Anthropic has delayed its initial public offering amid calls from New York City pension funds to pause listing due to autonomous AI safety risks. The move follows OpenAI CEO Sam Altman postponing OpenAI’s own IPO until at least next year, calling a listing under current market conditions ill-advised. Despite Anthropic generating $16.8 billion in Q3 2026 revenue—reaching an annualized run rate of $67 billion—its prospectus highlights existential risks to humanity and reveals that a quarter of its 2025 revenue relied on just two customers.
The delays occur alongside a broader tech selloff driven by soaring infrastructure expenditures and the rapid rise of open-weight alternatives. Gartner projects that cumulative AI data center spending through 2030 will reach $8.5 trillion against at most $3.6 trillion in attributable revenues. Furthermore, open-weight models like DeepSeek R1 are eroding closed-model pricing power, offering comparable performance at significantly lower API costs.
Financial analysts warn that justifying Anthropic’s proposed $2 trillion valuation would require sustaining an immediate 10% after-tax operating margin for a decade alongside $358 billion in annual after-tax operational income. As enterprise adoption of open-weight models jumps to 67%, proprietary AI vendors face increasing structural pressure on their long-term economic margins.
Why it matters
Open-weight models are slashing enterprise API costs by up to 90%, threatening closed-model pricing power.
Soaring AI data center infrastructure costs are outpacing projected revenues, raising concerns among institutional investors.
AI safety risks and concentrated revenue streams are delaying major tech IPOs despite strong top-line revenue growth.
Source: news.crunchbase.com



