Anthropic closed a $65 billion Series H on a $965 billion post-money valuation, the company announced today. Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital co-led the round, with Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN named as additional co-leads. The strategic stack underneath is where the round stops looking like a venture transaction and starts looking like an industrial financing: $15 billion came from hyperscalers and infrastructure partners, including a $5 billion check from Amazon and direct investments from Micron, Samsung, and SK hynix on the memory and storage side. Blackstone, Brookfield, Fidelity, General Catalyst, Jane Street, and T. Rowe Price round out the financial-investor tier.
The valuation prints a record. At $965 billion post-money, the company has roughly fifteen-x'd its March 2025 Series E mark and is sitting three months past its February 2026 Series G. Cadence is itself the story. Anthropic disclosed run-rate revenue crossing $47 billion in May 2026, the figure CFO Krishna Rao cited as the underpinning of the round. The use of funds, per Rao's commentary, advances safety and interpretability research, expands compute capacity, and scales the Claude Code and Cowork product lines. The named co-leads collectively represent the crossover and late-stage growth pools still willing to write nine-figure private checks at this valuation rather than wait for a public-market entry.
The compute side carries equal information. Anthropic has previously disclosed 5 gigawatts of compute capacity locked in with Amazon, approximately 3.5 gigawatts of TPU capacity through Google and Broadcom, and GPU access through SpaceX. The hyperscaler portion of the cap table is, in effect, a financing layer on those infrastructure agreements. Amazon's $5 billion equity check runs alongside the company's existing AWS Trainium commitments. Micron, Samsung, and SK hynix joining the round directly is less common: memory suppliers have historically taken volume commitments rather than equity. Their participation implies Anthropic's high-bandwidth memory consumption now justifies cap-table positioning, with multi-year supply commitments sitting behind the equity stake.
For the broader frontier-AI market, three things will be visible over the next twelve months. First, whether OpenAI's next financing prints at a comparable multiple of revenue, which would set a normalized private-market frontier-AI ratio. Second, whether four-way co-leadership becomes the standard structure for rounds of this size; co-leads at this scale signal that no single firm could absorb the allocation, which has implications for every subsequent megadeal. Third, whether the hyperscaler-equity model becomes contractually binding around future compute supply. The structure emerging here looks closer to Hollywood slate financing than to a traditional venture round, and the precedent set will shape how the next decade of frontier-model capital structures itself.