Anthropic Series H: $65 Billion Round Approaches a Trillion

Anthropic has closed its Anthropic Series H at $65 billion, pushing its post-money valuation to $965 billion. With annualized revenue of $47 billion and compute infrastructure agreements covering more than 10 gigawatts of capacity, the company is now a few dozen billion away from a threshold no AI firm has crossed.

Key Takeaways

• $65 billion raised in Series H, post-money valuation $965 billion, annualized revenue $47 billion
• Round co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia; Amazon alone contributed $5 billion
• Claude now available on AWS, Google Cloud, and Microsoft Azure simultaneously (a first for a western frontier AI model)

The Biggest AI Round Ever, One Step from a Trillion

The round was co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. The numbers are straightforward: $65 billion raised, a $965 billion post-money valuation, and an annualized revenue run rate of $47 billion as of May 2026. That last figure deserves a second read. It is the current run rate, not a year-end projection.

Within the funding breakdown, $15 billion came directly from cloud providers. Amazon contributed $5 billion on its own. This structure, where strategic customers become direct shareholders, is unprecedented at this scale in the AI sector. It also signals a level of confidence from AWS that goes well beyond a standard cloud partnership.

CFO Krishna Rao specified the capital allocation: safety research, computing capacity, and expansion of products including Claude Code and Cowork. No mention of consumer marketing campaigns. The message targets technical teams and institutional investors, not general audiences. That computing capacity gets contracted site by site, as when Riot Platforms leased 191 MW of Texas data center space to the lab.

The timing alongside the Claude Opus 4.8 release on the same day is deliberate. The new model outperforms GPT-5.5 on most agentic benchmarks and requires 35% fewer tokens on real-world tasks. For the full breakdown of what Opus 4.8 changes in practice, the complete analysis of Opus 4.8 and its dynamic workflows is available on Horizon. The funding round and the model release form a single, consolidated message to the enterprise market.

This valuation places Anthropic in a category few tech companies have reached without going public. The market pricing it at nearly a trillion dollars is not betting on a niche player. It is betting on structural dominance of a critical layer in enterprise software.

Infrastructure Built for the Next Decade

The most strategically significant detail in the announcement is not the amount raised but what it funds: more than 10 gigawatts of computing capacity secured through agreements with Google, Broadcom, and SpaceX. This is not server rental. It is the construction of proprietary infrastructure designed to last years, not quarters.

Partnerships with Micron, Samsung, and SK hynix extend the picture further. Anthropic is not just securing inference capacity in production. It is taking a position across the full hardware chain, from memory components to deployment. A vertically integrated supply chain assembled through contracts rather than acquisitions, but structurally equivalent in effect. The full picture is available in Anthropic’s announcement of its $65B Series H funding.

The simultaneous availability of Claude on all three major cloud platforms closes the last operational argument against adoption. AWS, Google Cloud, and Microsoft Azure all offer Claude natively. According to Anthropic, this makes it the first western frontier AI model with this level of distribution. Whatever cloud infrastructure a company runs on, Claude is available without migration or technical friction.

This infrastructure push is paired with accelerating geographic presence. The recent Seoul office opening came with adoption figures Anthropic described as record-setting. The analysis of that Korean expansion and what it reveals about the Asia-Pacific strategy puts the scale of this deployment in perspective. Partnerships with Korean and Japanese memory component manufacturers fit the same pattern.

Also on Horizon:

• Opus 4.8: Anthropic Ships Fast and Rewrites the Rules
• Apple Intelligence iOS 27: Siri Goes After ChatGPT
• Anthropic Korea: Seoul Office Opens on Record Claude Usage

Short Term, Medium Term: What Actually Changes

In the short term, the most tangible signal is directed at technical decision-makers. A $965 billion valuation paired with a $47 billion revenue run rate places Anthropic in the category of critical infrastructure. For CTOs and IT leaders who were still hesitant to standardize Claude in enterprise workflows, this financial signal removes a real blocker. The vendor risk question leaves the room. For context, see our earlier piece on Horizon: Anthropic Closing In on a $900 Billion Valuation.

The Opus 4.8 pricing confirms the broader strategy: $5 per million input tokens, $25 per million output tokens, identical to Opus 4.7 despite better performance. Holding those prices after the Anthropic Series H and its $65 billion raise is a deliberate choice. Anthropic is maximizing adoption over near-term margin. That is the posture of a company consolidating market position, not extracting short-term revenue.

Over the medium term (3 to 6 months), the roadmap is readable. The 10-plus gigawatts of compute and the dynamic workflows in Opus 4.8 converge on a single objective: running hundreds of agents in parallel within continuous work sessions. The infrastructure investments being made today are sized for that type of load, not for traditional chat interfaces.

For the players already positioned on the agent layer (Microsoft, Google, AWS), the message is unambiguous. With $47 billion in annualized revenue and a three-cloud distribution footprint, Anthropic is no longer the predictable challenger. The 2021 startup has become critical infrastructure for a significant share of the global enterprise market.

Follow the story on Horizon.