Transcript
SPEAKER_1: Anthropic is reportedly raising fifty billion dollars at a nine hundred fifty billion dollar post-money valuation. That would make it the highest-valued private company in history, by a significant margin. SPEAKER_2: And what makes that number even more striking is the trajectory. When Anthropic last raised at a sixty-one point five billion dollar valuation back in twenty twenty-four, its ARR was under one billion dollars. It's since grown revenue more than forty times in roughly eighteen months. SPEAKER_1: Forty times in eighteen months. That's an extraordinary growth rate. What's actually driving that acceleration? SPEAKER_2: Two things, really. First, API consumption. Enterprise developers are building production workloads on Claude, and consumption compounds as those workloads scale. Second, Claude for Work, which is the teams and enterprise tier. Those are seat-based contracts with high retention, and the move from pilot to production happened faster than most analysts expected. SPEAKER_1: And Ramp's enterprise AI spending data shows Anthropic capturing around seventy-three percent of first-time enterprise AI buyers. That's a dramatic shift from what was nearly an even split with OpenAI earlier this year. SPEAKER_2: Right. Claude's performance on coding and reasoning benchmarks, combined with longer context windows, is converting buyers who started on ChatGPT and moved after hitting capability ceilings. That win rate is the core of the bull case at this valuation. SPEAKER_1: So what does the structure of a fifty billion dollar raise actually look like? This isn't a traditional Series D. SPEAKER_2: No, it's not. Sources indicate it involves a mix of primary capital and secondary liquidity, with sovereign wealth funds and the largest institutional LPs as anchor investors. At nine hundred billion pre-money, Anthropic would be valued higher than most S&P five hundred companies at IPO. SPEAKER_1: The only real comps mentioned are late-stage OpenAI and the peak twenty twenty-one valuations of SpaceX and Stripe, and those were in the hundreds of billions, not approaching a trillion. SPEAKER_2: Exactly. And the round is expected to close before the end of Q two. If it closes at those terms, the post-money cap exceeds nine hundred fifty billion dollars. SPEAKER_1: So why does a model company need fifty billion dollars in the first place? SPEAKER_2: Compute. Training next-generation models requires datacenters that cost billions to build, forward contracts with chip manufacturers running years out, and the ability to absorb that capital expenditure before the revenue catches up. OpenAI has Microsoft. Google has its own infrastructure. Anthropic needs to own its compute destiny. SPEAKER_1: And the founders you talk to are apparently treating Anthropic-scale capital as its own category now. SPEAKER_2: That's right. The AI labs aren't competing with startups anymore. They're competing with sovereign nations for GPU allocation. That's the context you need to understand why this raise is structured the way it is. SPEAKER_1: And the central question for LPs taking that bet at nine hundred fifty billion post-money is whether Anthropic can maintain its enterprise win rate when every major cloud provider is shipping their own frontier model. SPEAKER_2: That's exactly it. Forty-five billion in ARR growing forty times in eighteen months at a nine hundred billion valuation is either the most justified price tag in startup history or the biggest bubble since twenty twenty-one. That one question is what separates those two outcomes. SPEAKER_1: Let's shift to NVIDIA. The company holds over forty billion dollars in equity positions across AI startups, specifically the companies it supplies chips to. And Jensen Huang designed this intentionally. SPEAKER_2: He did. The portfolio reflects NVIDIA's strategic priorities precisely. There are positions in AI infrastructure companies like CoreWeave and Lambda Labs, which rent out NVIDIA hardware and need to keep buying more to scale. There are stakes in AI labs and model companies like Inflection and Mistral. SPEAKER_1: And it extends into robotics and physical AI as well, with positions in Figure and one X Technologies, companies building on Isaac and Omniverse. Plus autonomous systems like Waymo and Aurora, which depend on NVIDIA's DRIVE platform. SPEAKER_2: The unifying logic across all of it is that every equity investment is in a company that needs more NVIDIA hardware to grow. So the portfolio appreciates in direct proportion to AI adoption. SPEAKER_1: Walk me through how the flywheel actually works mechanically. SPEAKER_2: NVIDIA sells chips to Company X. Company X raises a Series C at a higher valuation. NVIDIA's equity stake is now worth more. NVIDIA uses that appreciation to fund R&D for its next chip generation. The next chip is faster, so Company X buys more of it. Then you repeat the whole cycle. SPEAKER_1: And what makes this structurally different from typical corporate venture? SPEAKER_2: The lock-in effect. A portfolio company that has built on NVIDIA's stack, meaning CUDA, NIM microservices, the full software layer, faces switching costs that compound with time. By the time they're at a ten billion dollar valuation, rebuilding on AMD or custom silicon is a multi-year migration with no guaranteed outcome. SPEAKER_1: Now let's cover the other major moves from this week. OpenAI had a remarkably busy stretch. SPEAKER_2: All in one week: they ran a six point six billion dollar employee secondary at a four hundred billion dollar valuation, finalized a ten billion dollar private equity joint venture, and restructured into a new Deployment Company. SPEAKER_1: And on the IPO and funding side, Cerebras upsized its IPO to four point eight billion dollars. Ramp raised seven hundred fifty million at a forty billion dollar plus valuation. And Kalshi, Helsing, and Isomorphic Labs all closed nine-figure rounds. SPEAKER_2: It was a dense week for capital markets across the board. SPEAKER_1: On the security side, Chrome patched its fourth zero-day of twenty twenty-six. And researchers found AI-generated code failing basic security checks at scale. SPEAKER_2: The concern there is that the attack surface is widening faster than the defenses. That's the framing worth holding onto. SPEAKER_1: And for founders navigating all of this, Startup Intros has two events coming up. May twelfth is Agentic AI Founders' Night-Out, focused on where agentic AI is heading, who's building, and who's backing it. Spots are limited. SPEAKER_2: And May fifteenth is Caffeine and Capital, their morning office hours format. You bring your fundraising questions, product challenges, and growth blockers. The mission is simple: help tech founders navigate everything in the early stage, from compliance to capital. SPEAKER_1: To recap this week: Anthropic is on track to be the highest-valued private company in history, powered by forty-times revenue growth in eighteen months; NVIDIA has engineered a forty billion dollar equity flywheel that locks in its customers structurally; OpenAI restructured and raised across multiple vehicles in a single week; and the security risk from AI-generated code is growing faster than the defenses. Next week we'll go deeper into what Anthropic's valuation means for the broader AI funding landscape.