Jason Lemkin | 20VC Summary

1. Key Themes

The Optimist's Premium: Gross Margin Negativity Is Not a Death Sentence in Hypergrowth Markets

The Cursor deal crystallizes a core investing truth: in a rapidly expanding market, early-stage gross margin problems get swamped by the upside of market position. Rory framed the Cursor arc perfectly — it went from being mocked as margin-negative to a $60B exit.

"Your gross margin problem is my revenue opportunity for my Colossus cluster. So it just shows the negative issues didn't go away. They just got swamped by the optimistic take. And that's why, you know, it's the old cliche we talk about. Pessimists sound smart. Optimists die rich." 00:00:00 — Jason Lemkin

Agility as the Ultimate Competitive Advantage in AI

Cursor's journey — starting as an email client, nearly dying when Claude Code launched, pivoting to multi-model, and exiting at $60B — is the clearest case study in the episode that surviving in AI requires radical adaptability.

"The one thing I just kept thinking is how important it is to be beyond agile. Because I think so many teams would have given up on that journey, right? Oh my God. Fucking Claude Code came out. I got to build my old LLM. I got to do this. And it's just so hard to keep up with the rate of change. And there's probably been three different Cursors since we started." 00:05:25 — Rory O'Driscoll

The Roadmap Arms Race: If You're Not Deep Into 2027, You're Already Losing

Companies deploying AI effectively aren't just catching up — they're compressing entire multi-year roadmaps. The benchmark has shifted from quarterly delivery to being years ahead.

"I had two board meetings in the last week where they finished the roadmap for the year. They're into 2027. If you're not into your 2027 roadmap, deep into it by August of 2026 in the agentic world, your team is not good enough to survive today. This is your last chance to make changes." 00:00:37 — Rory O'Driscoll

The $100K Per Engineer AI Token Budget: The New CFO Playbook

A very specific and actionable prediction emerged: companies will normalize around $100,000 of AI inference spend per engineer, in exchange for reducing headcount 30–40%. This has enormous implications for AI infrastructure and model company revenue projections.

"I think we'll give each of our best engineers $100,000 of tokens. And in return, we'll cut the size of our dev teams 30%, 40% effectively... $100,000 here for running inference 24/7 with 10 agents in parallel." 00:00:37 — Rory O'Driscoll

The TAM Reality Check: $200B Is Plausible, $600B Is Mathematical Bollocks

The group stress-tested Anthropic's own projected revenue figures with hard demographic math. The conclusion: the $200B target is aggressive but achievable; the $600B number runs into real limits of addressable knowledge worker population and willingness to pay.

"U.S. is typically 50% of the world's software budget because we're 50% of the world's high-end knowledge workers... There are about 1.8 million people doing coding in the U.S. And then including QA and all the other people, there's around 5 million people that do software-related shit... And they get paid in total, grossing up about $600 billion a year. $200 billion means you're replacing a third of them. That's a lot." 00:33:15 — Jason Lemkin

SaaS Is Not Dead — It's Entering Its PE Harvest Phase

The Silver Lake / Workday deal at 5.3x revenues and ~16x trailing EBITDA signals that top-tier SaaS is transitioning from a growth story to a cash-generation story. This defines the ceiling and floor for SaaS buyout multiples.

"What this gives you is a sense of what the baseline is for best-in-class LBO takeouts... What Workday says is if you've got 30% operating margins, modest growth, but you're a system of record where you really can believe in the next five years, then if you're lucky, you get 5.3 times revenues. That's the bid-ask spread right now." 00:49:13 — Jason Lemkin

Closed Systems of Record Are PE's Best Defense Against AI Disruption

The episode surfaced a nuanced thesis: closed, hard-to-integrate systems of record like Workday are actually more insulated from AI disruption than open platforms like Salesforce, precisely because their closure limits the surface area for agentic substitution.

"Workday is so closed, they've got a leg up, right? Try building on Workday. It ain't so easy... If you're remotely open and you can produce a better agent yourself or a third party, the value will extract to the agent even if the system of record is retained." 00:50:56 — Rory O'Driscoll

PLG-to-Enterprise Is Still the Winning Playbook — Even for AI-Native Companies

Lovable and Higgsfield both demonstrate that the classic bottom-up SaaS motion still works in the AI era: build mass consumer adoption, then layer on enterprise features and mid-market sales.

