this: don’t misread the market. Triple-triple, double-double is good for 20% of VCs, but only 20%.

If you’re doing great, if you’re growing 100% at double digits in revenue, and you deserve every kudo on the planet—you deserve it—you will build an iconic company if that revenue is durable, right? Expect 80% of VCs to say no. Eighty percent today.

It’s a gold rush, and in a gold rush, no one’s mining silver. I don’t even know where there’s gold, but there must be an expression in a gold rush: no one’s mining silver. No one’s firing up the old copper mine during the gold rush, because it’s just so easy to pull gold out of the river, right?

That’s the reason. It’s not that VCs don’t think you can make money in triple-triple, double-double. They’re looking at the public markets and saying, “I don’t want to be stuck with a $3 billion IPO in 10 years.”

So you—

“I want to put the money into Anthropic.”

So Jason, what do I do?

“I want to put $100 million into Anthropic and turn it into 5 million in eight months. I don’t want to invest $10 million in your startup and wait 20 years for a $2 billion exit. I just don’t want to.”

VCs just don’t want to do it, right? And so, yes, just like durability, we’re changing it. For now, in this gold rush, top 1%—that’s still top 0.1%—triple-triple, double-double. I think most of them are—you’re not unfundable, but you’re unfundable by 80% of the folks we know.

Eighty percent don’t want to touch it. They don’t want to touch it.

I spoke to one friend who’s an investor in one of the large LLMs today, and he said, “Listen, our price was X and it’s now Y. We’ve had a 12X increase in price.”

Yeah.

“We’re actually 3.8X up in actual returns because of the insane dilution levels that we’ve seen.” That is something one doesn’t see, and it’s been really, really damaging for a lot of the LLM investors, just because the level of cash that’s required is so seismically different.

Well, yeah. OpenAI’s new round is double-digit dilution, right?

Totally.

And employee dilution is high, right? So that’s probably approaching 10% a year in some of these companies.

It’s because I don’t have the data, but if the best LLM companies are paying their engineers $600K to $1 million a year, okay? If it’s double or triple the market rate, doesn’t the equity have to be...? You could argue the equity’s lower because it’s so highly valued, right? I’m sure they give you a Black-Scholes analysis.

But net-net, the dilution has to be higher. So if a typical startup’s doing 5% to 6% dilution a year, what is a hot AI startup that has more than 3 employees? Well, if it’s 1 employee, there’s no dilution, right? But I bet the dilution at some of these companies is 10% a year too, right?

So what do we do when we think about how we package up traditional enterprise software companies or SaaS companies better for a fundraising market? We both have existing portfolios, and there are a lot of SaaS founders who’ll be listening. What do they do to make themselves more attractive or package themselves more correctly in a world of your Macaws and your Scales, where insane growth is so inherent in investors’ minds?

Yeah.

4. Founders Need Loose Processes

I think what’s going to come for folks doing really well who aren’t at the AI growth levels but are still in the top 0.1% is a vibe like the old pre-seed vibe, which is, “It only takes 1.” What it means is that I think you need to build relationships over a year, do that great monthly update to everybody, build it, and get excited—and do it for real.

I think that folks who believe in you, especially folks who are confident and have a good hand, will be there when you cross a moment. That may be when they haven’t gotten a deal done this quarter or this year. There’ll be a moment in time, and if you get to know 50 investors who believe in you, then over a year I think your odds go way up, right?

If you’re trying to—I hate this term, Harry. I hate it because it’s great advice for the top 0.1%. I hate this “run a process” term.

Why?

I think it’s some of the worst advice founders get. I love Y Combinator, but I think it’s some of the worst advice Y Combinator gives their founders too, which is, “Run a tight process.”

Why?

“Run a tight process.” Because when you can run a tight process, it’s great, but most founders can’t. They need to run a loose process.

I can’t tell you how many founders I know who are like, “Here’s our data room.” Why do you need a data room? You’re a Series Seed founder, Series A—what’s your data room? Just give me your deck.

“Here’s our data room, and most of the round is full, and you have until Friday.” And you open up the data room: “Well, we’re growing 82% at $4 million in revenue.” I’m like, “Maybe you played that one a little too hard,” right?

But they’re trying to put some form of time constraints around it so VCs—

Don’t. My advice is: don’t. Don’t put time constraints on if there is no time constraint, okay?

As an investor and as a founder, if you do this proven playbook of getting a list of 10 or 20 VCs every month who you know personally, who you’ve met, and who believe in you, and give them an update every month, they see it every month. This month you’re growing 8%, and they’re like, “It’s good. Good job, Harry,” right? Then 6%, then 8%, then 10%, then 12%.

You’ve seen it for 5 months, so you have more confidence that it’s not fraud. And there’s a lot of fraud in AI today. We can talk about that. It’s 5 months of this. Eventually, you’re going to reach back out and say, “Harry, hey, want to catch up?”

Now, that may not lead to 20 term sheets, and it may lead to a non-optimal valuation, but I think that’s a higher chance, for real, of getting a deal than not, right? Because if you run the hysteria-froth playbook, you better deliver. You better have the numbers. You better have the numbers to back it up, right?

And literally, I just did a company that we’re both investors in, but they haven’t announced a round. They just did a round at $500 million.

One of our friends’ investors said, “Don’t even send it to me because I don’t want—I want more than 1 hour. I want more than 1 hour.” One of our old friends just said, “Don’t even send me the deal because I won’t have time.” I’ve followed this company because they’re an LP in my fund, so I’ve followed it since $30 a month in revenue, since I was the first investor in 2018. They said, “Don’t even send me the deal because I won’t have time.”

So just make sure if you run this hyper-aggressive playbook, you realize that. I’ll opt out of it 99% of the time. I’m a quirky investor. But 99% of the time, if you send me this exploding timeframe, I’ll say, “More power to you, Harry. Go close it, man. It’s not for me.”

What about folks who can’t work at that pace or just did a deal last week? What if you did a deal last week? You go from hungry to full, right? So anyhow, that’s why I just think you’ve got to be careful.

If you’re not Cursor, Podium, Lovable, and friends, maybe be a little humble. Just a hint: humble. Confident but humble in your process, right? Maybe it’s okay to have a valuation 30% lower if the deal happens.

Listen, when things go faster, cracks appear. You mentioned more fraud’s taking place in AI.

Yes.

What are you seeing before we have Bhavin join in 5 minutes?

