SAN FRANCISCO — The paperwork has not been made public, and the company has said almost nothing, but the filing landed this week and it is the most anticipated document in technology since OpenAI last raised money. Anthropic has submitted a confidential IPO registration to the Securities and Exchange Commission, according to people familiar with the matter, beginning the process that could bring the AI company to public markets at a valuation approaching one trillion dollars.
A confidential filing is the standard first step for a large company preparing to go public, and it tells investors nothing about timing, size, or pricing. What it tells them is direction: Anthropic is moving toward the public market, and the private-market valuation that has been floating around the company is about to meet the scrutiny of public investors who will demand evidence behind the number.
The stakes are larger than one company. Anthropic and OpenAI form the two poles of the commercial AI industry, the pair that defines frontier model quality and the pricing that follows it. OpenAI’s structure, with its unusual governance and its deep ties to Microsoft, has made a conventional IPO complicated. Anthropic, by contrast, has a cleaner shareholder base, and its filing could set the template for how the AI industry enters the public markets. If Anthropic goes public first, its opening price will become the reference point for every AI company that follows.
The valuation question is the heart of the matter. A near-trillion-dollar price tag would put Anthropic in the same weight class as the largest technology companies on earth, with a fraction of their revenue. The company’s defenders point to growth rates that would embarrass most public software companies and to enterprise demand for Claude that shows no sign of slowing. The skeptics ask the same question they ask about every AI company: how much of this value is durable, and how much is the market’s belief that artificial intelligence will eventually be the most valuable industry in history?
Anthropic’s narrative has always been different from OpenAI’s. The company was founded on the principle that AI safety comes first, and it has built its brand around restraint, refusing some business, publishing safety research, and positioning itself as the responsible alternative. That story has been effective with enterprise customers and regulators. The question is whether it survives contact with the public markets, where the incentives run toward growth and the disclosure requirements run toward candor about risk.
The competitive timing is awkward and unavoidable. OpenAI has been reported to be exploring its own path to public markets, and the two companies are racing to be the first with a compelling public story. Being second in an IPO race has real costs: investor attention, valuation benchmarks, and the narrative of leadership all belong to the first mover. The AI duopoly is about to become a public-market duopoly, and the order of entry will shape how both are priced.
The broader market is preparing for a crowded year. With Anthropic filing and other AI companies at various stages of IPO preparation, 2026 is shaping up to be the most active year for AI listings since the industry’s founding, according to bankers who work on technology deals. The pipeline reflects both the industry’s capital hunger and the private markets’ appetite for exits, and it will test whether public investors share the private markets’ enthusiasm for AI valuations.
The risks in the filing will be the focus of scrutiny. Anthropic’s costs are enormous, driven by the compute required to train and run frontier models, and its dependence on cloud partners for that compute is a concentration risk investors will weigh. Its revenue concentration, with a relatively small number of large enterprise customers, will be examined line by line. The safety narrative cuts both ways in a prospectus: it is a brand asset, and it also calls attention to the company’s own research warning of serious risks from the technology it sells.
The private market has already been pricing the outcome. Anthropic’s shares have changed hands in secondary transactions for years, and the implied valuations in those trades have climbed with each new model release, according to people who track the market. The confidential filing gives those trades a public anchor, and it gives employees, who hold a large share of the company’s equity, a date to work toward. The human question behind the numbers is straightforward: the people who built Claude want to know what their work is worth, and the market is about to tell them.
For now, the confidential filing is a formality that is anything but formal. It tells the market that Anthropic has chosen the public path, that its bankers believe the numbers can work, and that the most important valuation question in technology will get a public answer. The exact timing remains the company’s secret, but the direction is now clear: the AI industry’s second giant is coming to the market, and the trillion-dollar test begins.
