We’re back! Harry, Rory and Jason this week. The venture landscape has fundamentally shifted. A $4.5B exit is no longer enough for mega-funds, ownership targets are getting crushed across the board, and if your SaaS company didn’t see AI-driven reacceleration in 2025, you’ve failed. Meanwhile, the best Series A environment in years is hidden behind pricing pressure, and every “mature” software company better be capturing dollars from the AI spend explosion—or they’re headed for a 3x revenue PE buyout.
Top 5 Takeaways
1. The Exit Bar Has Been Raised—And It’s Brutal
Navan’s IPO at $700M revenue, 32% growth, and a ~$5B market cap should be a massive win. Instead, it feels like a disappointment. Why? Because mature SaaS companies are now trading at 6-7x NTM revenue — or less — and with fund sizes exploding, VCs need $10B+ outcomes to move the needle. As Jason put it bluntly: “Is a $4.5 billion exit good enough today? It’s a horrible thing to say, but if we’re talking about in VC and inside baseball, it’s a bonafide question.”
The math is unforgiving. Lightspeed invested $257M into Navan and returned ~$1B for a 4x blended return. Their early rounds probably returned 20-30x, but following with $200M+ in late-stage capital diluted that spectacular early return down to “just” 4x on the total investment. For a $2B fund, even a billion-dollar return only gets you a third of the way to 1x.
2. Ownership Is Getting Slashed—And Everyone Has to Adjust
Benchmark—the gold standard of ownership discipline—only secured 10% of Mercor at a massive valuation. Jason’s last three deals came in at 6-8% ownership despite his rule being double-digit ownership of two winners per fund. Y Combinator has institutionalized 10% rounds at demo day (3 on 30, 4 on 40), leaving seed VCs scrambling for 5-6% unless they “way overbid and basically do two rounds at once.”
The paradox: Both hyper-capital-efficient companies (that don’t need much money) and hyper-capital-inefficient companies (that need billions) result in low ownership for VCs. As Rory noted: “A founder’s optimized fundraising is a VC’s below ownership target.”
3. If You Didn’t Reaccelerate with AI in 2025, You Might Need to Fire Half Your Team
Jason delivered one of the spiciest takes of the episode: “You better have gotten a few nickels out of the AI expenditures.. Some growth from all that massive AI spend in 2025. If you didn’t, if you’re not growing faster at the end of 2025 than at the start, then as a founder, I give you an F minus.”
The data backs him up. Twilio went from single-digit growth to 15% growth fueled by voice AI customers (top 10 voice AI startups up 10x). MongoDB is benefiting from every AI company needing databases. Snowflake has re-accelerated. Stripe is booming from AI customers. Even if you can’t become Harvey, you must co-attach to the AI spend explosion.
The window is closing fast. “You had 18 months to ship a product that mattered in this world. Where’s your agent? Where’s your reacceleration? F. No more excuses after Thanksgiving.”
4. This Is Actually the Best Time to Be a Series A Investor (If You Can Handle the Pressure)
Despite the pricing pain, Jason argues Series A is in a golden age: “This is the best of times to be a series A investor because there is just an explosion of seed AI startups… It is a gift that thousands of founders are in San Francisco raising seed funds from multiple accelerators.”
The top of funnel has never been better. YC, Neo, South Park Commons, and others are creating an unprecedented pipeline of smart, vetted founders. Yes, competition is fierce—Rory noted losing deals to Andreessen Horowitz and Kleiner Perkins (firms that would have been “slightly early” five years ago). But the direction of travel is clear: “Some form of agentic software… The mission has been assigned.”
5. The Wealth Concentration in Silicon Valley Is Unprecedented—And Changing Everything
Buried in the discussion of Navan’s “disappointing” IPO was a striking observation: “The amount of wealth in Silicon Valley is just unprecedented in our lifetimes. It’s just gone up dramatically the last 18 months. In a very concentrated fashion.”
This creates a strange dynamic where VCs are getting richer than ever (Oren Zev as a solo investor turned $150M into $1B+, Lightspeed’s $257M became ~$1B) while simultaneously feeling the pressure to deliver even bigger outcomes. A $50M post-seed valuation for a company that “has to be way better than Navan” feels both expensive and necessary.
