May 14, 2026 · 1h 19m · news
Anthropic Buys Compute From Elon & Commits $200BN to Google | Cerebras IPO | Ramp Raises at $40BN · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode, host Harry Stebbings, along with guests Jason Lemkin and Rory O'Driscoll, analyze the shifting landscape of tech, tracking Anthropic's massive infrastructure deals, the highly anticipated Cerebras IPO, and the intense market realities reshaping public SaaS, fintech, and founder psychology.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 10.7% of the talking time here. How this is scored →
Harry as informed peer 3.4 Guest teaching 5.3 Guest disagreement 2.9 Harry pushing back 2.2
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 53:31 Rory calls out host's unciteable claim
Rory directly refutes Harry's claim about Benchmark holding twenty percent of Cerebras by citing the S-1 filing and mocking Harry for relying on Twitter chatter.
Hardest push from Harry ▶ 22:02 Host counters guest premise on token wasteHarry explicitly challenges Jason's claim that token usage is being wasted by web developers by quoting Atlassian CEO Mike Cannon-Brooks on infinite software demand.
Biggest teaching moment ▶ 53:31 Rory corrects host using primary SEC filing dataRory schools Harry on cap table facts, replacing Harry's social media rumor of twenty percent ownership with actual S-1 data showing eight to nine percent.
Harry holds his own ▶ 22:02 Host deploys industry CEO quote to challenge guestHarry demonstrates domain tracking by deploying Atlassian CEO Mike Cannon-Brooks' exact perspective on infinite feature demand to push back against Jason.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Restrictions on Anthropic Secondary Sales | 3 | 6 | 5 | 2 | Harry introduces Anthropic secondary market restrictions, but Jason reframes the topic as a social media nothing burger while Rory explains cap table mechanics. | |
| Anthropic's SpaceX Compute Deal | 2 | 5 | 2 | 1 | Harry prompts the guests regarding Anthropic's compute deal with Elon Musk, allowing Rory and Jason to elaborate on data center capacity and business unit P&Ls. | |
| Anthropic's $200B Google Cloud Commitment | 3 | 6 | 2 | 2 | Harry frames Google's circular economy positioning, prompting Rory to break down how Anthropic represents forty percent of Google's cloud backlog. | |
| AI Agent Token Consumption and Goldman Sachs Report | 3 | 5 | 3 | 2 | Harry asks about a Goldman Sachs report on token growth, leading Jason to challenge the conservative twenty-four times estimate in favor of parallel agent usage. | |
| Token Supply, Engineering Quality, and Goodhart's Law | 6 | 5 | 5 | 6 | Harry pushes back against Jason's argument about developer token waste by quoting Atlassian's CEO regarding infinite demand for software features. | |
| Vertical AI Agents and Platform App Layer Dynamics | 5 | 6 | 4 | 5 | Harry argues that key vertical agent markets like legal and support are winnable by model providers, while Jason and Rory contend app layer software remains defensible. | |
| SaaS Public Market Earnings: Monday vs. HubSpot | 3 | 5 | 2 | 1 | Harry introduces SaaS earnings dynamics, leading Jason and Rory to compare Monday's guidance bump with HubSpot and Cloudflare's market reactions. | |
| ZoomInfo Case Study and Clay's Disruption | 3 | 5 | 3 | 2 | Jason presents ZoomInfo as a case study of AI stealing growth via Clay, and Rory clarifies how waterfall data architectures commoditize pure data providers. | |
| Cerebras IPO & Early-Stage Venture Capital Incubations | 4 | 8 | 5 | 3 | Rory directly corrects Harry's assertion regarding Benchmark's Cerebras ownership percentage, pointing out actual S-1 data over social media rumors. | |
| Founder Greatness and Andrew Feldman's Journey | 2 | 4 | 1 | 0 | Jason and Rory reflect on founder greatness and Andrew Feldman's long journey to navigate early market timing with Cerebras. | |
| Fintech Multiples: Ramp's $40B Valuation & Parker's Collapse | 3 | 5 | 2 | 2 | Harry brings up fintech valuations, prompting Rory to break down Ramp's forty times revenue multiple relative to public comparisons. | |
| Lime's IPO Prep and the Musk vs. Altman Trial | 3 | 4 | 1 | 1 | Harry transitions through Lime's IPO prep, the OpenAI trial, and SaaStr format changes where Jason explains why live demos are replacing podcast-style speakers. | |
| DRAM Stocks and the AI CapEx Boom Duration | 4 | 5 | 3 | 2 | Harry asks about DRAM stocks and discusses founder sacrifice, leading Jason to articulate how years of startup intensity permanently rewires a founder's brain. |
Statements from this episode (32)
Lemkin: Software without a role in an agentic world faces terminal decay
“There are categories of software where if they don't have a reason to exist in an agentic world, they will go into a terminal state of decay.”