I’ve seen several AI startups over the last couple of months go from zero to millions in revenue in a couple of months, and from zero to 8 figures in less than a year, where the founders quite honestly showed me under the hood that the AI was barely there.

It was often humans running a prompt and then shipping that prompt to an ignorant customer. It was often humans running a report in a BI tool with a little bit of AI and shipping it to customers that are unable to do this type of analysis, right?

There’s so much demand for increased efficiency from AI that sometimes the definition of AI has been stretched to implausibility, right? There were a couple of founders I love. They’re SaaStr super fans. These guys are great, okay? They went from zero to $2 million in 60 days.

I asked them, “Let’s—show me a demo.” And they’re like, “We’ve got to be honest. There’s really nothing to demo.” I mean, what do you mean there’s no demo? They just opened it up, and they had built a nice little wrapper around ChatGPT and just wrote content for big, ignorant customers.

I’m not saying that’s not real revenue. It’s earned, right? But when you move this fast, you might not even know. You might not even know.

Do you think VCs are doing their diligence like they should? Do you think we learned from 2020?

No. No. One of these deals just raised at an insane valuation, but I think we’ve become addicted to top-line growth. We don’t care about gross margins. We don’t care about profitability, right?

OpenAI said they’re not going to be profitable until $127 billion in revenue. $127 billion, okay? So we don’t care about margins. We don’t care about profitability, right? A handful of folks care. And even when we care, then the deal gets hot and we all want to do super pro rata.

We cared yesterday, and then when we have to give up our pro rata in the deal, now we don’t care about burn rates or gross margins or anything, right?

Well, I was walking in the park with a founder this morning, and he said, “Hey, if you can’t get your company profitable in an age of AI after the seed round, honestly, you’re running your business wrong.” Do you think we actually see a reduced role for VCs in a more efficient, company-run world with AI?

Yeah.

No. I don’t think—I think very few people believe that, other than a small circle of folks in the SF Bay Area, and I think there will be a few of these. One, there’s way too much competition. You can’t stand still.

There’s no question that RevenueCat, where we’re both investors, did an analysis recently, and they got 2X the productivity using AI tools—using Cursor and Codeium. That’s huge at their scale. Forty percent of all mobile apps run on RevenueCat. They’re lean and mean. They’re cash-flow positive. It’s a tiny team for their scale, okay?

They got 2X productivity. But what do they do with the 2X? I’m like, “Can I talk about how you got 2X productivity?” They’re like, “Well, sort of, but we plowed it all into new hiring.”

So they took the productivity, and they didn’t keep the head count flat. They just hired twice as many people who are twice as productive. So I don’t know whether that’s 4X or whether it cubes out to 6X to 8X or 16X.

But how can you sit still with your little 2-person company in San Francisco when the best of the best are reinvesting those gains? They’re reinvesting those gains, which is stressful because the bar has gone up. So I’m like, okay, that’s a good, nuanced point, right? Twice as productive, but we’re reinvesting all those gains.

So you’re not going to—you can’t keep up. And this is agentic AI, where there are no moats.

If there's no moat, you better run faster. This one guy, one person with Lovable, is great if there's no competition, a huge moat, no bugs to fix, and no other issues.

But dude, I literally was at a board meeting yesterday with a company I invested in in 2019 that's at about $20 million now. Great company, I love the founders. Not a rocket ship, but great, right? They had 2 competitors that I knew of, 1 of which got acquired and 1 of which wound down. Pretty good, right? I mean, not perfect, but you know. Yesterday, the board slide couldn't fit on the slide. He had 11 columns with different categories because of AI, and he's like, “This is the world today. This is the world today, right?”

What does the board say to that?

3 to 100 competitors.

What does the board say to that? How do you respond?

It was divergent. I think 1 of the board members was very concerned. They're like, “My God, you gotta sell your company,” right? One was neutral. And the folks that have been around the longest—and honestly, Harry, the funds where it's already in a fund that's way up, including 1 other VC—were like, “Dude, you're hyper-committed. You know the market cold, right? The customers love you. Reboot the company. We all have to reboot our companies in 2025. Reboot it however you want.”

What does that mean—reboot the company?

You gotta—whether you're AI-native or AI-whatever—you gotta be part of the future, right? Because that's where the customer pull is, right? That's why Moveworks was acquired for $2-something billion. It's where the market pull is, right?

Totally. I completely agree with you there. Was the founder receptive to that?

Yeah, I think so. For sure. It helps that they're cash-flow positive, too, right? At $20 million.

That's super helpful. I think we've got—

Super helpful, right?

I think we've got Bhavin joining us.

Yeah, he's right here.

Amazing. There we go. I mean, perfect timing. We segued that one perfectly. Bhavin, it is great to have you, man. Thank you so much for joining us.

Hey, man. Congratulations.

Can I ask, Bhavin, how does it come to be? How does that acquisition come into existence? Does Bill McDermott call you up and say, “Hey, let's make this happen”? What does that look like?

5. Moveworks Joins ServiceNow

Well, there's a saying, I think—and you guys have referenced this in the past—that great companies are not sold, they're bought. We've been building this company for about 8 years, really with the idea of using AI to transform the workplace. As you guys know, I started with IT transformation and moved across the entire enterprise, especially in this new agentic era where we got really good at providing this agentic reasoner.

Long story short, over that period of time, about 250 of our 300 customers were also ServiceNow customers, and we could see ourselves working really well together.

That's often a big part of the story, right? The overall—

Customers were sort of like, “You guys work super well together. We love both products. ServiceNow has this agentic platform. They've got a lot of work orchestration, systems of record, databases, and the like, and you guys have this great employee layer.” So, Harry, if I was probably honest, it was a conversation that naturally would've happened at some point. And it did.

I think I've been watching the market. We've seen tons of enterprise interest across the globe, especially in the last 9 months. I'd say in the fall, something happened and we started to see a lot of demand for our product. We talk about growth rates in venture and startups at board meetings of 100% or 200%, but that's still not fast enough.

I think if you look at the appetite of the world out there, I've got 350 customers. Bill and the team over there have 8,400 large enterprise customers on their platform. How do we actually go from where we are to that scale? I think this opportunity just felt so right that I decided it was the right move at this time.

Did you guys have that overlap? I think you said 30% of your customers were on ServiceNow or something like that. Maybe I misunderstood. Did you have a joint go-to-market motion that drove awareness of the deal? Or was it competitive in some ways? I don't know. How did that overlap drive the deal?