The Navan IPO: A Case Study in Modern VC Economics
Let’s dig deeper into why Navan’s $5B outcome feels bittersweet. The company survived a near-death experience (COVID wiping out travel), came back strong with $700M+ revenue at 32% growth, and delivered billions to investors. By any historical standard, this is spectacular.
But the lockup period reveals the gap between headlines and reality. Typical lockup: 6 months before you can start selling, and 18-24 months to fully distribute a large position. As Rory explained: “The IPO, I mean, it’s significant, but the economic significance is limited.” Translation: Those “billion-dollar returns” reported on IPO day often turn into $700M after 18 months of distribution.
Worse, the last private round priced at $9B. Anyone who bought at that valuation or in the IPO is now down 30-50%. As Rory noted, this actually disproves Bill Gurley’s “free money” thesis about IPO pops: “This is an example of an IPO where, you know, if you look at it from the perspective of the issuer of stock, not only did they not leave money on the table, they actually priced at what now looks like a high.”
The follow-on concentration strategy amplified returns but also amplified late-stage risk. Lightspeed’s early rounds probably delivered 20-30x, but their $257M total investment (including heavy late-stage follow-on) blended down to ~4x. For Orin Zev’s smaller fund structure with SPVs, the $150M-to-$1B return (6.7x) on a single deal is phenomenal. Same outcome, different fund architecture, wildly different impact.
The AI Adoption Dichotomy: Individual vs. Corporate
One of the most insightful discussions centered on adoption velocity. OpenEvidence grew to 300,000 doctors in one year—one-tenth the time it took Doximity. Individual adoption of AI tools is explosive.
But corporate adoption? Much slower. As Jason observed: “I don’t think everyone’s going to buy Agent Force or their equivalent in a year. It might be 5 years, not 10 like SaaS. It might be half the time of SaaS.”
This creates opportunity and risk. Opportunity: If you’re selling to enterprises (Salesforce, HubSpot), you have 5 years to get this right, not 1 year. Risk: Individual users are already using ChatGPT, Cursor, and other tools “legally or illegally.” Every associate is using AI to write briefs whether law firms have adopted Harvey or not.
The TAM question becomes critical. OpenEvidence might capture 300K doctors fast, but there are only ~1M doctors in America (half in-house, half external). As Rory warned: “You’ve got to be sure that once you get all the names that you have enough follow-on stories for those names. You don’t want to be one and done.”
Harvey faces the same math. To justify an $8B valuation at 30% growth (7x multiple at maturity), they need to hit $3B in revenue. With ~1M lawyers in America, that implies thousands of dollars per lawyer per year in subscription—not impossible (WestLaw is bigger), but a high bar.
The Harvey Deep Dive: When Does $8B Make Sense?
Speaking of Harvey: $150M ARR, 98% GRR, 170% NDR, 40% DAU/MAU ratio, raised at $8B valuation with minimal dilution (~2% for $150M).
Jason’s back-of-napkin math: “If it’s 400 forward revenue that’s 20x. That’s exactly what it is. 400 AR. That’s what they’re predicting.” Rory agreed: These low-dilution rounds at hypergrowth companies are great for early investors but create pressure on the ultimate exit.
The bull case: LLMs are perfect for law (language manipulation is literally what lawyers do), Harvey established market leadership fast, and legal had been a “pretty barren place” for software until now. The question mark: Is there a $3B annual spend in software for corporate law practices?
Rory’s framework: “You basically have to be such a big automation tool for these lawyers that they’re willing to spend, you know, equivalent dollars, you know, thousands of dollars per year in subscription to make the math work.”
Jason’s evolution: “I couldn’t believe that software companies could really capture dollars from replacing humans for real… Now we’re really seeing it. Agents are better than mediocre humans.” His conviction comes from using products, not Powerpoints. He’s running 20+ agents at SaaStr, replacing $40K workers with $10K/year agents, and seeing real results.
The Sam Altman Question: When $1.1 Trillion in Capex Meets Reality
The discussion of Sam Altman’s response to Brad Gerstner (“if you want to sell your shares, I’ll find someone for you in 60 seconds”) revealed deep concerns about governance and capital requirements.
Jason’s reaction: “When that was said to me by a founder, I never said a critical word ever again. I didn’t mean I didn’t realize I pushed too hard. I never said a critical word ever ever again when I was told there’s a market for your shares if you want to sell.”
