20VC x Saa Str: The AI Power Grab — Anthropic Blinks, Open AI Raises $110B, and Every CEO I Know Wants to Cut 40% of Their Team | Saa Str

Overview

Free e Books

	e Book: Hiring a Great VP of Sales
	e Book: Raising Capital
	e Book:  The First $1m ARR

University All Posts University Podcasts The Top CROs VC Fundraising Top Videos Q&A Best of Saa Str #1 Bestselling Book Search Everything Join the Community

Details

Free e Books

	e Book: Hiring a Great VP of Sales
	e Book: Raising Capital
	e Book:  The First $1m ARR

London 2025 Annual 2026 Events Overview Sponsors

	Event Sponsorship
	Media Sponsorship

Digital AI Day 2025 (Free) Speaker Submissions Speaker Requirements Overview

20VC x Saa Str: The AI Power Grab — Anthropic Blinks, Open AI Raises $110B, and Every CEO I Know Wants to Cut 40% of Their Team

Every week, Harry, Rory, and Jason sit down together to break down the biggest stories in AI and B2B.

Anthropic picks a fight with the Pentagon and loses. At least for the moment.

Open AI closes $110 billion privately — four times the size of the largest IPO in history.

Cursor doubles from

And underneath all of it: the B2B software transition is worse than we thought, and different than we thought.

Here’s the full breakdown — with all three perspectives, because they don’t all agree.

Anthropic vs. The Pentagon: A Three-Way Disagreement

Anthropic had a $200M contract with the Department of Defense. Negotiations broke down over two clauses Anthropic wanted to impose: no mass surveillance, no autonomous weapons. The Pentagon’s counter: we want to do anything that’s legal. That gap didn’t close. The Pentagon walked — and threatened to declare Anthropic a supply chain risk, which is anywhere from “mild contract loss” to something approaching thermonuclear, depending on how far the administration decides to push it.

Rory’s take: Dario was right to walk, but naive to have entered the deal at all.

Anthropic’s entire organizing principle — the “we’re the ones who can be trusted with AI” thesis — is what has kept all seven founders together while Open AI lost nearly all of theirs. That’s not an accident. Blowing that up for a

The analogy Rory reaches for is the Manhattan Project. General Groves humored the scientists, gave them what they needed, let them skip uniforms. And they weren’t in the room when the decision was made to use the bomb. That’s how state power and private expertise have always interacted.

Jason’s take: He had no choice — his team would have quit.

The real reason Dario walked wasn’t just personal principle. It’s that he couldn’t face his team and say “at least we’ll make autonomous weapons safer than XAI would.” That framing stopped working. You can only tell researchers “we’re making this safer by being the ones building it” for so long before you can’t say it with a straight face anymore. Sam Altman proved the point — he took the deal when Anthropic walked, and his team immediately revolted. He had to publicly backtrack on the contract terms the next day.

At the frontier AI labs right now, labor has extraordinary power. Researchers walk away from eight figures in unvested stock because they want to work somewhere else. This is unprecedented. Whatever management thinks, the team decides. Dario knew that. He made the calculation that preserved his company.

Where both converge: The state is more powerful than Anthropic, and Anthropic is just discovering this. The Defense Production Act is real. Supply chain risk designations are real. You can probably beat them in court eventually — but that is a brutal way to run a company. As Rory put it plainly: “AI might be scary in the theoretical. The state is scary in the real sense — we have laws, we have men with guns to enforce those laws, and we can take your company if we want to.”

As an Anthropic shareholder right now: same expected return, wider variance. That is categorically worse.

Harry asked the right question: is there any money left in the world to fund these businesses?

Sam Altman closed $110B. Give him full credit — the man is an extraordinary capital raiser and this is a legitimate force of nature. But Rory drilled into the structure, and the structure is interesting.

Amazon committed

Harry’s read on where this goes: “You’ve got to believe the next round for Open AI, Anthropic, and Space X are all public offerings.” The private capital pool is essentially exhausted. Soft Bank is in. Amazon is in. Nvidia is in. The sovereign wealth funds are in. There isn’t another round at this scale available privately. The next round is an IPO. Harry’s calling Open AI at $1.5T by October.

Rory on the valuation math: Open AI is at least understandable — it’s growing explosively and you’re paying 40-50x revenues on a hypergrowth company, which has precedent. Space X is harder. The $18B Starlink business growing at 20% would be worth maybe 5x revenues as a normal company. You’re paying 100x because you’re underwriting Starship, direct-to-cellular Starlink, and orbital data centers all at once. That’s a lot of next-generation technical risk in a single bet, even if the person executing that risk has the best track record on the planet.