Yeah, so we were integration partners for 7 of the last 8 years.

Yep.

We had the ability to work very well within their system. Just for the sake of numbers, it's 250 out of 350 customers that are ServiceNow customers, so quite a bit of overlap.

Yeah, that's pretty high overlap.

Yeah.

Most. Let's call it most.

Most. What we found is that we were continuing to focus on our trade, which is this agentic employee experience layer that could do search and deep automations.

We even had to figure out a whole new architecture around how you build automations in the agentic era. How do you think about this translation from ambiguous language to precise APIs? It's not easy. In fact, you have to build separate reasoners, slot-filler models, and what we call a manifest generator to figure out which plugins to select. There's a lot of detail that goes into this.

As we did that, I think we started to find ourselves, especially at some of the largest companies, like Palo Alto Networks, CVS, Siemens, Unilever, Hearst, and Honeywell. The list goes on and on. We started to get very good at the top end of the market to solve not just employee productivity, which is what I call weak ROI. It's like you save Bob 3 hours a week—who cares? You're still gonna pay my salary.

The real strong ROI is where you can actually transform core business processes. Everyone's talking about MCP servers and all this stuff right now, but we had to figure out how you actually integrate with SAP, Workday, SuccessFactors, ServiceNow, Salesforce, Concur, Jira, Freshservice, and all of these systems, do it well, and actually be able to trigger deep core business processes. That's where you get the strong ROI, and I think that was why this partnership felt so right for both parties.

Can I ask, with the 250 out of 350 in terms of the overlap there, I actually have a company where they have a partner who is much, much larger than them. Frankly, they will and do have the opportunity to be acquired by them, and their fear is that if we don't say yes to the acquisition, they could go with someone else—one of our competitors—as the partner.

We're almost strong-armed into an acquisition because we are so reliant on that partnership. Respectfully, was that the case here? What would you advise me, advising this founder on how they should approach it, given they are very nervous that if they don't say yes, that partnership could go?

A slightly different story for us. You all know my last company, Refresh, was acquired by LinkedIn because we were using a lot of data from LinkedIn and Facebook and these sorts of deep web databases, right? That was probably more like what you're describing, where you sort of have to—you know, it's an IQ test of what decision to make.

For us, though, we have been building this platform as an independent service across all these different systems. We have been supporting ServiceNow, but we're also supporting Jira, Freshservice, Microsoft, and all these other services. Even though we have customers that use ServiceNow, they're also using all those other products, too.

We will continue to support and expand the reach of our platform with this, even past this acquisition, because that is where the gap is in the market. That's where people want something that interoperates with everything.

In your scenario, I don't know the details, but sometimes there is a situation where that is very true. In our case, ServiceNow was 1 of many deep partnerships that we had across the enterprise ecosystem. They were just a company that I admired a lot. What Bill McDermott has done—they're 1 of the most underrated companies, I think. They don't get discussed very much on podcasts or brought up very much in the circles that we all spend time in.

When I started Moveworks, I didn't even know ServiceNow.

We don't talk enough about the deep enterprise. It's just not in the interest area of most folks, right?

I felt the same.

Yeah, I mean, I went and built them.

That’s why I bought Standard a month ago, and now that I’m down 25%, I’m like, “Oh.” But yes, I totally agree. Bill is incredible. Can I ask, how was Bill as a negotiator? When you’re discussing the deal with him, what was that like? He is a generational leader. How did that discourse come into play, and how does one come up with a price today?

Obviously, I won’t share private conversations or what shouldn’t be shared. But I will say this: Bill is exactly like you see him. He is very warm, very generous, very honest, and what he says, he sticks with. I think I’ve learned a lot just from being around him and watching him over the years, but now more closely.

I think it was a very welcome and warm conversation, if you want some color around how these things got done for us. I don’t think all of these types of conversations happen that way across the industry. But for us, the partnership was already there in many ways, and so I think we could both speak each other’s language on the first conversation.

What he says is often stuff that I’ve said, and vice versa. I think it was a real treat to spend more time with him. Then, of course, there’s Amit Zavery, who’s now leading—he’s their president and chief operating officer—who worked very closely with us on this as well, really thinking about how to bring this agentic layer to every corner of the business.

I think the 2 of them have had a strong thesis around where they can take ServiceNow and what we can do together. This isn’t the last you’ll hear from me. I think, as I said, we’re going to continue to build this future together with ServiceNow, and Moveworks will continue to proceed forward. When this all gets closed with regulatory approvals, my team will be pushing really hard and aggressively toward this vision.

Was there any concern for you about the regulatory approvals? We’ve seen, obviously, quite a stringent regulatory environment. I’m intrigued, when we look forward over the next 12 to 24 to 36 months, whether that plays into founders’ mindsets and their willingness to engage in processes or not. Did you think about that? Were you concerned about that? How did that play into your mindset?

It was the first time I’d been through something like this where that factored in. I took the counsel of our experts and the folks involved. But I think we do see it as a very beneficial thing for the market, and we’ve already seen the benefits. A lot of times, you don’t know because there’s no precedent, but we have a lot of customers who have been very excited about working with both of us, which has proven that.

Obviously, I’ve got to let the regulators do what they do, and they’ll make that final decision. But look, with any sizable deal, all these factors go into it. Ultimately, as a founder, you have to do what makes natural sense. For us, I think Moveworks can become a worldwide phenomenon, and I think this was the best path to get there.

We’ll trust that the regulators will do what they need to do to ensure that this is configured the right way. But it isn’t the primary thing that we start with.

Can I ask 1 question, Harry?

Yeah.

I’m just curious. I remember when, right after AppDynamics got bought by Cisco for $3.7 billion, Jody Bonsall came to Aster. He was very specific about this; he was very tactical. He said, “Well, we did the math. At $3.7 billion versus 3 more years to IPO, with dilution and time, the IPO equivalent was $6 billion.”

The IPO equivalent of $3.7 billion, given time risk and dilution, was $6 billion. I don’t remember when that was. I’m dating myself; it was probably 2017. That was a big deal back then, AppDynamics for $3.7 billion. He said, “We couldn’t pass up the deal because of the…” I suspect Wiz went through that.

Did you guys, with your investors, go through that analysis and say, “Holy cow, it’s not just $2.85 billion, the headline price, right? But compared to what?” What would you—did you have this logical distance discussion? Because that’s even more money than it sounds from a founder perspective, because of time, dilution, and everything else, right?

