The February benchmark is official: Anthropic said it raised $30 billion in Series G funding at a $380 billion post-money valuation in February 2026 . A reported $850 billion valuation would be about 2.2x that level; $900 billion would be about 2.4x. Because the new round has not closed and final structure is not public, those are scale comparisons rather than final deal terms.
A valuation jump of that size is easier to understand through the AI compute race. Anthropic is trying to secure capacity for both training and serving Claude, and its announced and reported partnerships form the backbone of the investor narrative.
| Partner | Announced or reported capacity | Why it matters |
|---|---|---|
| Amazon | Anthropic announced an expanded Amazon agreement for up to 5 gigawatts of capacity for training and deploying Claude, including Trainium2 capacity in the first half of 2026 and nearly 1 GW of Trainium2 and Trainium3 capacity by the end of 2026 . | Gives Anthropic a large committed supply path through Amazon infrastructure and chips. |
| Google and Broadcom | Anthropic announced a Google and Broadcom agreement for multiple gigawatts of next-generation TPU capacity expected to come online starting in 2027 . Reuters, citing The Information, separately reported a $200 billion Google Cloud spending commitment over five years . | Extends Anthropic’s custom-chip capacity beyond Amazon and ties the Claude roadmap to long-term cloud supply. |
| SpaceX | Gigazine and StorageReview reported a SpaceX compute agreement giving Anthropic access to Colossus capacity, including more than 300 megawatts and more than 220,000 Nvidia GPUs . | If the reporting is accurate, it adds a major GPU source outside the usual hyperscaler channels. |
| Akamai | Akamai disclosed a $1.8 billion, seven-year cloud deal with a customer it described as a leading frontier model provider; Bloomberg identified the customer as Anthropic, and The Next Web reported that Akamai’s stock rose 27% after the disclosure . | Suggests Anthropic is also diversifying into additional cloud and distributed infrastructure providers. |
The bull case is a matching problem: Anthropic needs enough demand to monetize huge capacity, and enough capacity to meet demand. On the demand side, Anthropic said in February that run-rate revenue was $14 billion and that the number of customers spending more than $100,000 annually on Claude had grown 7x over the prior year . TechCrunch later reported that Anthropic’s annual revenue run rate had surpassed $30 billion .
Those numbers help explain why investors might tolerate an exceptional private valuation. If Claude demand keeps expanding and compute remains scarce, reserved infrastructure could let Anthropic convert more enterprise usage into revenue. The caveat is important: run-rate revenue is not the same as audited full-year revenue or profit, and the available reporting does not disclose enough about margins to know whether the economics are durable.
The reports point to offers and discussions, not a closed financing. That matters because a $900 billion-plus valuation would reset expectations from February’s $380 billion post-money valuation to a level more than twice as high within months . Until a round closes, valuation, timing and terms can still change.
The same infrastructure that makes Anthropic attractive may be expensive to carry. A reported $200 billion Google Cloud commitment, an Amazon agreement measured in gigawatts, a reported SpaceX capacity deal and a $1.8 billion Akamai contract all point to a business where scale requires enormous upfront and ongoing spend . The open question is whether Claude usage can produce enough gross profit after chip, cloud, networking and power costs.
TechCrunch described the potential round as likely Anthropic’s last private raise before an anticipated IPO later in 2026, with proceeds aimed at massive computing needs . A richer private valuation can help fund that buildout, but it also raises the public-market hurdle. An IPO case would need investors to believe revenue growth can keep compounding while unit economics improve; the available reporting does not yet answer that.
If a round closes anywhere near the reported terms, Anthropic’s valuation story will be inseparable from its compute strategy. The case is not simply that Claude demand is growing; it is that Anthropic has been moving to secure the infrastructure required to serve that demand . The risk is just as clear: near-trillion-dollar private AI valuations leave very little margin for disappointment if compute costs stay high, revenue growth slows, or public investors demand clearer proof of profitability .