Lemkin: ZoomInfo's market growth was stolen by Clay and competing tools
“Zoom Info's growth was stolen from it from Clay and friends, and it's a brutal case study.”
O'Driscoll: Software valuations will never reach 20x revenue multiples again
“Give me 30% growth, give me profits, give me a story that's got some future in it, and I'll get you back to five times. I'll never give you 20 times again. I won't fall for that one this time.”
Lemkin: Anthropic secondary sales restriction is an overblown social media non-issue
“To me, this story seems like a nothing burger made up on social media.”
Lemkin: Standard startup charters have required board transfer approvals since 2020
“Maybe four or five years ago, every set of charter documents say no transfer without the board permission, just like anthropic and open eye.”
Lemkin: Anthropic publicly warned Menlo Ventures against SPVs last year
“Anthropic last year warned Menlo, one of its best investors and others, enough with the SPVs, they said publicly, enough with the damn SPVs.”
O'Driscoll: xAI data center was only 11% utilized
“I believe the data center was 11% utilized, right?”
O'Driscoll: SpaceX revenue run rate is around $20 billion
“For context, the total SpaceX revenue run rates around twenty billion.”
Lemkin: Anthropic has secured more compute capacity than OpenAI
“And now Anthropik, I hate the five D eight D chess metaphor, but now it's gonna, it's figured out a way to have more capacity than open AI.”
Stebbings: Anthropic committed $200B to Google Cloud over five years
“They also did a deal committing two hundred billion to Google over, I believe it was a five-year period.”
O'Driscoll: Anthropic deal makes up 40% of Google's future backlog
“I think now the anthropic rev commit is about 40% of Google's total future backlog.”
O'Driscoll: Long-term AI value will accrete primarily to model providers
“I think, and I could be wrong, I think in the end the value will accrete mainly to the model providers and everyone down the stack that's selling to them, even though they're all making out like bandits today, starting with the memory guys all the way up to th…”
O'Driscoll: Microsoft does not have an AI model worth a damn
“And in Microsoft case, I really wish I even had a model that was worth a damn, which I don't.”
Lemkin: WSJ data shows Claude enterprise share jumped to 48%
“The wall street journal today published the market shares in the enterprise for open AI, Claude, Gemini and Grok. Grok is a rounding error going to the prior conversation, right? It's not making any progress. But everything is so multimodal that it said Gemini…”
O'Driscoll: AI token output per dollar increases 10x every two years
“Every 18 months, you know, the raw performance of the chip gets, you know, roughly three X faster. Then on top of that, other optimizations, how they run LLMs, how they do all the quantization, all the other clever stuff that, you know, you can read about and …”
Lemkin: Companies will clamp down on token waste within one year
“In a year from now, we're not gonna be wasting tokens on mediocre web developers playing with stuff, and we're gonna clamp down on it because it's a huge waste.”
Lemkin: Anthropic will not build end-user apps in legal or CX
“I don't think there's any chance that Anthropic is going to do applications in CX and actually, and I could be wrong. I will bet you a lot of money. They're not going to go all the way in legal at the app app.”