"What both of them have done really well is parlay that kind of massive bottom-end demand for AI, in Lovable's case for website building and coding, in Higgsfield's case for video... start with a PLG motion and then add mid-market and enterprise products on top. It's a well-trodden path." 00:59:10 — Jason Lemkin

The "First to Market" IPO Advantage Is Especially Critical When Capital Needs Are Enormous

The standard wisdom that IPO order doesn't matter long-term gets inverted when both companies have hundreds of billions in ongoing capital requirements. Anthropic going first is not just a PR win — it's a strategic financing necessity.

"There are no two companies on the planet that need more capital than these guys. In a world where you do need the capital, big second sucks... I think you will regret not being able to access the capital markets this year." 00:41:29 — Jason Lemkin

Regulatory Laws Designed for One Era Become Landmines in Another

The DOJ's Section 8 Clayton Act investigation of Andreessen Horowitz's overlapping board seats on Databricks and Fivetran prompted a broader warning about AI regulation: laws passed for one context persist and cause unintended consequences decades later.

"You have to remind yourself. These regimes go on forever, right? And if we pass some law about AI regulation now, you've got to be really careful. The unintended consequences, you know, months, years, and decades later, these, once the regulatory law is passed, it doesn't leave." 01:08:55 — Jason Lemkin

2. Contrarian Perspectives

The Cursor Valuation Was Actually Cheap for SpaceX, Not Expensive

Conventional wisdom says $60B is a huge price. The contrarian read: given SpaceX's stock currency trading at 40x revenues, acquiring Cursor at ~10x forward revenues was net accretive on day one, and turns an internal cost center (Colossus compute) into a revenue engine.

"Elon has the biggest advantage, which is his stock is trading... it's still 40 times revenues. Picking up a big asset at 15 times current revenues, maybe less than 10 times year-end revenues. Dirt cheap for him, right? A net accretive day one in a way that probably wouldn't be as true for Meta." 00:07:49 — Jason Lemkin

Anthropic's Profitability Was Inevitable and Mathematically Predictable — Not Impressive

While the media treated Anthropic's first profit as a surprise milestone, the panel argues it was a near-mathematical certainty given the revenue trajectory and the structural impossibility of scaling costs as fast as revenue.

"You can't add expenses below the line fast enough to stop yourself making money. So it's inevitable... You're not going to 14x headcount or below the line training costs in six months. So yes, I'm totally not surprised they are operating in composite." 00:00:18 — Jason Lemkin

OpenRouter Is a Niche Product, Not a Platform — Stripe May Have Bought a Dead End

The popular narrative frames OpenRouter as foundational AI infrastructure. The contrarian view: it's strong in two narrow use cases (developer tools and chatbot fallback routing) but irrelevant for the high-stakes enterprise workflows that matter most, where companies down-spec to two or three models and never need routing.

"For high reasoning models, people do frontier-esque outputs, right? People don't rotate through 11 models. And I don't think OpenRouter is the right product for that... In the world of routing, which everybody does, Databricks does, Replit does it, Lovable does it, Vercel does it — it's a niche product with two really good niches." 00:22:42 — Rory O'Driscoll

"I don't even think this product will exist in five years. But I think there's a high chance, more than 51% chance it builds into a 20 or 30% revenue stream for Stripe." 00:00:11 — Rory O'Driscoll

M&A Land Rushes After a Landmark Deal Are a VC Myth

The common assumption is that a landmark acquisition like Cursor triggers a competitive scramble for the next-best assets. The reality, according to Rory, is the opposite: acquirers who missed the deal simply elevate their existing #2 target — which is dangerous for founders who assume being #2 automatically means a bidding war.

"My experience is that doesn't happen... Usually the other thing happens. I wanted to buy Cursor. I lost Cursor. Now it pushes it up my existing priority list. That's how number two and number three get bought. Not because there's a panic for land grab, but because I didn't get what I wanted." 00:16:16 — Rory O'Driscoll

SBC in Hyper-Growth AI Companies Is Economically Meaningless and Should Be Ignored

The standard analyst move of penalizing hyper-growth companies for massive stock-based compensation is intellectually wrong for companies like Anthropic, because the SBC reflects windfall appreciation on grants, not sustainable run-rate comp that would need to be replicated in cash.

