Anthropic has closed a staggering $65 billion Series H round, propelling its valuation to a near-trillion-dollar mark of $965 billion. While skeptics continue to debate the existence of an "AI bubble," investor pools led by Sequoia, Greenoaks, and Altimeter Capital are writing checks backed by a tangible $47 billion in revenue (as of May 2026). The total includes previously committed funds of $15 billion from cloud giants, including a $5 billion tranche from Amazon. But if you thought this capital was destined for Claude's user interface, you are mistaken: Anthropic is pivotting from being a mere software firm to becoming the foundational owner of the computing layer.
Infrastructure Sovereignty and 10 Gigawatts
The strategy championed by CEO Dario Amodei and CFO Krishna Rao is now rooted in physical dominance. According to company reports, these funds will be allocated not just to AI safety and Claude Code, but toward securing 10 GW of power capacity. To put that scale in perspective: Anthropic has inked deals with Google, Broadcom, and SpaceX. The participation of memory giants Micron, Samsung, and SK hynix in this round confirms that the company is building a vertical stack—from custom silicon and orbital connectivity to direct power generation.
Essentially, Anthropic is treating AWS and Azure as mere distribution channels while simultaneously erecting its own technological fortress.
Key takeaways from Anthropic's new strategy:
Capital concentration is transforming this private company into a de facto state-level institute for AI safety. This is no longer a startup; it is a systemically important player with an appetite backed by a massive infrastructure foundation. While Claude remains a top-tier frontier model in the West, the real battle is shifting toward control over the physical world of computing.
Anthropic must now prove that $47 billion in annual revenue is not a ceiling, but a modest starting point for an entity aiming to manage the world's intelligence infrastructure. The bet is clear: in the future, success depends not just on algorithms, but on gigawatts and silicon supply chains.