But Rory pushed back: “If you are on the board of a company that’s planning to spend a trillion dollars and you only have 12 billion in revenue. It’s a totally appropriate board level question to say how are we going to do this.”
The stakes are enormous: “The health of the entire US economy depends on the answer to this question. It turns out ‘fuck off and sell your shares’ is not an acceptable answer.”
OpenAI needs to do $100B+ in revenue by 2027 to support the capex commitments. That’s not just impressive—it’s “hundreds of billions of dollars a year in a world where today only Amazon” operates at that scale. Google does $100B in a quarter; so does Amazon. OpenAI needs to reach that level.
Rory’s compassion for Sam: “There’s a little part of me that’s kind of a bit sorry for Sam Altman because of where he’s put himself. He is the poster child of the AI capex boom. And if the AI capex boom unravels and the world and media is looking for a villain to throw rocks at, it ain’t going to be a big search.”
The board’s job: “Help the CEO avoid things… provide the guardrails.” When someone invents numbers at the $5-10M level and they’re wrong, they disappear. “When you invent numbers that are wildly overoptimistic at the trillion dollar level and if it unravels, you just become the poster child in every economic history for the next 200 years of the great AI crash of 2026.”
The Public Market Tea Leaves: What AWS, Google, and Meta Tell Us
AWS: Grew 20% (reacceleration from 13%), secured OpenAI deal as latest partner, but the story is bittersweet. Five years ago, AWS dominated cloud compute. Today, they’re behind Microsoft ($250B commitment), Oracle ($300B), Google, Broadcom ($400B) in the AI race. As Jason noted: “It’s tough from AWS being having really created the category of cloud and being what we all grew up on. It is a major come down to be not even above the fold on the leaderboard.”
Google: The underappreciated winner. “Google is really good now at all levels. It’s really good at consumer AI. It’s really good in searches back. Search is growing for Google again. The TPUs are good. It’s a good partner.” Unlike Amazon, Google has the application layer (all the consumer apps we use), the infrastructure layer (TPUs), and is printing money. Stock up 53% from Q1 lows.
Meta: Core business performed great (20% growth), but stock dropped double-digits because they’re spending $70B/year on AI with “no revenue attached.” The market’s message: “You have this wonderful business and then you’re taking the entire cash flow and you’re building this AI stuff and unlike Google, Microsoft, or Amazon you don’t have an enterprise business to sell this shit to and unlike ChatGPT you don’t yet have an obvious AI forward app.”
Zuck’s response (paraphrased): “I think it’s relevant to be in this space. Thank you for your opinion. I refer you to the Artisan Corporation. I control this company. Have a nice day.” The market’s counter: “Hey you’re doing the thing you did in 2021-22 putting a whole bunch of capex into something that might not work and that’s why the shares are slightly down.”
The Salesforce and HubSpot Test: Enterprise Giants Need to Deliver in 2026
“If I were the CEOs of any of the CRM leaders, I might move on from half my leadership team if I don’t see real growth from AI agents by the middle of next year. Because there’s been enough time. You’ve just got the wrong people.”
The pressure is real: “There’s 2,000 people building AgentForce and we’ve deployed it.” said Jason. “It’s pretty darn good. It’s quite good. It’s very competitive with any other agent you’re going to buy. Time to monetize it in 2026. It works. It’s a good product. It’s not just smoke and mirrors. It’s really good.”
Why the urgency? Infrastructure companies like Twilio, MongoDB, and Datadog already saw the bump. Companies at the infrastructure layer get the AI spend first. Application layer companies need to prove they can capture it too—or risk being viewed as legacy players who missed the moment.
Rory initially pushed back on Jason’s harshness, then came around: “Sometimes when Jason is cruel and harsh, I disagree. And then sometimes I listen and I go, he’s absolutely right. This is one of the latter ones… If you’re not on this train now, you’re just not going to be relevant and you will be sold for three times one-year revenues to a PE firm who will smush you in with something else, never to be seen again.”
The contrast with Harvey is stark: Harvey has 2,000 people somewhere building legal AI. Salesforce has 2,000 people building Agent Force. One has 170% NDR and $150M ARR growing like a weed. The other needs to prove monetization in 2026 on a massive enterprise base.
Portfolio Construction in the Age of AI: Fewer, Bigger Bets
The ownership discussion revealed a fundamental tension in portfolio construction. Jason’s rule: “I feel like I can only make money if I own double digits of two winners per fund.” His last three investments: 6-8% despite that rule. His resolution for 2026: “Get my ownership up. We’ll see how I do it.”