Lemkin: AI legal startup Harvey costs $150k annually per law firm
“Now what does Harvey cost on average? 150,000 dollars a year per law firm.”
Lemkin: Traditional marketing automation software obsolete within one year
“I don't think in a year we're going to need any traditional marketing automation software. It's too dated. It doesn't work for agents marketing.”
Lemkin: Public markets fear terminal value is zero for decelerating SaaS
“But I think even though the SaaS apocalypse is behind us I think for many, there is still a worry, the terminal value is zero. I don't think that fear has gone away. If you're, if you keep decelerating, While budget is accelerating everywhere around you.”
O'Driscoll: High-multiple tech stocks remain vulnerable during paradigm shifts
“And I think what you saw in the case of AppLoving and Cloudflare is it turns out high-priced stocks At the start of a paradigm shift, no matter how amazing the quarter is, are just vulnerable to disruption.”
Lemkin: ZoomInfo trades at 1x revenue with 35% operating margin
“It's trading at one time revenue with 35% adjusted operating income.”
O'Driscoll: Cerebras IPO will trade amazingly well in the short term
“Look, it's going to go great. The fact that they've raised the range, which you only do when you're highly confident, especially that much, that you've got a killer IPO on your hands. There's no more new information. It's going to go out and it's going to trad…”
O'Driscoll: Benchmark, Foundation, and Eclipse each hold 8–9% of Cerebras
“Ah, you just, because you see, Harry, you just listen to Twitter, but if you go and you actually look up the S-one, they have eight or nine percent ownership, right? Because I do these things, because every VC reads an S-one the same way, you read the front pa…”
Lemkin: Foundation Capital did true VC work by incubating Cerebras
“For the first time in a long time, I was kind of jealous of foundation because they did the hard work. They, Steve and the team incubated this company.”
Lemkin: Real VC is early conviction, not brand-leveraged Series B entry
“This is the job of early stage investing is what foundation did. This is not using Mark Andreessen's brand to muscle into the B.”
Lemkin: SaaStr will switch from Brex to Ramp if Ramp's AI agents win
“Certainly we're going to rebuild our financial stack after Sastra annual and we will move from Brex to ramp if it is the most agent friendly and automating procurement is a huge bonus.”
O'Driscoll: Ramp's 40x revenue multiple sharply contrasts Brex's 6x valuation
“Fortyx runway revenues when the comp traded at six in terms of breaks, even albeit on a lower growth rate is a pretty healthy valuation.”
O'Driscoll: OpenAI will emerge from Musk lawsuit with deal intact
“My gut continues to be that even though everyone will look crappy that, you know, open AI get escape with their deal intact.”
Lemkin: Fireside chats and speakers are dead at tech conferences
“So I'll have to see, but we won't do, we'll never do Unless I'm forced to, I mean, if, you know, I do it for Sam Altman, but otherwise I'm not gonna do, there's, the firesides are dead and speakers are dead. There's just no point when podcasts are better.”
Lemkin: VCs never make real money investing in 'nice' founders
“Yeah, and every time you cut a corner, every time you invest in a founder that's really nice and really hardworking and really good guy, you never make any real money.”
Lemkin: Founders should accept buyouts unless they refuse $1 billion
“That's why I tell all founders to take the offer. I tell all founders to take it. The opposite of VCs, take it. It's, it is an intensity test. And if they push back and say, no, a billion is not enough. Great. Then go for it. But if they're not sure you may no…”
Shorts cut from this episode
▶ The Problem With Token Quotas · 20VC with Harry Stebbings (@23:15)
▶ The End of Traditional Software? · 20VC with Harry Stebbings (@0:31)
▶ 24x token consumption by 2030? · 20VC with Harry Stebbings (@16:43)
▶ Any normal human would sell · 20VC with Harry Stebbings (@1:15:28)
▶ Anthropic, Elon & AI’s Biggest Week · 20VC with Harry Stebbi (@0:00)