Anthropic Is Now Worth More Than OpenAI

The Scoreboard Has Flipped

For years, OpenAI was the unquestioned center of gravity in artificial intelligence. It had the brand, the backing, and the narrative. Now, by nearly every financial metric that matters ahead of a public offering, Anthropic has pulled ahead, and the gap is widening fast.

This week, Anthropic announced a new $65 billion fundraising round led by Altimeter, Dragoneer, Greenoaks, and Sequoia Capital, pushing its pre-IPO valuation to $965 billion. That figure tops OpenAI's $852 billion valuation, which was itself set by a record $122 billion round closed just in February. In a matter of months, Anthropic has gone from a well-regarded challenger to the more valuable company in the room.

The valuation headline is striking on its own. What makes it a genuine business story is everything else sitting behind it.

The Revenue Gap Is the Real Story

Anthropic's annualized revenue run rate has climbed from $10 billion at the end of 2025 to over $47 billion today, a move that would be remarkable in any industry. OpenAI, by comparison, has gone from $13 billion to $25 billion over the same stretch. Both companies measure revenue slightly differently, and that accounting nuance has generated its share of sniping. But as this episode makes clear, the underlying trajectory divergence is hard to explain away with accounting methodology alone.

From a capital efficiency standpoint, Anthropic has also done more with less. It has raised roughly $144 billion to reach this point versus approximately $180 billion for OpenAI. In a sector where burn rates are enormous and profitability remains theoretical, that spread matters.

Momentum Compounds in Strange Ways

What is harder to quantify, but equally important, is narrative momentum. Anthropic has it. OpenAI is losing it.

Consider the recent sequence of events in Anthropic's favor. Claude Mythos, its most powerful model, is so capable at detecting software vulnerabilities that Anthropic has restricted access to roughly 40 top enterprises globally. Even a significant security lapse, in which a group of Discord users reportedly guessed their way into Mythos files within Anthropic's codebase, did not dent investor confidence. Big tech firms followed with an additional $30 billion in deals.

Then came Andrej Karpathy. A co-founder of OpenAI who later led AI work at Tesla, Karpathy announced on social media that he is joining Anthropic to help build out pre-training. It is the kind of talent signal that analysts and engineers both read closely.

And then there is the Vatican. An Anthropic co-founder attended the Pope's unveiling of a new encyclical on AI, aligning the company publicly with concerns about how technology is being wielded by the powerful. Whether one reads that as genuine conviction or calibrated positioning, it is a brand move that OpenAI is not making.

OpenAI's IPO Trap

OpenAI is preparing for a public offering, with Goldman Sachs and Morgan Stanley reported to be advising. But the timing question is becoming genuinely difficult. The company needs to go public before Anthropic potentially absorbs more of the available investor appetite. At the same time, its underlying metrics, including a missed target of one billion weekly active users, suggest the business may not yet be at its strongest point for a debut.

Reports indicate that CFO Sarah Friar had argued internally for waiting until 2027. That view appears to have lost out, at least for now. The tension between IPO timing and business readiness is a classic trap, and OpenAI appears to be sitting squarely in it.

Altman's Retreat and What It Signals

The episode closes on a detail that may prove to be the most telling. Sam Altman, speaking at a conference in Sydney, walked back his previous warnings about AI-driven mass job displacement, suggesting the disruption will be less severe than he had argued. The charitable read is that he is managing IPO optics by softening a politically charged message. The less charitable read is that AI adoption, across consumer and enterprise, is more fragmented and slower-burning than the original thesis assumed.

That fragmentation is visible in the market. Google's Gemini had a strong 2025 and continues to grow. An Uber executive recently said publicly the company is not seeing the ROI it expected from AI. The demand is real, but it is not concentrating around a single winner the way OpenAI's original compute strategy assumed it would.

Anthropic, for now, is in the unusual position of winning a race that is still very much being run. The next chapter will depend on whether its revenue trajectory holds, whether Opus 4.8 delivers on its benchmarks in neutral testing, and whether OpenAI can stabilize its narrative before the IPO window closes.