Anthropic's Rise Is Giving Some OpenAI Investors Second Thoughts About the $340B Valuation

The Financial Times has surfaced something that has been discussed quietly in Sand Hill Road conference rooms for months: at least some of the investors who participated in OpenAI's most recent funding round at a $340 billion valuation are experiencing what one described as "valuation anxiety" as Anthropic's competitive trajectory has become impossible to dismiss. The story, confirmed through multiple investor conversations by TechCrunch, reflects a broader recalibration in how sophisticated capital is thinking about the frontier AI race — one that is less about which lab has the best model today and more about which has the right strategy for the next phase of the industry.

The Anthropic Case

Anthropic's case to investors in 2026 is structurally cleaner than it was eighteen months ago. The company has a clear enterprise customer base, growing revenue, a differentiated safety narrative that is gaining traction with regulated industries (banking, healthcare, government), and — with Mythos — a product that has no direct OpenAI equivalent at the capability level relevant to professional security buyers. The CoreWeave infrastructure deal announced in April provides compute security at a scale previously only available to Microsoft-backed OpenAI, addressing a longstanding concern about Anthropic's ability to compete on model training at the frontier. The cumulative picture is of a company that has substantially de-risked its operational position while OpenAI has been managing internal turbulence and the complexity of its transition from nonprofit to capped-profit to whatever its current governance structure actually is.

The OpenAI Defense

OpenAI's defenders argue that the $340 billion figure is justified by ChatGPT's consumer distribution moat, the Microsoft partnership's structural advantages, and the company's lead in agentic product deployment. The counterargument is that consumer distribution in AI is fragile — switching costs are low, models are commoditizing, and enterprise buyers evaluate on capability and reliability rather than brand. The Microsoft relationship, meanwhile, is increasingly complex: Microsoft has its own AI ambitions, and the terms of the OpenAI partnership have evolved in ways that do not obviously benefit OpenAI's independent valuation. None of this means the valuation is wrong — but it does mean it is a bet on continued execution in a competitive environment that has grown materially more difficult since the round was priced.

What Investors Are Actually Saying

The FT's sourcing is careful: "second thoughts" does not mean panic, and the investors quoted are not suggesting the round was a mistake. What they are articulating is something more specific: that the implied assumptions at $340 billion — specifically, the assumption of a path to IPO at a price that justifies the round — require OpenAI to maintain competitive differentiation in a market where Anthropic, Google DeepMind, and potentially open-source models are all improving faster than the bull case anticipated. One investor's comment that "justifying the round required assuming an IPO valuation" is a statement about the fragility of the logic, not necessarily its incorrectness. The window for OpenAI to establish durable competitive moats is narrowing, and that is what the most thoughtful investors are watching.