"Someone who was hired with a million dollar package in 23 ended up making 51 million four years later. That doesn't mean you'd have to pay the next guy 51 million... The other 50 million is just dumb luck. You got lucky. It's not a run rate. I actually think it is OK in a hyper-growth company to look past a good slug of the SBC and normalize it out." 00:00:18 — Jason Lemkin

3. Companies Identified

Cursor

AI-native coding IDE, recently acquired by SpaceX for $60B all-stock. Mentioned as the defining acquisition of the AI era, and as a masterclass in pivoting through near-death experiences (Claude Code's launch) by going multi-model early. Revenue trajectory heading toward $6B by end of year.

"Cursor goes multi-model really early. It changes everything and rockets to a 60 billion outcome... And it went from something earlier in the show, it looked like it would be gross margin negative... It's a pretty darn good business model selling at 10x forward revenue. Elon was a shrewd buyer." 00:04:55 — Rory O'Driscoll

Anthropic

Foundation model company, first profit reported on $11.5B of Q2 revenue. On track for potential $200B+ ARR target. Mentioned as the company best positioned for an imminent IPO and as the enterprise AI leader.

"They did $4.5 billion last year... you go from $4.5 billion in a year to $10 billion in a quarter, that means you have $4 billion of gross margin, right? And that's literally in two quarters." 00:28:29 — Jason Lemkin

Stripe

Global payments infrastructure company. Acquiring OpenRouter for $7B to enter AI token routing. Also reportedly in talks for a PayPal acquisition. Highlighted as unusually good at M&A for a private company.

"Stripe actually appears to be very good at acquisitions. It's how it accelerated into crypto... If you're good at M&A and this is 5% of your market cap plus cash and you want it tomorrow, it makes sense." 00:20:48 — Rory O'Driscoll

OpenRouter

AI model routing infrastructure company, founded by Alex Atallah (also co-founder of OpenSea). Acquired by Stripe for ~$7B. Series B was at $1.3B valuation just four months prior. Praised as elegant and instantly usable software, but flagged as niche.

"I love OpenRouter. I talked about it on the show like six months ago. I'm a customer. I'm a user. It was brilliant. It was one of these pieces of software like ElevenLabs, which is just instantly easier to deploy. It's just elegant. It was just a beautiful piece of software." 00:21:16 — Rory O'Driscoll

SpaceX

Elon Musk's aerospace and AI company. Acquired Cursor. Now described as 80–90% an AI story despite its rocket and satellite origins. Has more compute than nearly anyone else via the Colossus cluster.

"He took SpaceX from literally a year ago being a really amazing rocket and satellite connectivity story to being... at least in terms of the future prospects, 80%, 90% an AI story." 00:12:01 — Jason Lemkin

Lovable

AI-powered app/website builder. Raised a new round from Menlo at $13.3B valuation with ~$600–700M ARR. Highlighted as having evolved from a weak product into a genuine platform with production-grade security and deep features.

"Lovable and Replit have become talent magnets... these layers are not impenetrable, but they start to get thick and crusty, this crust around them." 00:01:12 — Rory O'Driscoll

Higgsfield

AI video generation company. Raised at $5.5B valuation from DST at $700M ARR. Mentioned as employing some of the best mathematical talent globally (described as top mathematicians from Kazakhstan). Praised for its enterprise video market potential.

"Higgsfield is like the smartest mathematicians in Kazakhstan... I was one of the first 10 customers, I think. Now you can make a full-length motion picture... it's just so hard." 01:00:44 — Rory O'Driscoll

Workday

Enterprise HR/finance SaaS. Silver Lake reportedly circling at $43B take-private bid (5.3x revenues, ~16x EBITDA). Praised specifically for its closed ecosystem as a defense against AI disruption and agentic extraction.

"Workday is so closed, they've got a leg up, right? It is like LinkedIn, right? It is intentionally barely open... it also means you're going to capture more budget overall in your ecosystem than you would for others." 00:50:56 — Rory O'Driscoll

Databricks

Data and AI platform. Mentioned in context of the DOJ's Section 8 Clayton Act investigation of Andreessen Horowitz's overlapping board seats. Implicitly valued at ~$200B.

"If Martin Casado had to step off the board of Fivetran after exiting Cursor at 60 billion, it's probably OK, given their position in Databricks. We're sitting on $200 billion at Databricks. Fivetran, good luck. We'll switch to an observer seat." 01:08:33 — Rory O'Driscoll

Fivetran

Data pipeline company. Mentioned as the other company in the DOJ's Clayton Act investigation against Andreessen Horowitz.