You’ve got to raise another round—3 more rounds—hire another 1,000 employees, and that’s another 25% dilution, 10% from the IPO. Did you guys do that kind of trade-off with the future?

I think everyone does those calculations continuously as you’re building the business, deciding whether to raise more capital, and thinking about what it’s going to take. As I said before, I think ServiceNow is underrated. What I felt strongly about is that joining them to go do this and create this vision was going to be more accretive—

Yep.

—for all my employees than perhaps going at this stubbornly alone and continuing on the path. Don’t get me wrong: I think founders start companies with strong conviction about what they want to see eventually.

But as the market has evolved and I’ve seen the appetite increase so dramatically, the number of large companies that have come onto our platform in just the last 6 months has been insane. I realized we couldn’t catch up to the speed at which everything was moving, and so this made the most sense.

Of course, you do all those calculations, and so does everyone. You think about what the opportunity cost is. So, look, I think—

Can I ask you a follow-up?

—I think there are still opportunities for startups, but in this AI world, we do have distribution moats that are very important and that exist. I think they’ll still be important to think about as you want to build your brand and your vision.

I always worry that we drastically overestimate adoption in the short term and underestimate it in the long term. Do you think we are overestimating adoption in the short term for some of the largest companies in the world? You mentioned some names earlier. I’m thinking of European incumbents who still don’t know what Slack and Notion are, let alone are adopting AI-first principles.

Do you think we are overestimating adoption in large enterprises or not, given what you just said about the speed at which those customers are joining?

It’s actually a good question because there is nuance there. We’re seeing a lot of large enterprises, 2 years after ChatGPT came out, finally say, “Okay, we’ve got to do something, and we want to do something.” But I think there is still a very long journey in terms of true change management, the integrations, and leveraging these capabilities throughout their entire business workflows.

I was on a panel the other day, and someone asked, “What do you think we’ll be talking about in 2 years?” I said, “From the enterprise standpoint, exactly what we’re talking about right now,” which is the adoption curve, getting things in, and moving things along.

Again, we’ve seen this where customers absolutely are ready to go, but their own internal processes may take 6 to 9 months. You get the approval, then you start the implementation, you roll it out to a subgroup, and then it goes from there. I think we are seeing that transformation. It’s just happening slower—not because of the technology. To your point, I think the technology is there, but it’s a people problem.

Yeah.

I think change management is a very real thing that we underestimate, and it’s where startups and companies like ours have really focused our attention in trying to help these organizations figure this out and support them.

We did a lot of work over the years to really perfect this large-enterprise motion. We just got FedRAMP authorization 2 months ago, the first agentic platform to get that, because we had to build a lot of the security measures, protocols, and infrastructure on GovCloud. It just takes time.

Bhavin—

—and I think that’s where people overestimate or underestimate the work that’s required.

You mentioned distribution. If I were to push you into 1 camp or another, does AI benefit incumbents with incredible distribution to huge existing companies, or startups with speed, agility, and the ability to move much faster and think much quicker? Which one does AI favor more in the next 2 to 3 years?

On the leading edge, startups are always the first to bring out the newest capabilities. We saw this with DeepSeek, and we see this with search moving into research and some of the capabilities that are now possible.

But I would say that as that technology gets absorbed quickly into the market’s zeitgeist, the larger companies will continue to play a very big role in titrating the cadence at which this stuff really gets into the hands of millions of users. We have 5 million users on our platform.

ServiceNow has 150 million-plus. So if you think about just the reach that these organizations have, I think the incumbents of previous generations are different from the ones today because I think they're very innovative. I think all of the hyperscalers are very innovative. I think all of the large SaaS companies are very innovative. And so I think there are going to be winners on both sides of it. Sometimes, like Wiz and Moveworks, we combine forces. But I still think there's going to be a healthy opportunity for new companies to emerge and, if they choose to stay independent, do so.

Bhavin, can I ask just one meta question that I think Moveworks might be very helpful around AI, which is where the money comes from? What percentage of revenue are you seeing as budget for AI? Are you stealing from other sources of budget in the enterprise, maybe more dated workflows or systems? And how much is new AI budget that CIOs and others are giving to innovation or simply new budget? How much of your revenue is new budget versus stealing it from other systems, dated vendors, and so on?

I don't think there's one answer because every company is doing its budgets differently.

Yeah.

I will say that at the start of 2024, we started to see solid numbers inside budgets for people to deploy AI systems—agentic systems—to help their employees and drive business transformation. It wasn't just coming out of a slush fund or a one-off situation. These things are now in the budget. I've seen a lot of analyst reports where they think about where some of that is coming from and how it's being repositioned.

I'll give you an example. Broadcom started with Moveworks 6 years ago. They had 10,000 employees. Today, they have 50,000 after buying CA, Symantec, and VMware, and they have the same-size support team across IT, HR, and a bunch of other functions with the help of Moveworks. When you think about the budget, oftentimes the budget doesn't have to be looking backward. It can be looking forward: How do we reduce our cost of growth while the business grows? Does everything have to grow linearly with it, or can it be sublinear? Can it give us more leverage?

That's what we're seeing a lot with our customer base: they're able to do more with their budgets because we're able to keep certain costs lower and have the efficiency of computers doing it, where I think in other cases it was more labor-intensive.

Maybe that's good just to summarize. I think it is interesting that about a year ago, you started to see CIOs budget agentic AI as essentially a line item in the CIO stack. However you define it, that's growth. That is growth in the overall spend, whether it's coming out of not growing the headcount or coming out of getting rid of half of the support team. Forget about where; that's a separate issue, but it's budgeted. That is incremental.

I think that's the profound question there. Are we getting incremental budget? Because that's what's got to fuel this growth—not just for hyperscalers, but at the application level. Is there more budget? Otherwise, it's just moving chess pieces around the board. If it's incremental, that's why it's the gold rush. That's the gold rush: it's incremental budget.

Bhavin, dude, listen, I want to say thank you so much for agreeing to hop on. Thank you so much for providing liquidity to this ecosystem. What we forget is that when you sell—

Totally.

—when you sell, LPs get money back, and they reinvest it in managers like me and Jason. And, listen, it's been an incredible journey, so thank you so much, and well done again. Awesome. Thank you both. Great to see you.

Congratulations.

Thank you. Amazing. Thank you. Love that.