Harry on the founder premium question: Remove the person and see what happens to the valuation. Tesla at

Jason came in with a strong view: almost every public B2B software company is going to miss numbers for the rest of the year, and it’s worse than it looks. Rory pushed back — not on the conclusion, but on the mechanism. And the mechanism is what matters.

Rory’s framing of what actually happened: For 15 years, public software companies averaged 30% growth at 6x revenues. COVID: two years of 40% growth at 20x. Then deceleration — not back to 30%, but down to 15%. And the markets, weirdly, kept giving premium multiples anyway. The revenue multiple chart looked like normalization. But the growth had fallen out of the bottom and nobody said it out loud. Then the market woke up and said: this isn’t temporary. It’s permanent. AI is structurally changing the topline trajectory. We’re repricing from 6x revenues (which used to mean 30% growth) down to 4x revenues (which means 15% growth, permanently). Mongo dropped 25% after a strong quarter because they guided to the low 20s and were trading at 40x EBITDA. That math simply doesn’t work anymore.

Jason’s deeper point — the one that stings: The amount of enterprise spending that has flowed to cloud and LLMs since 2023 is unlike anything we’ve ever seen. Public B2B software companies captured almost none of it. They’ve been sitting on their ARR, taking 6-8% price increases, rolling out beta AI connectors, and hoping their customers wouldn’t notice. You can smell the lack of belief through the earnings calls. When Marc Benioff talks about Agentforce he believes it — he’s been in the building, he’s seen the deployments, he’s all in. Most of the other public software CEOs don’t believe their own AI story. It shows.

Rory’s counterpoint: It’s genuinely hard to do what Intercom did. Owen came back as CEO, gutted the product architecture, bet everything on Finn, and spent a year where customers were asking why he wasn’t focused on the real business. That takes a specific kind of founder bloodymindedness that most professional managers at public companies simply don’t have. It’s not that they ignored AI. It’s that the organizational will to make those changes — bet-the-company changes — is extraordinarily rare.

Where all three land: Salesforce and Shopify are doing the work. Most others are heading toward what Jason calls the event horizon: a state of terminal decline where the ideas aren’t there, the team doesn’t want to do the work, and the only remaining move is to cut the team and call it AI efficiency.

Harry’s framing: Block lays off 40% — is it genuinely AI efficiency, or a bloated organization that needed resizing, with AI as the mask?

Both Rory and Jason essentially agreed: it’s both, but the AI framing is mostly cover.

Block’s top line is growing at roughly 3%. You only lead a press release with “gross profit up 27%” when you can’t lead with revenue growth. They’ve given up on returning to growth. And once you give up on growth with a cost structure built for 50% growth years, the math has one answer: cut, get profitable, make the stock move.

Jason’s call: “Every single CEO I talk to doesn’t think they need 40% of their team.” The Block playbook will become the default for companies that have given up on growth.

Cursor’s $2B: How the VC Echo Chamber Got It Wrong

Harry opened this one with genuine confusion: everyone he talked to had moved off Cursor to Claude Code. Then Cursor announced they’d doubled from

Jason’s honest answer: We generalized from our portfolios. His fastest-growing companies had moved to Claude Code. One board meeting had a CTO joking that Cursor was the “grandpa tool.” That narrative was real — for a specific cohort of technical founders at cutting-edge AI startups. It was not representative of how most of the enterprise market buys software.

Rory’s explanation: 60% of Cursor’s revenue is enterprise. Banks. Regulated industries. Large organizations that need SSO, role-based access controls, no data retention, audit logs, legal review, and purchase orders before a tool can be deployed at scale. These organizations are not doing a 48-hour tool eval and switching. Barclays just rolled out Cursor as the first approved agentic coding tool in the bank. That matters. Banks have enormous budgets and extreme conservatism — being first-approved at Barclays is worth more in durable ARR than a thousand Twitter takes about which tool is technically superior this week.

Rory’s Azure parallel is sharp: Azure held on for 20 years against AWS partly on the strength of “Microsoft has a relationship with your CTO.” Enterprise software has always rewarded trusted and integrated over technically optimal. That doesn’t change because AI is faster.

On the competitive dynamics: Jason’s point is that in a TAM this large, growing this fast, you can be losing marginal share and still grow explosively. As Rory framed it: “The knife fight doesn’t start until the TAM is 60-70% saturated.” AI coding tools are nowhere near saturation. Cursor and Claude Code can both explode right now without truly competing. One note worth watching: 60-70% of Cursor’s model calls still run through Anthropic’s API. Margins are real but complicated. And Claude Code will keep building enterprise features — not to target Cursor specifically, but because that’s what you build. The space shrinks. Cursor just has to keep expanding faster.

