Anthropic filed confidentially for a U.S. initial public offering on Tuesday, according to three people familiar with the matter, after a recent funding round valued the artificial intelligence company near $965 billion. The filing positions Anthropic to test public-market appetite for frontier-model labs at valuations once reserved for entire semiconductor sectors.
Chief executive Dario Amodei and president Daniela Amodei informed employees that bankers would begin drafting prospectus language over the summer, with a listing targeted for late 2026 or early 2027 depending on market conditions. Anthropic declined to comment publicly; the people spoke on condition of anonymity because the filing is not yet public.
Revenue and Customers
Anthropic reported annualized revenue exceeding $14 billion in May, driven by Claude Enterprise contracts with financial institutions, pharmaceutical manufacturers, and federal agencies holding FedRAMP High authorization. API consumption grew faster than seat licenses as developers embedded Claude into agent workflows.
Unlike consumer chatbots that monetize through subscriptions alone, Anthropic's mix skews toward six- and seven-figure enterprise agreements with usage overages. Bankers pitching the IPO told investors that gross margins on inference improved after Anthropic negotiated reserved capacity deals with Amazon Web Services and Google Cloud.
Competitive Timing
The filing lands weeks after OpenAI began preparing its own listing at a valuation executives have discussed internally near $1 trillion. SpaceX's parallel IPO planning has raised questions about whether equity markets can absorb multiple mega-cap technology debuts in a single cycle without crowding out secondary offerings from smaller issuers.
Goldman Sachs and Morgan Stanley are expected to lead Anthropic's underwriting syndicate, the people said. Existing backers including Google, Spark Capital, and the Qatar Investment Authority would face lockup periods typical of venture-backed debuts.
Regulatory Overhang
Securities and Exchange Commission reviewers will scrutinize revenue recognition on multi-year cloud commitments and the company's safety research spending, which Anthropic has framed as a competitive advantage rather than a cost center. Civil liberties groups asked whether prospectus risk factors adequately describe model misuse scenarios in cybersecurity and biological design.
Congressional staff tracking AI policy said Anthropic's S-1 could become a public document shaping oversight debates if the company discloses government contract values and export-control compliance procedures in greater detail than private companies typically share.
Investor Questions
Portfolio managers must decide whether Anthropic's constitutional AI training approach justifies a premium to OpenAI or whether both firms converge on similar enterprise economics as inference commoditizes. Memory-supplier earnings and hyperscaler capex guidance in July will influence whether the IPO window stays open.
For employees holding options struck at earlier 409A valuations, a listing would crystallize gains that have already made Anthropic one of Silicon Valley's most closely watched cap-table stories. The confidential filing is the first formal step toward converting paper wealth into liquid shares.
Bankers pitching the deal highlighted Anthropic's enterprise renewal rates above 140 percent in the first half of 2026, a metric institutional investors treat as evidence that Claude deployments expand after initial pilots. Several Fortune 100 chief information officers told Credence Wire that procurement committees now shortlist Anthropic alongside Microsoft and Google rather than treating the company as a niche safety vendor.
Securities lawyers said confidential filing allows Anthropic to revise prospectus language if July technology volatility persists into roadshow season. Venture backers with board seats have not signaled secondary sales beyond standard IPO float mechanics, according to two investors briefed on governance discussions.