6. Private Markets Delay IPOs

As I said, dude, I think people forget the multiplier effect on liquidity and how LPs getting that back, and the recycling that comes from it, is so important. It actually goes to one of my questions that I wanted to ask you, dude, which is—

Yeah.

—I think we're going to see a drastic reduction in the number of IPOs. When you look at Stripe, when you look at Databricks—

I owe you $50,000, but keep going. Yeah.

—you do. I think it's actually $75,000, but, you know.

Okay, I owe you. I owe you. Here's the $75,000.

But my question is: are we going to see a drastic reduction in the number of IPOs with these case studies of, bluntly, the extension of private markets favoring founders in very effective ways?

I think we don't know, Harry. Look, first of all, IPOs are tough to pull off, no matter what it looks like in the media. There aren't that many buyers. It's a very niche thing to buy in the IPO when anyone can buy the next day. Most tech IPOs are flat 2 years out. So why would you buy in the IPO? It's very—you think venture is a niche asset class, so is whatever people buy in IPOs. A lot of stars have to align for it to be worth people's time overall.

Having said that, I think it's a gold rush. I think whatever we feel today—Stripe, Figma, Chime, the list of folks north of $500 million, Canva at $3 billion—that are going to IPO in the next 18 months, it may not be at the pace of 2021. I need someone smarter than me to put it in a spreadsheet and look at the number of IPOs in 2021. Intuitively, I actually think it should be higher just because of the scale today, but it's going to be a gold rush, and these concerns we have around liquidity in the ecosystem, I think, will remain. I think it's a huge issue, but we're going to forget about it starting next year.

Do you really think—

We're going to get so drunk. We're going to get so drunk on getting our Moveworks and Wiz cash back, we're going to forget that our cash is stuck in 300 unicorns. We're going to forget about it because no one wants to. You can't make money as a Debbie Downer in venture, can you? There's no money to be made.

So many folks we know, after a decade and a half in venture, become Debbie Downers, don't they? “Oh, everything sucks. Seed's overpriced. Ownership is down. Founders quit. There's too much fraud.” I've fallen into that trap once in a while. But you can't make money in that trap.

Sure. My question is: do you really think Canva and Figma will go public in the next 18 months? When you look at them, they have incredibly strong investor and consumer brands.

Yes.

They can continue to run in the private markets, so you can see the maturation of secondary markets mean early employees get liquidity and investors get liquidity. Do you think they really do go public?

My limit—I don't know, but I can tell you just one insight. It's a small one. One of these top folks that we talk about going public, but that might not need to, recently tried to acquire one of my investments for a relatively high price. They shared that their goal, quite plainly, was to IPO in the coming teens of months.

If they're telling that to a target just so they can understand the trade-off of cash and stock in the deal, I think that—listen, I'm not on every board. We need to get everyone on and ask them the same question, all the leaders, your Sequoias and your Mamoons. But I think there are some CEOs for whom, in the next 18 months, it's just time.

Yes, you can do endless tender offers, but a lot of that is driven by taxes in the US. It's not really driven by liquidity for investors. It's driven by this brutal tax situation of RSUs vesting. They're vesting on employees, and they're becoming taxable events before the IPO. That issue lurks: taxes and headaches for employees.

Anyhow, I don't know, but this was just one story of them saying, “Listen, it's going to come in the next 18 months.” And if they say 18 months, that means it's being planned, right?

One thing I think that we overestimate in venture is PE coming in and saving the day, so to speak. I think a lot of venture investors are like, “Ah, PE is going to be the savior.” And it's like, actually, Salesloft was a great deal, and that was one of the few where PE really paid up. But generally, it doesn’t actually work out that well for actual venture investors, I find.

Yes.

I'm more worried—

Yeah, Salesloft—I was one of the first investors.

Oh.

That was like the last deal of the last era, right?

Well done, dude.

$2.5 billion.

That was awesome.

Last one. Yeah.

What was the multiple on that?

I don't know. The seed was at 8, and then what was an A but really would be a C today was at 29. Kyle tells that story a lot, right? I forget after that. But I don't think it was 100X, right? So it's probably 40X or something. I don't know. It'd come back, right? But it is interesting: we talk about 100X, but even that doesn't deliver.

With dilution and time, it doesn’t really deliver. So it’s probably 40X, right?

Mm.

A lot of learnings. But my worry, Harry, I’ll tell you, is this: I’m bullish on the golden age, okay? Of my investments, there was Salesloft at $2.5 billion, but last year, one of the last of that group of investments was a company called Logical. They got bought for almost $300 million by PE, okay? So there are other examples.

My first investment was Pipedrive. That was PE for $1.4 billion. In my portfolio, I have companies that are north of $20 million, that are cash-flow positive, that have strong assets and strong NRR. I haven’t gotten any offers. I haven’t heard any offers—$50 million, $100 million, no offers.

So I know there are 400 private equity firms. I know Vista and Thoma Bravo have raised new funds. You have all these firms, and I’d love to get you to get them on your shows and ask them this question, because I’m not seeing the kind of tire-kicking that I saw a year ago. I’m not seeing it. Maybe other people with broader portfolios will tell you differently, but that’s what worries me.

I don’t know. We’re waiting for PE to bail us out, right? Forget about the multiples and the waterfall. I’m not sure they have the appetite. I’m not sure they want to be stuck with these assets where they’re like, “God, you know, Salesloft was a rocket ship. It’s none of my business how it’s doing today, but I don’t think it’s growing, right?”

They merged it with Drift, and Gainsight just merged with Skilljar yesterday. All these PE firms are mashing up these things to produce these Frankensteins to go public, and I think it will work. But does that mean you want to buy another one?

I don’t know if you’re seeing it in the US, but in Europe, the hottest thing in this weird intersection of PE and venture is this services play with an AI bolt-on. It’s: How do we buy services businesses and then juice up margins with an AI or technology play around them? That is the hottest thing in PE today.

Yeah. But are they buying unicorns that are growing 11%? I don’t know. I’m just not even seeing the tires being kicked. That’s my worry. I’m not even seeing the tires being kicked.

Yeah. No, listen, I totally agree. I think we’ve got Andrew joining us.

Now we can find out the truth about everything here.

I had Andrew on the show recently. He’s fantastic.

It was a great one.

Honestly, I got so many messages about that show. Andrew, it’s great to see you, man. I was just saying to Jason, I love doing our show. I got so many comments off the back of it, Andrew. I honestly just asked everything that I needed to know, and then I felt there was so much more that I wanted to ask you.