Rory’s framework on late-stage follow-on: Early-stage funds aim for one deal to return the fund. Late-stage funds play a different game: “We got 4 billion. We’re going to put it to work. The average good deal will be at 3 to 5x. We’ll have a low loss ratio and over time we’ll get our two two and a halfx net. It’s not small numbers of big heads business. It’s a very different business. It’s moving money at scale.”
The challenge: When companies are capital-efficient, they don’t need to sell much equity. When they’re capital-inefficient (OpenAI, Anthropic), they need billions, so even $100M buys you only a few points. “Both of those, interestingly enough, would be pretty good deals… Your mental rules of thumb have been smashed to pieces.”
Harry’s counter: “You have to turn the next card to see sometimes. And it’s not obviously a $10 billion company on day one. And actually value can accrue in increments over time. And you could miss some great ones by being flippant in being like, oh, it’s not a $10 billion company.”
Jason’s response: “If your first check for me is a very large percent of the fund, I don’t have a lot of margin for error. I don’t have the other 145 million that he had in Trip Actions in Navan. For me, that first one has to work.”
The “Only Mortals Need Apply” Controversy
Jason’s most provocative statement: “I don’t even want to take meetings with mortal founders. I don’t even want to take them, especially if the ask is aggressive.”
Context: At $50M post (with everything in), Jason needs a 100x return to make money after dilution. That company needs to be “even better than Navan” to justify the price. “Do I really believe this deal I just did is for sure going to be worth more than Navan? I don’t know. It’s such an insanely high bar.”
Rory’s gentle pushback: “I recoil from the ‘I don’t do mortal humans.’ It’s a bit cuz I think, you know, it sounds a little judgy.”
Jason’s clarification: “Maybe it was always hard but now like you got to see these 10 billion exits but if they’re not utterly breaking the mold… it’s hard to really believe it’s going to be worth north of 10 billion.”
The sobering reality: “You now have to assume that 400 million, 500 million in revenue is the threshold for an IPO. And if you say to yourself that you only want to do deals where you at least have the upside of an IPO, the IPO potential, then the bar to what a doable successful venture-backed deal with upside has gone up.”
Rory’s synthesis: “The business has got harder. It has. But also people are getting richer at the same time. The amount of wealth in Silicon Valley is just unprecedented in our lifetimes.”
The VCs-as-Yes-Men Problem: When Boards Stop Providing Guardrails
One of the most important exchanges revealed a dirty secret about modern venture boards. Jason: “The better the company is doing, the more everyone’s a grin fucker. There’s just never a critical word said.”
His example: “I have a company that’s going batshit and it’s going too batshit… The board members will not intervene in any way to protect the shareholders because of the bad NPS that would come from damaging that founder relationship. That to me is one of the most egregious escapes from fiduciary duty.”
Harry agreed: “There is so much fear among VCs of getting out of step with the most successful founders. There’s so much fear.”
Rory’s take: “If it were a normal startup with the VCs we work with and they had had to come up with 1.2 trillion, I think everyone would be saying, ‘Sounds good, Sam. Sounds good, Sam. Keep going. Good month.'”
The contrast with OpenAI’s actual board: Brett Taylor (likely not a yes-man) and previously Larry Summers (“Larry Summers is many things, but not a yes-man”). Having strong, independent voices matters enormously when the stakes are this high.
Jason’s reflection on being told to sell by a founder: “That was a teaching moment for me. I’m like, okay, I crossed the line. Like I didn’t mean I didn’t realize I did.”
But Rory’s point stands: “You’re as a board member meant to provide the guardrails. I think when I look at this one, I go, ‘Hm, I hope someone’s providing really good guard rails because if it hits, it’ll hit hard.'”
The lesson: Young founders make mistakes, but experienced board members who stay silent are morally (if not legally) more culpable. “Do we really have an audit, Sam Bankman? Do we really know where the money is?”
Two Paths Forward: Playing to Your Strengths in 2026
The episode crystallized two viable strategies for software companies navigating the AI transition:
Path 1: Go Where Others Aren’t (The Roger Ehrenberg Strategy) As Harry noted: “Let’s go where others aren’t. Get 20% in actually reasonably priced assets. Let’s not do AI dictation tools from YC and get 4%.”