Rory’s closing line on this: “The prize for winning is to reinvent the company from scratch and the product from scratch every six to nine months. Congratulations. It’s a fun game.” Cursor went from autocomplete → IDE → agent → swarms of autonomous agents. That’s the model. If your product story isn’t fundamentally different every six to nine months, you’re falling behind. The prize for winning each round is just that you get to play the next one.

Labor vs. Capital vs. the State. At the frontier labs, labor holds all the cards — researchers leave eight-figure unvested packages because they can. At Block, capital has all the power and labor is fungible or unwanted. And in between: the state, which has the monopoly on legitimate force and is just starting to flex it in this industry. Anthropic learned that lesson the hard way this week.

Legacy B2B vs. AI-native. The Saa S model — brute-force revenue growth through human sales and marketing, high NRR, multi-year contracts — is transitioning. Not dying overnight, but transitioning. The companies with the stomach to rebuild from the product up will come through. The others will cut 40% and call it AI efficiency and hope the market forgives them.

Private vs. Public capital. The private capital pool for frontier AI is essentially exhausted. The next rounds for Open AI, Anthropic, and Space X are public. That matters enormously for how the rest of the market prices itself — both the AI leaders and the B2B companies trying to survive the transition.

The opportunity is enormous and the rate of change is accelerating. What Amelia built in six months at Saa Str — a fully autonomous AI VP of Marketing running our entire operation — would have been science fiction 24 months ago. What gets built in the next 12 months will make that look simple.

But the prize for winning is just that you get to reinvent the company from scratch and do it again.

“You’ve got to believe the next round for Open AI, Anthropic, and Space X are all public offerings.”

“You’ve got to believe the next round for Open AI, Anthropic, and Space X are all public offerings.”

“Tesla trades at a trillion today. I think if Elon died tomorrow, it’d trade at

“Tesla trades at a trillion today. I think if Elon died tomorrow, it’d trade at

“Block lays off 40% — is it truly AI efficiency, or a bloated organization that needed resizing, with a mask of AI?”

“Block lays off 40% — is it truly AI efficiency, or a bloated organization that needed resizing, with a mask of AI?”

“Whatever it is that’s in the water at Anthropic, it’s working — and it has created unity.”

“Whatever it is that’s in the water at Anthropic, it’s working — and it has created unity.”

“The knife fight doesn’t start until the TAM is 60-70% saturated. Until then, let’s all grow — one big happy family.”

“The knife fight doesn’t start until the TAM is 60-70% saturated. Until then, let’s all grow — one big happy family.”

“The prize for winning is to reinvent the company from scratch and the product from scratch every six to nine months. Congratulations. It’s a fun game.”

“The prize for winning is to reinvent the company from scratch and the product from scratch every six to nine months. Congratulations. It’s a fun game.”

“Every single CEO I talk to doesn’t think they need 40% of their team.”

“Every single CEO I talk to doesn’t think they need 40% of their team.”

“Vibe coding is one of the most minor threats to software companies in our lifetimes. I vibe code constantly. The real threat is they’ve lost the way to growth.”

“Vibe coding is one of the most minor threats to software companies in our lifetimes. I vibe code constantly. The real threat is they’ve lost the way to growth.”

“Three kids from Stanford figured out how to use Opus. Why couldn’t you? You’ve had 16 months.”

“Three kids from Stanford figured out how to use Opus. Why couldn’t you? You’ve had 16 months.”

Anthropic May Never Catch Open AI. But It's Already 40% as Big.

The 6 Metrics in Open AI's New Enterprise AI Report Worth Knowing — And Why Most of It Is Noise

Anthropic May Never Catch Open AI. But It's Already 40% as Big.

The 6 Metrics in Open AI's New Enterprise AI Report Worth Knowing — And Why Most of It Is Noise

Industry News

Get from

Key Takeaways

• Free e Books

    e Book: Hiring a Great VP of Sales
    e Book: Raising Capital
    e Book:  The First $1m ARR
  

• University All Posts University Podcasts The Top CROs VC Fundraising Top Videos Q&A Best of Saa Str #1 Bestselling Book Search Everything Join the Community

• Free e Books

    e Book: Hiring a Great VP of Sales
    e Book: Raising Capital
    e Book:  The First $1m ARR
  

• London 2025 Annual 2026 Events Overview Sponsors

    Event Sponsorship
    Media Sponsorship
  

• Digital AI Day 2025 (Free) Speaker Submissions Speaker Requirements Overview