"Andreessen's on the board of Databricks and they're also on the board of Fivetran, right? And now they're competitors. So now this has been percolating and now they're investigating." 01:05:55 — Jason Lemkin

Etched

AI chip company. Raised at $10B valuation roughly three weeks prior, then raised $700M at $21B from Jane Street, Kleiner, Sequoia, Andreessen — doubling valuation in four weeks.

"Literally like three weeks ago, Etched raised at 10 billion. Today, they've announced they've raised 700 million at 21 billion from Jane Street, Kleiner, Sequoia, Andreessen. Four weeks after, double the price." 01:03:03 — Harry Stebbings

Replit

AI coding platform. Mentioned alongside Lovable as having evolved from a weak product into a true platform and talent magnet, and as a model routing user proving OpenRouter's niche.

"You walk into Replit, I mean, these are the smartest people that Amjad could recruit for years." 01:01:12 — Rory O'Driscoll

Cognition

AI software engineering company. Flagged as the likely next M&A target now that Cursor has been taken off the table, particularly for acquirers like Microsoft who missed Cursor.

"Is there a knock-on effect for the second player in market, which I think arguably now would be Cognition?" 00:13:56 — Harry Stebbings

Rippling

HR/finance SaaS. Cited as a real-world example of enterprise model rationalization — down-speccing to just two models (Opus 4 and GPT medium) rather than routing across many, illustrating OpenRouter's niche limitations.

"Rippling posted their view as a B2B player of what models they use... they said the rest isn't worth it for Rippling today... they downspec to two at a time." 00:24:36 — Rory O'Driscoll

Scale AI

AI data labeling company. Mentioned as an analogy for how Stripe's OpenRouter acquisition might evolve — initially seeming niche but potentially becoming a major revenue stream over time.

"I think it'll be like the Scale acquisition. It will be the start of something that gets bigger." 00:00:11 — Rory O'Driscoll

Airtable

No-code database platform, under Bending Spoons ownership. Used as a contrast to Workday to illustrate how price hikes in less sticky, more open systems accelerate churn.

"A lot of folks are like, okay, I'm lifting off Airtable now. And they're like, Bending Spoons is going to raise my prices 3x, right?... Bending Spoons may lose 20% of Airtable's customers who finally spend a week lifting off of Airtable." 00:56:29 — Rory O'Driscoll

GitHub

Microsoft's developer platform. Mentioned as now a "trailing edge product" — a strategic loss for Microsoft that makes a coding AI acquisition existential.

"The fact that they've lost that connection with developers, that GitHub is now a trailing edge product, is to me over the medium term, a pretty significant loss." 00:15:00 — Jason Lemkin

ElevenLabs

AI voice/audio platform. Briefly mentioned alongside OpenRouter as an example of instantly elegant, easy-to-deploy software.

"It was one of these pieces of software like ElevenLabs, which is just instantly easier to deploy. It's just elegant." 00:21:16 — Rory O'Driscoll

Salesforce

CRM giant. Used as a contrast to Workday — praised for openness and platform extensibility, but flagged as more vulnerable to agentic substitution precisely because of that openness.

"Salesforce is out there working their frigging tails off because they are a much more open platform. You can build your own agents on top of Salesforce tomorrow." 00:49:58 — Rory O'Driscoll

PayPal

Payments company. Mentioned as a potential acquisition target for Stripe — a consolidation play that would add consumer wallets and dramatically increase Stripe's payments market share.

"They're talking about a PayPal acquisition, which is very much, we own this space already. Let's buy these guys, fold them into what we already have and just make a shit ton of money consolidating." 00:26:32 — Jason Lemkin

Ramp

Corporate spend management platform. Cited for data showing extreme dispersion in AI spend per employee: top 1% spending ~$7,000/month, median dramatically lower.

"You see the Ramp data that says the top 1% of their sample, which in turn, obviously, is a biased sample of tech forward people are spending $7,000. And then the median is spending like $100,000." 00:36:31 — Jason Lemkin

Wiz

Cloud security company. Dali Rajic, highlighted as a top CRO talent, came from Wiz. Used to illustrate the value of hiring sales leaders from hyper-competitive, technical companies over traditional SaaS backgrounds.