I’m looking at a lot of financings today, and when we chatted before, I asked you where value is in this ever-changing world, and you said compute and hardware. I know I’m looking at a lot of my friends who are putting a lot of money into compute and hardware startups, and I wanted to ask you: Should I be following them? What should I be looking for in them? I’m seeing some incredibly young founders. I was with a 19-year-old founder trying to take on NVIDIA this morning, raising $20 million for a pre-seed. How should I be thinking about this, Andrew?

7. Hardware Demands Experience

Well, if you don’t know a lot about hardware, I wouldn’t invest in hardware. I think, Harry, it’s probably the same in many things: hardware is not an easy place to make money. It’s a place that has historically rewarded experience, both from investors and from entrepreneurs. I think the number of different technologies involved in designing a chip is extraordinary.

Not just the logic, which is what most people think about when they think about chip design. That’s just the front-end part. That’s writing in very low-level software. The selection of tools—we pay millions of dollars a year in tools—the selection of geometry, which fab, and having a relationship with a fab: you’re going to pay $20 million or $30 million in NRE, and if you have a bug in your chip, you’ve got to pay it again, right?

The back-end design and timing closure are entirely different skills that companies like Google outsourced, broadly, to Broadcom. It is an extremely complicated and difficult place to be, and it hasn’t historically rewarded 19- and 20-year-olds.

Where the market has rewarded extreme inexperience is where founders look like the customer, right? The reason social networking was phenomenal for them, as you and I discussed last time, was that they were selling to their friends. This is something where experience might actually have negative value, right? We don’t understand what is top of mind for the 18-to-24-year-old, or 16-to-24-year-old, whatever. That category is extremely difficult to reach unless you’re in it, unless you understand it.

Historically, when graduate students write tools for their friends, they go through the roof. When people in college write tools and things that their friends want to use, VCs take that very, very seriously. When there hasn’t been a history of success in the same way in infrastructure software—in databases, for example—those companies haven’t come from very young entrepreneurs.

Sales force automation has historically come from salespeople. If you look at the history of Salesforce, back to Siebel, these were people who knew their audience. That’s where young entrepreneurs have a tremendous advantage. Perhaps in these extraordinarily complicated domains, such as chips and enterprise infrastructure, those of us with a little gray hair, who may be a step slower in our 40-yard-dash time, have some accumulated wisdom, and that pays some dividends.

Andrew, I’m happy to dive in. Can I ask you: Does CoreWeave open up a wave of excitement—a wave of investor excitement—from here? Does it open up a wave of net-new M&A and IPOs? How does going public change sentiment, change investor appetites, and change the ecosystem?

I think CoreWeave’s—it wasn’t an easy path to an IPO. They are an unusual business. They’re a creative business. Many of their innovations are in the finance structure. They were among the first to use the type of debt that they’re using, to recognize that you could borrow against GPUs and use their access to leverage an advantage.

For them and for their team, what an enormous success to get out the door. This is hard. The number of people who tell you you can’t do it, the number of parasites that jump on to try and nibble a little bit—you tell somebody, “I want a great slide deck,” and they say $40,000. You say, “It’s for an IPO roadshow,” and they say $125,000.

There’s just this relentless nibbling away at you, trying to tell you that you’ve got to do it their way or that you’ve got to be sure their clients get the big part of the first-day bump. As entrepreneurs, we’ve got to be focused on the people who are behind us, the teams we’ve built who’ve invested their careers behind us, and our investors who’ve been with us for long periods of time.

I think, to a person, the entrepreneurs inside CoreWeave are extremely happy with the outcome, and yesterday’s performance was phenomenal as the short sellers got squashed. I think their longstanding investors are extremely happy as well.

After that, Harry, the proof’s in the pudding. What you need to do is go out and execute. What you did on the first day doesn’t really matter. What matters is where you are in 6 or 8 months and where you are in 2 or 3 years, and whether you’re able to build the business and use your new capital structure to generate value.

I don’t know what you guys think, but it seems to me that we make a big deal of day 1. I think we should make a big deal of day 180, day 365, and day 450. Does this new money and the new capital structure allow entrepreneurs to execute in this new environment? I think those are the things that matter.

Whether your bank made money or whether it was hard to get out the door, I don’t think entrepreneurs should care very much about that.

There’s nothing weirder than being an employee the day after the IPO, right? You go back to work. You turn on your computer. Actually, you usually have a lockup. You’re not liquid, right? Even if you are, you’re vesting, right?

Everyone’s like, “Go back to work.” You feel weird. You don’t know: Should I be doing my normal job? Is it over? Is it just beginning? It’s just a weird day—the day after the IPO as an employee. Weird day.

I think that’s exactly right. It is not the end; it’s the beginning. It’s the beginning of adulthood in a company’s life, right? In a particular form of adulthood.

And you now have to execute. You have traded lower-cost capital for a different level of discipline and oversight brought on by both the markets and the regulatory infrastructure. What you need to do is go back to work and crush it every day.

Yeah.

Nobody talks about that. What they talk about is, “Oh, look, Goldman did this,” or “Citi did that.” What matters is: Did you manage to keep your engineers from being defocused? If your stock goes up a lot or goes down a lot, do you keep them from being—

Super distracting.

—an emotional whipsaw, right? I remember when we took a company public way back in the day at Riverstone. There was a day our stock was down and individual engineers lost more than their fathers made in their entire life. It is extremely difficult to work through that and not be affected.

Yeah.

You have to prepare your team, and you have to prepare yourself. Guys used to tell me, “We know what our stock’s doing based on the look on our CFO’s face.” That’s a bad place to be. Guys wrote little scripts so their net worth was streaming across their monitor as each trade happened, as their stock went up or down. You have to get past that. None of that is helpful.

Jason, how did you analyze CoreWeave?

Well, listen, I love what Andrew’s saying. They’ve found a very creative and clever way of using debt to monetize a gap in the marketplace, right? They’re able to provide cutting-edge GPUs and make them accessible and affordable. I have to say, I’m very confused by the financial engineering. I’m very confused by the debt loads being off-balance-sheet. I’m very confused about whether they will have the cash flow to pay it off.

What I really don’t like, going back to some of these old-days stories Andrew shared, is this put option that the last-round investors have to put back almost $2 billion of stock, right? I’m worried about that because I lived through it once as an employee, and it created a death spiral at a company.

Can you explain that? What does that mean for everyone listening? What is a put option?