This means finding categories with less competition, being willing to take concentrated positions, and having the patience to build ownership over time. The infrastructure layer companies (Twilio, Datadog, MongoDB) benefiting from AI spend without being “AI companies” exemplify this.
Path 2: If You’re Slow, Pick Slower Categories Jason’s advice: “Only a couple percent of [Salesforce’s] customer base is fully ready for AI today… If you are a little slow, maybe play to your strengths. If you’re a little slow, maybe maybe go into retail or manufacturing or areas where it hasn’t changed overnight.”
The key insight: Individual adoption is explosive (OpenEvidence to 300K doctors in a year). Corporate adoption takes 5 years, not 10, but still takes multiple years. If you’re a smaller team or moving slower, find industries where the decision-making process is naturally slower.
But no matter which path you choose, the core message remains: You must be relevant to AI in 2026. Whether that’s replacing humans, automating tasks, or enabling AI workloads, you need a story. Companies without one will get “sold for three times one-year revenues to a PE firm” and disappear.
The Kalshi vs. Polymarket Question: Regulatory Risk in Prediction Markets
The episode closed with a fascinating debate about prediction markets. Both agreed the space is interesting, but differed on which bet to take.
Jason’s lean toward Kalshi: “Based on my limited knowledge, I’m going Kalshi because I feel like it is a safer long-term bet than someone that is riding the current political vibes which are all in favor of everything here.”
The risk: “There is a chance there will be a new administration that will be less sympathetic to this category… The first act could be to undo everything everything that Sachs and Buddies have done. That could be the January 1st whatever.”
Rory’s prediction: “I predict that any reregulation won’t happen because of any new administration. I think the real challenge to sports betting like this will actually be the leagues themselves wrestling with the fact that when you have sports betting, you have sports cheating.”
The Brian Armstrong example: Using specific phrases during an earnings call to tilt the scales on a Kalshi bet creates all sorts of weird incentive problems. “When you have these prediction markets and then the person can tilt, put their finger on the scales of who wins… There’s going to be a lot of weird stuff that happens.”
The broader point: Regulatory risk isn’t just about current administration views. It’s about second-order effects (league commissioners dealing with cheating scandals), insider trading concerns (policy bets moving an hour before official announcements), and the fact that “everything is legal” right now doesn’t mean it stays that way.
Quotable Moments
Jason Lemkin:
On the exit bar: “The horrible question in venture and startups, it is horrible. Is a $4.5 billion exit good enough today? It’s a horrible thing to say, but if we’re talking about in VC and inside baseball, it’s a bonafide question.”
On AI urgency: “You better have gotten a few nickels out of the AI expenditures… If you didn’t, if you’re not growing faster at the end of 2025 than the start as a founder, I give you an F minus. There’s so much money and you don’t have to be Harvey to get a little piece of it.”
Rory O’Driscoll:
On ownership: “A founder’s optimized fundraising is a VC’s below ownership target. That’s right. If Cursor had needed 40 million to get rolling, then Reedrock would own 20%. If they only need 20 million, then there you are.”
On Sam Altman: “There’s a little part of me that’s kind of a bit sorry for Sam Altman because of where he’s put himself. He is the poster child of the AI capex boom, right? And if the AI capex boom unravels and the world and media is looking for a villain to throw rocks at, it ain’t going to be a big search.”
Harry Stebbings:
On board dynamics: “There is so much fear among VCs of getting out of step with the most successful founders. There’s so much fear. And you guys are going to disagree with me, but I see it all across my portfolio. The better the company is doing, the more everyone’s a grin fucker. There’s just never a critical word said.”
On adoption velocity: “OpenEvidence grew to 300,000 doctors in one year, which is one-tenth of the amount of time it took Doximity. That was interesting just in terms of market pull and adoption rate.”
Final Thought: The venture landscape going into 2026 is simultaneously the most exciting and most unforgiving it’s ever been. Pricing is high, ownership is compressed, exit expectations are brutal, and the speed of AI adoption creates winners and losers faster than any previous technology shift. But for those who can navigate it—who can find the right companies, get meaningful ownership, help them capture AI spend, and provide real board-level guidance—the wealth creation is unprecedented. Just don’t expect it to be easy, and definitely don’t expect a $5B exit to feel like a win anymore.