"If you step back for a minute, I'd rather have someone from Wiz that is close to technology, that is in a hyper-competitive space, rather than asking how many seats of Slack you want." 00:44:27 — Rory O'Driscoll

OpenAI

AI foundation model company. Discussed extensively in context of IPO timing, executive churn (departure of Denise Dresser as CRO), and competitive position vs. Anthropic.

"OpenAI is going to be able to sell stock next year at 1.8. People will do it... If you're the smaller market cap company and you have the bigger capital need, which right now OpenAI does because they have a more ambitious capital target." 00:42:00 — Jason Lemkin

Silver Lake

Private equity firm. Circling a $43B take-private bid for Workday. Discussed as a sophisticated financial buyer making a precise LBO bet on sticky SaaS cash flows.

"A very financially oriented, wildly savvy buyer is willing to bet money that they can buy this at a constrained price, lever it, and generate a return because the revenues are sticky enough to allow them to pay down the debt over five years." 00:45:22 — Jason Lemkin

UiPath

Automation software company. Mentioned briefly as an analogy for bringing in a "knobs and dials" operating co-CEO alongside a returning founder.

"He hired his successor when times were easy, just before AI. He brought in a great knobs and dials co-CEO, you know, like our friends at Daniel at UiPath." 00:47:57 — Rory O'Driscoll

Andreessen Horowitz (a16z)

Venture capital firm. Mentioned for winning big on Cursor and OpenRouter, and as the subject of a DOJ Section 8 Clayton Act investigation over overlapping board seats on Databricks and Fivetran.

"Well done, A16Z. Well done, Menlo." 00:20:04 — Jason Lemkin

Menlo Ventures

Venture capital firm. 12x return on OpenRouter (Series B at $1.3B, acquired at ~$7B). Also the existing investor who led the Lovable round at $13.3B.

"It's a 12x for Menlo and Andreessen." 00:17:37 — Harry Stebbings

Capital G

Google's growth equity fund. Mentioned as a winner in the OpenRouter acquisition.

"Well done, Capital G." 00:20:04 — Jason Lemkin

Thrive Capital

Growth equity firm. Mentioned as a winner in the Cursor acquisition; was part of the $50B round that was superseded by SpaceX's $60B bid.

"Neo, OpenAI's startup fund, is a mega winner. He invested $6 to $8 million very early... And many others, Thrive and Andreessen, most importantly, netting huge returns." 00:04:22 — Harry Stebbings

Jane Street

Quantitative trading firm. Participated in Etched's $700M raise at $21B, noted as a potential strategic customer relationship.

"They've announced they've raised 700 million at 21 billion from Jane Street... Jane Street wants to be a customer too or something." 01:03:03 — Harry Stebbings / Rory O'Driscoll

Bending Spoons

Mobile app/SaaS acquirer. Running the Airtable price hike playbook — cited as an example of the risk of aggressive monetization in less-sticky SaaS versus the Workday model.

"Bending Spoons is going to raise my prices 3x, right? Let me start doing it now... Bending Spoons may lose 20% of Airtable's customers who finally spend a week lifting off of Airtable, when they triple prices." 00:56:29 — Rory O'Driscoll

4. People Identified

Elon Musk

CEO of SpaceX, Tesla, xAI. Orchestrated the $60B Cursor acquisition in under a week, built the Colossus compute cluster from scratch. Described as possibly the most effective operator on the planet at industrialization and AI.

"He went from a standing start to owning more compute than pretty much anyone else and owning the most important product to fill that compute in less than a year, just over a year. That's just wildly effective management. It's world class in getting shit done." 00:12:01 — Jason Lemkin

Michael (Cursor CEO)

Co-founder and CEO of Cursor. ~24 years old at time of deal. Described as a paper decabillionaire who chose to sell to SpaceX over taking a $50B round, apparently motivated significantly by who he'd be working for.

"Imagine you are Michael at Cursor, right? And things are going pretty well. You've got a term sheet from, you're what, 24? You're a paper decabillionaire... I would much rather initially work for Elon than for Zuck personally." 00:12:37 — Rory O'Driscoll

Alex Atallah

CEO of OpenRouter, previously co-founder of OpenSea. Just interviewed by Harry Stebbings. Mentioned as the architect of OpenRouter's elegant developer experience.

"Alex Atallah, CEO, who I just had on the show, he founded OpenSea before. It's obviously the leading LLM routing company." 00:17:07 — Harry Stebbings

Dali Rajic

New CRO of OpenAI, replacing Denise Dresser. Came from Wiz. Described by Chad Peets as the best sales leader in the business.