Well, look, having read it, it’s discussed but underdiscussed. The last-round investors obviously weren’t fully aligned on the price, right? There’s a structure, and I don’t think this is a bad thing. It’s okay for late-stage investors to be more conservative on price than founders.

They agreed on a deal: “Look, if we don’t trade up 70% from the IPO in 2 years, you have to buy all our stock back. You have to buy all our stock back.” On paper, that might make sense to a late-stage investor. It’s not like they’re making a profit, right? It’s not like getting your money back is any profit for the VC.

But the problem is, what tends to happen if the company isn’t generating massive cash flow? Where are they going to get $2 billion? CoreWeave doesn’t have the $2 billion per se, right? I think they’ll probably get it. I would imagine OpenAI or Microsoft would give it to them.

But if it doesn’t, what happens? Andrew, I think it puts massive pressure on the stock price when people don’t think it’s going to happen. It sets you up to have your head cut off in 2 years, right? The shorts come in hard. They come in hard. I just worry about it. It’s stressful.

I’m not a public-market investor, and my comments weren’t about the deal structure.

Yeah.

I think there is a lot of complexity in their deal structure. For me, I’ve tried to avoid complexity. I think if you can get a term sheet in a page or 2, that’s really good. If you can describe very simply what both sides are getting, right? You’re getting—if you invest behind me, you’re getting our passion, our drive, and our innovation. You’re getting everyday drive. In return, I’m trying to deliver a return for you in line with the risk you’re taking.

You have to be able to make that very, very clear and simple. There are other people who’ve made a great deal of money and do really well in very complicated transactions. That’s not historically where I’ve chosen to participate.

I’m not an expert in the debt. I know the guys at Coatue did some of that debt, and they’re investors behind us. I know some of the guys at Altimeter did some of the equity, and they’re investors behind us. They’re extremely sophisticated investors. I’m sure they did their diligence and were very careful.

But in general, for entrepreneurs who I think are Harry’s audience, getting out the door is extraordinarily hard, and kudos to them for doing it. Kudos to them for being creative. Now the ball’s entirely in their court to execute. They have to deliver.

Yeah.

My fear with the put option is that it puts such pressure on them that you no longer have a long term. You have a 2-year horizon. You have a date that you have to deliver to, and that can create bad incentives, but I’m sure they’ll manage that.

The only question on the table here is: Can they move from being what they are today, which is interesting financial engineering and good execution, to something bigger and better? They bought Weights & Biases. I thought that was a step in the right direction to build a comprehensive software layer and provide some new value.

As an entrepreneur, you see people doing creative things and getting out the door, and you say, “Godspeed. We wish you well.” Now go and execute. Even though you guys deliver primarily NVIDIA gear, we’re 100% behind entrepreneurs doing interesting things.

Andrew, before we—

Harry, can I ask a different question? I know you might want to do more on CoreWeave, but because we have Andrew here, it is certainly trendy, in a sense, to try to sell to defense, right? All of a sudden, entrepreneurs are excited. I’m sure Harry met with a 16-year-old entrepreneur yesterday who wants to sell to defense. You’ve just closed a deal with DARPA, right?

Yeah.

My limited life experience as an entrepreneur is that this stuff is hard. It takes a long time. The money’s there. The money is there, right? But the qualification times and the time to market—I mean, my God. Maybe the world’s changed, but what have you learned? Is this a great place for VCs and founders to go? What have you learned selling into defense through this DARPA deal?

8. Defense Procurement Stays Broken

I think one of the real challenges for our Department of Defense is that their procurement structure is designed for them to buy without trust from Lockheed Martin and huge primes. When small, innovative companies come, they’re not parts of the Department of Defense that you bolt onto neatly, right?

When you go to sell to an enterprise, you say, “I’m looking for the senior vice president who’s responsible for AI and analytics.” That person is responsible for this, this, this, and this. Who’s responsible in the Air Force for innovative solutions using AI across X? There isn’t a person, right? It’s extremely difficult to find, so the sales process is really, really hard.

Yeah.

Then the contracting is mind-numbing. If Elon and the Trump administration could do something to change the contracting, that would help. When you contract with zero trust, you end up with Bible-sized contracts, right? If you don’t trust the other side to do anything well or to have any integrity, you end up trying to specify every case.

You tack on all these policy dimensions, and you end up with contracts that are literally this big. It’s a very difficult place to sell. There’s a reason why, for many of these, we partner. We’re doing a project right now with the Canadian military, and we partnered with Dell.

That’s why HPE Federal, Dell Federal, and the big integrators are often the vehicle in. They help you overcome this one piece, which is that the contracting is extraordinarily difficult. Finding the right people who are bold and innovative and looking for new technologies isn’t easy.

I don’t think it’s an accident that the people who’ve had success have done it before. You don’t have to be the founder who’s done it before, but your co-founder could have, or your head of sales needs to, or your COO needs to. You need to find people who have built a career in that community.

That takes a tremendous amount of effort and work, and they aren’t hanging out at coffee shops in Palo Alto, right? That’s not the—

Is having the Rolodex really transformative?

Huge.

Transformational.

Hugely important in DC. Hugely—

Yeah.

—important. And I think it's not just the Rolodex. It's a mindset of: This is how the military procures, or this is how commerce procures, or this is how the FBI, the CIA, the NSA, or the National Nuclear Security Agency goes about doing business. It looks very different. The sales cycle is very long, and the contract is very complicated.

Often, they will make requests. We are in the process of meeting some of these requests, but they might make requests that you have cleared people. How are you going to support your equipment if it's on-site at a secure facility? Either you need partners that are cleared, or you need to get cleared people. Sometimes they want your manufacturing site to be cleared, or they want your building to be cleared. What if you have Persian employees? Sometimes you have to put them in a building next door. The complexity of selling to the military—not just the U.S. military, but the Five Eyes, so the Canadian, Australian, and English as well—is hard.

Do you have to consider, when you're selling, that given cost-plus and, bluntly, their willingness to pay traditional pricing and operate in traditional pricing structures, and given your ability to sell to any customer on the planet, there's a real opportunity cost of your time and resources for something that might not actually be worth it, given the sales cycle, the procurement process, and then the cost-plus structure at the end of it?

Yeah. Look, cost-plus sounds like a good path, but it is a terrible way to do business.

Yeah.

