Anthropic has received several offers from venture capital firms to raise a new round at approximately $800 billion or higher, according to Bloomberg. [2] The company has so far rebuffed all of them. [1] [4] In February, Anthropic closed a $30 billion round at a $380 billion valuation. In two months, the market decided that price was roughly half of what the company is worth.

The numbers arrive one day after this paper examined how Anthropic and OpenAI staked out opposing positions on Illinois's AI liability bill — the "open" company demanding legal immunity while the "safe" company argued that liability makes AI better. That story framed the two firms as ideological opposites building the same product. The valuation story makes the frame concrete: the company that insists it should be held accountable for its technology is now almost as valuable as the one insisting it should not.

This is the safety-company paradox, and it is worth $800 billion.

Anthropic's run-rate revenue reached $30 billion by the end of March, up from $9 billion at the end of 2025 — a tripling in roughly ninety days. [3] The company now counts over one thousand business customers spending more than $1 million annually, a figure that doubled in less than two months. [1] On the Caplight secondary exchange, shares trade at a $688 billion implied valuation, up 75 percent in three months. [1] These are not the growth metrics of a safety-focused research lab. They are the growth metrics of a company that has found the most lucrative market position in the history of technology: the responsible alternative.

The structural comparison to OpenAI is irresistible and misleading in equal measure. OpenAI raised $110 billion at an $852 billion post-money valuation. [3] At $800 billion, Anthropic would trade at roughly 94 percent of OpenAI's last round — a near-parity that would have seemed absurd eighteen months ago, when Anthropic was widely regarded as a well-funded research outfit running a distant second. The gap closed not because Anthropic caught up on capabilities — both companies ship frontier models — but because Anthropic created a brand proposition that enterprise customers find more comfortable: the AI company that wants to be regulated.

The paradox is that safety, as a market position, works precisely because it is expensive. Anthropic has committed $50 billion for data centers and $30 billion on Microsoft's cloud infrastructure. [3] Capital expenditures at this scale are normally associated with companies trying to outrun competitors. Anthropic is spending to keep pace with OpenAI while simultaneously arguing that the race should have speed limits. The market reads this not as contradiction but as strategy — a company that can match OpenAI's output while maintaining the safety brand commands a premium that pure capability cannot.

The IPO conversation makes the premium tangible. Anthropic is in talks with Goldman Sachs, JPMorgan, and Morgan Stanley about a potential public offering as early as October 2026, which could raise more than $60 billion. [2] An IPO would be the first test of whether public markets value the safety positioning as highly as private investors do. Public investors tend to discount narrative and reward revenue growth. Anthropic has both — the narrative of responsible AI and the revenue curve of a company growing at triple-digit annual percentages.

The timing is not accidental. Anthropic announced Mythos, its newest model, described as its "most capable yet for coding and agentic tasks." [1] The release is the company's answer to the persistent criticism that safety and capability trade off against each other. Mythos is Anthropic's argument in product form: you can build the most capable model and still be the company that opposes liability shields. Whether this is true in the long run is an open question. In the short run, it is a $800 billion argument.

The brand inversion from the Illinois fight deserves a second look in light of the valuation numbers. OpenAI — the company with "open" in its name — lobbied for legal immunity under SB 3444. Anthropic — the company whose name evokes caution about human-level AI — argued that developers should face liability for foreseeable harms. The market's response to this positioning is unambiguous: investors are paying near-parity prices for the company that accepts accountability over the one that rejects it. This does not mean investors care about safety in the abstract. It means they believe enterprise customers care, and enterprise customers are the ones writing million-dollar annual contracts.

Anthropic has also drawn up plans to donate $20 million to support candidates who favor AI regulation. [3] This is lobbying, but it is lobbying in the opposite direction from the industry consensus. Most technology companies spend to prevent regulation. Anthropic is spending to encourage it — on the theory that regulation creates barriers to entry that favor well-capitalized incumbents with compliance infrastructure already in place. The safety positioning is, at bottom, a moat strategy. The castle happens to have a conscience. Whether the conscience survives the moat is the question that an $800 billion valuation cannot answer.

The VCs offering $800 billion are not buying safety. They are buying revenue growth, model capability, and enterprise traction that happens to be wrapped in a regulatory philosophy that makes customers feel better about deploying AI in sensitive contexts. Strip away the philosophy and Anthropic is a company tripling revenue, doubling its enterprise base, and preparing for the largest AI IPO in history. The safety brand is the packaging. The product is the same thing OpenAI sells — intelligence on demand, at scale, for a price.

What makes Anthropic's position genuinely interesting is that the packaging may have become the product. A company that opposes liability shields, funds pro-regulation candidates, and turns down $800 billion rounds is making choices that constrain its future behavior. These are not marketing decisions. They are commitments — or they are nothing. The market is betting they are commitments. At $800 billion, the market is betting a great deal.

-- THEO KAPLAN, San Francisco