"Dali Rajic has replaced her. For those that don't know Dali, he's one of the most respected CROs. He was a freaking master at Wiz. And I think the best CRO or sales leader in the business is Chad Peets, says he's the best of the best." 00:43:31 — Harry Stebbings

Dario Souza (Anthropic CEO)

Referenced in context of Anthropic's first profitability and Gavin Baker's comments about it being the "final private company."

"This is also in a week where Gavin Baker said about Dario saying he believes that they will be the final private company." 00:28:08 — Harry Stebbings

Anil (Workday CEO/Co-founder)

Co-founder who returned to run Workday. Highlighted as potential upside in the Silver Lake deal — a founder back in the saddle could create the agentic version of Workday and transform the base case.

"I think if he does, there's real upside to that. Maybe instead of their 20% IRR, it could be a game changer. If he creates the agentic version of Workday, they at least have the founder back in the saddle doing it." 00:48:25 — Rory O'Driscoll

Greg Brockman

OpenAI President. Mentioned as having taken over and brought in the Wiz hire (Dali Rajic), signaling a preference for technically-adjacent sales leadership over traditional enterprise SaaS pedigree.

"Greg Brockman took over, right? And brought in the Wiz guy. Just had enough of the Salesforce crap, right or wrong?" 00:43:58 — Rory O'Driscoll

Martin Casado

Andreessen Horowitz general partner. At the center of the DOJ Section 8 Clayton Act investigation for sitting on boards of both Databricks and Fivetran. Mentioned as someone whose position would be easy to resolve given Databricks' valuation.

"If Martin Casado had to step off the board of Fivetran after exiting Cursor at 60 billion, it's probably OK, given their position in Databricks." 01:08:33 — Rory O'Driscoll

Amjad Masad

CEO of Replit. Mentioned for his success in recruiting elite talent, turning Replit into a talent magnet.

"You walk into Replit, I mean, these are the smartest people that Amjad could recruit for years." 01:01:12 — Rory O'Driscoll

Scott Wu (Cognition CEO)

Mentioned by Harry as a candidate to be CEO of Microsoft (in a tweet he cited), framed as one of the best technical product leaders in AI.

"I've got a fantastic tweet that said Satya should buy it and then make Scott Wu CEO of Microsoft... I actually thought Scott would be a rather brilliant CEO of Microsoft." 00:15:55 — Harry Stebbings

Annie Partovia

Investor (Neo / OpenAI Startup Fund). Mentioned as a major winner in the Cursor acquisition, having invested $6–8M very early.

"Minting 1,000x returns for the likes of Annie Partovia. Neo, OpenAI's startup fund, is a mega winner. He invested $6 to $8 million very early." 00:04:22 — Harry Stebbings

Chad Peets

Sales leader / commentator. Cited as having called Dali Rajic the best CRO in the business.

"I think the best CRO or sales leader in the business is Chad Peets, says he's the best of the best." 00:43:31 — Harry Stebbings

Noah Smith

Economics writer/blogger. Cited for writing a piece saying "only a fool denies that Elon Musk is wildly effective," notable because Smith is described as center-left and not an Elon fan.

"I was reading Noah Smith, who's kind of a damn left to center, but moderate centrist blogger, who's not an Elon fan, just wrote a great piece about a year ago that says, only a fool denies that Elon Musk is wildly effective." 00:11:32 — Jason Lemkin

Satya Nadella

Microsoft CEO. Discussed as the person who most should have acquired Cursor given Microsoft's existential developer positioning challenge, but who may have missed the window.

"I thought of you when you said before about Elon Musk giving Sam billions of dollars through gritted teeth... Satya should buy it and then make Scott Wu CEO of Microsoft." 00:04:22 — Harry Stebbings

Gavin Baker

Investor. Mentioned for publicly stating his belief that Anthropic will be the final private company.

"Gavin Baker said about Dario saying he believes that they will be the final private company." 00:28:08 — Harry Stebbings

5. Operating Insights

Cap AI Spend Per Engineer and Force Trade-offs Against Headcount

Rather than letting AI spend grow unchecked, the emerging CFO playbook is to explicitly budget $100K of inference per engineer and simultaneously reduce dev team headcount 30–40%. This creates accountability, prevents runaway token costs, and turns AI from an expense into a measurable productivity swap.