It's a way of doing business that says, “We don't trust you, and you're going to slam as much cost as you can into the cost side. We're going to try and run around and make rules about what goes in and what doesn't.” We have a system for this called accounting. We have generally accepted accounting principles, and while they're not perfect, they're not bad. It is a mechanism that says, “Nobody in this process trusts anybody, and so what we're going to do is pretend that it's cost-plus.”

Now, if you've ever built a home and your contractor runs cost-plus, you know exactly what happens, right? The contractor doesn't negotiate with his window guy at all, because he's making 17% on top of whatever the window guy charges. The incentives are dead wrong. It is exactly the same.

What you want is innovative people. You want a structure that can take a little bit of risk, where if they buy innovative technology or invest behind innovative technology, they're not fired if it doesn't work. What innovative technology means is that the probability of failure is higher. That's what it means. That's why VCs need higher rates of return at earlier stages. You guys are taking more risk.

If you want to do truly innovative things, then your procurement arm needs to be able to withstand some false starts. They need to reward big wins rather than reward not getting burned. You need to have a mechanism for smart people to participate. It is a real challenge for the military to hire and retain world-class AI scientists.

I honestly don't know how the military or the NSA does it. I don't know how they recruit them, right? I don't know how.

It's hard.

I don't know how.

It's really hard. It's hard. I think these are all challenges. Now, what Anduril has shown, and I think what the war in Ukraine has shown, is that commercial technology can really have an impact in a way that was unexpected.

The drone warfare, using basically drones you could buy at Costco—I mean, these aren't heavyweight military drones, but they're drones that were designed for hobbyists—and 16-year-old kids are playing a meaningful role against Russian tanks. I think that is a wake-up call to the military and where interesting technology could come from.

It doesn't have to come through the traditional primes, the traditional Lockheed Martin or Leidos or these other guys that are so entrenched in building weapons systems. It can come from other places, and that creativity and innovation can meaningfully impact ground warfare.

I mean, there's a big push in the U.K. that when you look at defense budget spending—I can't remember what the number is, but it's 20%, I think we're pushing toward—that it goes outside of primes. It goes to net-new providers to encourage this kind of diversification of supply from different providers.

I'll tell you a funny story that happened to us. There is a mechanism in the U.S. government's contracting system designed for smaller companies. It's called an OTA, an Other Transaction Authority, and it was designed to simplify the contracts and make it easier to contract with us.

We have a big project with the government. The procurement agent agreed that this would be an OTA. The procurement agent took the standard contract and cut and pasted it into the OTA, so we had the exact same contract as we would have had without the OTA.

There needs to be a cultural change. I think that takes real time, and I hope that the success of Anduril pushes that. I hope Elon's success in showing the world that you don't need to be NASA to build rockets—turns out that really smart engineers in Silicon Valley, Seattle, and L.A. build rockets too.

It turns out that creative people under tremendous pressure in Ukraine can use commercial things that are just above toys to have a meaningful impact in war. I think all of these ought to be in the new calculus of how we organize the acquisition of innovation and the procurement of it by the government, and particularly the military.

Andrew, listen, dude, I cannot thank you enough for jumping on with us. I so appreciate hearing your thoughts. It's great to have you here.

I enjoy talking to you guys. I really appreciate you making a few minutes to chat. Anytime you want, I'm happy to jump on and talk about technology at any depth. We love it.

You're a star, dude. Thank you so much, man. Awesome.

Be well, my friends.

Stay well.

Cheers.

Bye. Awesome, my friend. Listen, before we wrap up, I just want to finish—

Yes.

—with a bet.

Okay.

You can have a bet with me. We've done bets before. I'm up. Whatever it is, $50,000 or $75,000.

Up.

Choose.

I've got to whittle it down.

You've got to whittle it down, baby. What's your bet going to be?

What's my bet?

Yeah, what do you think we should bet on?

How many deal executives leave in the next 90 days? Okay. You want me to come up with another one?

Uh—

I can come up with more.

No, you can do it. You go for it. What do you think it's going to be?

3. I say 3 executives. I'll let you define “executive” loosely so that you come out ahead on this bet. I'm going to say 3 executives leave in the next 90 days.

I think it's going to be 1. And—

Oh, you're going to lose. I'm going to do $50,000 on this. You're going to say 1 executive leaves?

No, I'm not betting $50,000. I'm betting about $100.

Okay. How about £50,000 instead of dollars?

Oh.

I don't know which way that goes now.

Then we've got to do another bet.

Yeah.

This is a sacrificial lamb.

I see. Okay. It's a gentleman's bet.

Okay, I say 3.

The one thing I think happens, though, now—and this is a new phenomenon that we've seen in the last few years—is that news cycles move faster than ever. With the rise of Trump, the rise of Elon, and the rise of Sam Altman and AI, bluntly, this is super hot today. It will not be super hot tomorrow—or when I say tomorrow, I mean—

Even next week.

—next week. It'll become stale very quickly.

It's a weird world. People just don't care.

It becomes stale.

Yeah. And so that's the only thing that I think. It's like, this too shall pass. I'm not saying that's acceptable, and I'm not passing comment on it, but cycles move so fast. I think, to your point on how many leave, given the speed of cycles, fewer than people think.

I'm still going for 3. I have thoughts, but I'm just being practical. I think it's just for the shareholders. I think it's going to be a very successful company going forward, right? Founder-led, but I think 3 folks have got to go. I think 3 folks have got to go.

But the news cycle—I mean, I'll tell you, we could break... You know what? You're a news expert. I can't believe we don't even talk about the fact that two people tried to assassinate Trump. We forgot about that in a week. If this was five years ago, we'd be talking about this for four years, right?

Dude, your president—

We'd be talking about President Clinton and Monica Lewinsky for a decade. We're not even—politics aside, we're not even talking about multiple assassination attempts against the president, right? I mean—

Your—

We forget in an hour today. We forget in an hour.

Dude, your president's flogging Teslas on his front lawn. And then JD Vance is trashing the European leaders of the world, saying, “Hey, you need to defend yourself.”

Yeah.

And everyone's moved—

We'll forget in a day. We're going to forget in a day. When these tariffs end, we're going to forget we had them. We're not going to talk about them. The day the tariff ends, next week we're not going to talk about tariffs, are we?

Ja—

It's all we're going to talk about this week.

Jason, thank you for doing The Daily Deal with me, my friend. I've loved doing it. You're a star, and I look forward to the next one.

All right, man. You're the best. Talk to you soon, Harry. Thank you.

Stay well, brother.