Anthropic Closes $65B Series H at $965B Valuation, Surpasses OpenAI and Previews Mythos

The AI funding story of 2026 just found its exclamation point. On May 28, Anthropic closed a $65 billion Series H funding round at a post-money valuation of $965 billion — making it the most valuable private technology company in history and the first AI startup to breach the near-trillion-dollar threshold. The round eclipses OpenAI’s own record $852 billion valuation set in March, reshuffling the leaderboard of the two companies that are most intensely competing to define the AI era.

The Mechanics of the Mega-Round

The $65 billion Series H is notably larger than the $30 billion-plus range that earlier reporting had forecast. The discrepancy is explained in part by the structure of the deal: approximately $15 billion of the total represents investment commitments that had already been made by hyperscale cloud partners — including a $5 billion tranche from Amazon, which has separately committed up to $10 billion in total to Anthropic under a broader cloud and model partnership. The remaining $50 billion represents fresh equity capital flowing into Anthropic’s balance sheet.

The round was led by Altimeter Capital, Dragoneer Investment Group, Greenoaks Capital, and Sequoia Capital. Co-leading the round alongside them were Capital Group, Coatue Management, D1 Capital Partners, GIC (Singapore’s sovereign wealth fund), ICONIQ Growth, and XN. The breadth of the investor base — crossing traditional venture capital, growth equity, sovereign wealth, and asset management categories — reflects the degree to which Anthropic has been repositioned not as a speculative AI startup but as an infrastructure company with durable revenue.

Revenue That Justifies the Price

The valuation is anchored by numbers that would have seemed fantastical eighteen months ago. Anthropic’s annualized revenue run rate had crossed $47 billion by May 2026, up from approximately $10 billion at the end of 2025 — a roughly 4.7x increase in five months. That trajectory, driven primarily by Claude’s penetration into enterprise workflows and the explosive growth of the Claude Code developer platform, has convinced investors that the company can sustain hypergrowth even at scale.

For context, the $965 billion post-money valuation implies a forward revenue multiple of roughly 20x on the current run rate, or more conservatively, 8-10x if analysts accept management’s guidance for further acceleration through the end of the year. Those multiples are high by traditional enterprise software standards but increasingly standard benchmarks for frontier AI infrastructure companies where switching costs are rising and integration depth is compounding.

Anthropic’s co-founders, CEO Dario Amodei and President Daniela Amodei, are each now worth approximately $8 billion based on their equity stakes — wealth that was largely theoretical until the company’s revenue inflection made external benchmarks credible.

Claude Opus 4.8: Safety as Competitive Moat

The Series H closing was accompanied by the immediate release of Claude Opus 4.8, the new frontier version of Anthropic’s flagship model. Anthropic’s framing of the release was notable: the company described Opus 4.8 primarily through the lens of safety and behavioral alignment rather than benchmark scores or capability claims.

According to Anthropic’s internal evaluations, Claude Opus 4.8 is meaningfully less likely to deceive users, cooperate with misuse scenarios, or assist with tasks that conflict with its stated values than its predecessors. The model incorporates updated constitutional AI training and a revised reinforcement learning from human feedback process designed to make alignment properties more robust to adversarial prompting.

For enterprise customers, the alignment story is increasingly commercial: regulated industries including finance, healthcare, and law have consistently cited concerns about AI model unpredictability and potential liability exposure as their primary barriers to deeper deployment. A model that demonstrably fails less often at safety-critical junctions is, in that context, a product differentiation with dollar value.

Claude Code, Anthropic’s agentic software development tool that has been the fastest-growing revenue driver over the past two quarters, was simultaneously updated to run Opus 4.8 with elevated effort defaults, meaning the agent will spend more compute per task on verification and self-correction before completing coding assignments.

Mythos: The Model Behind the Warning Labels

The most consequential revelation of the May 28 announcement may not be the funding round or the Opus 4.8 release, but rather Anthropic’s public confirmation that its Mythos-class model — previously known primarily through internal safety evaluations and limited government previews — is coming to general availability “within weeks.”

Mythos has been the subject of intense speculation since Anthropic disclosed in earlier safety documentation that its capabilities included not just advanced code generation but what the company described as “autonomous vulnerability chaining”: the ability to identify security weaknesses in software systems, combine multiple vulnerabilities together, and construct sophisticated multi-step attack sequences — all without human guidance between steps.

That capability profile is why Anthropic adopted a tiered access protocol for Mythos deployments, requiring customers to undergo enhanced vetting before gaining API access. The forthcoming general release will come with what Anthropic described as “significant safeguards, monitoring, and use policy restrictions,” including real-time evaluation of request intent and automated throttling for patterns associated with offensive security applications.

The company has worked closely with U.S. government partners — including the National Security Agency, Cybersecurity and Infrastructure Security Agency, and several defense contractors — to allow vetted access to Mythos capabilities for legitimate defensive security, vulnerability research, and penetration testing use cases.

Anthropic’s decision to preview Mythos’s general availability in the same announcement as a $65 billion funding round is itself a strategic communication: the company is signaling to enterprise customers, government partners, and investors alike that the capability frontier and the safety investment are advancing in tandem, not in opposition.

The IPO Clock Starts Ticking

The sheer scale of the Series H — and the sovereign wealth and institutional co-investors it brought in — suggests that Anthropic’s trajectory toward a public offering is accelerating. TechCrunch reported that sources familiar with the fundraise indicated Anthropic was pursuing the round “ahead of an IPO” that the company is exploring for as early as 2027.

A public offering at or near the current $965 billion post-money valuation would make Anthropic’s IPO one of the largest in technology history, comparable in scale only to Alibaba’s 2014 debut and Saudi Aramco’s 2019 listing in different sectors. Whether the public markets, which have already absorbed OpenAI’s confidential S-1 filing and SpaceX’s IPO preparation in 2026, will sustain those multiples for an AI infrastructure company remains to be seen.

What is certain is that Anthropic has crossed from startup into something closer to a pillar of the AI economy. Its models run inside Goldman Sachs, PwC, Bristol Myers Squibb, KPMG, and hundreds of other large enterprises. Its developer platform supports a fast-growing ecosystem of agentic applications. And its safety research, however contested at the margins, has become a de facto reference point for the policy conversations happening in Washington, Brussels, and Beijing alike.

The road from four-year-old company to near-trillion-dollar valuation has been historically fast. The harder question — whether the company can deliver the scientific, engineering, and governance outcomes its valuation implies — is the one that the next several years will answer.