"The last 60 days are every single scale up is capping their AI budget for real... I think we'll give each of our best engineers $100,000 of tokens. And in return, we'll cut the size of our dev teams 30%, 40% effectively." 00:35:25 — Rory O'Driscoll

Run Your Core Systems Headless with an Agent Layer

Rather than logging into Salesforce (or any system of record) directly, Rory's firm runs it entirely via an internal AI agent that interfaces with Salesforce under the hood — dramatically increasing utility while enabling connection to any data source, workflow, or competitor tool.

"We run Salesforce entirely headless. So we have our own agent, 10K, our own AI VP of revenue. It runs Salesforce under the hood. Pro is it makes Salesforce much more powerful than it ever was. Like I didn't log into Salesforce for seven years. Now I log in every day because I have an agent." 00:53:17 — Rory O'Driscoll

Correlation AI Spend to Output Velocity — Then Shame the Laggards

Jason describes running a portfolio-wide survey on AI adoption and spend per head. The actionable next step he identified on the show: cross-reference spend data with output velocity (e.g., roadmap compression) to prove the ROI, and use that data to pressure underperforming teams to accelerate.

"What I didn't do, and actually now that I think about it, I should do, and I will do, is go back and see if you could find a strong correlation between output and spend, you know, can you justify the spend? Then you should be saying to the laggards, you're just going to fall behind." 00:39:19 — Jason Lemkin

In M&A, Be the Buyer Who Moves in Days, Not Months

The Cursor deal illustrates a precise M&A tactic: when you want an asset, enumerate all objections and remove them in one conversation. SpaceX bid $10B above the existing term sheet and added a $10B reverse breakup fee. Deliberate, fast, decisive.

"What does it take? So he bid $10 billion more, right? You've got a deal at $50. I mean, Zuck did Instagram and WhatsApp like in an hour on the back of a napkin and paid high, right? But I think Elon did even better. What's it going to take? I've removed all the objections from the deal." 00:10:16 — Rory O'Driscoll

Founder Brand Matters More Than Price in M&A — Manage It Deliberately

A frequently overlooked M&A lever: founders choose acquirers partly on who they want to work for. The emotional appeal of working for Elon vs. Zuck materially affected the Cursor outcome. Acquirers should curate their employer brand with founders the same way they curate deal terms.

"It is incredibly emotionally important to founders to land in something they want to land in. And I would not want to land at Meta today... I'll take Elon over that one." 00:13:36 — Rory O'Driscoll

6. Overlooked Insights

The Amazon / AWS Inference Revenue Windfall From Claude Code Is Already Happening — and It's Underappreciated

The panel briefly mentioned, almost in passing, that Amazon is capturing a huge slice of AI inference revenue from Anthropic's Claude Code without owning the model — simply by being the cloud provider for inference. This is structurally significant: AWS is effectively a silent winner every time a developer uses Claude Code, with no model risk, no training cost, and no competitive exposure.

"For Amazon, you know, they're in the AWS business. They got lots of contracts with Anthropic. They basically have the compute for Claude Code. So they're basically getting the inference side revenue for that. They don't own the model, but it's important, but not imperative." 00:14:04 — Jason Lemkin

This one sentence buries a major investment thesis: cloud providers who lock in foundation model inference contracts are extracting enormous, low-risk revenue from the AI boom without taking model risk. The market has not fully priced AWS as a direct beneficiary of Anthropic's growth at the inference layer — separate from its general cloud revenue.

The Real Existential Threat to Microsoft Is Already Happening — and No One Is Panicking About It

Microsoft owns GitHub, the world's largest developer platform. The panel noted, with almost no emphasis, that GitHub is now a "trailing edge product" and that Microsoft has "lost the connection with developers" — the most strategic demographic for software's next era.

"The fact that they've lost that connection with developers, that GitHub is now a trailing edge product, is to me over the medium term, a pretty significant loss... if you wanted to name people who should want to own a leading state-of-the-art coding product in this brave new world, clearly the number one person is Microsoft." 00:15:00 — Jason Lemkin

Microsoft's market cap still reflects its status as a developer-beloved company via GitHub and VS Code. If the developer mindshare — the foundation of its cloud and tooling flywheel — is genuinely shifting to Cursor, Lovable, and Replit, the damage to Microsoft's long-term enterprise software position is structurally underpriced. The Cursor acquisition by SpaceX rather than Microsoft may be the moment historians point to as when Microsoft permanently lost developer primacy.