Anthropic eyes october ipo to outrun openai in $60 bn sprint

Anthropic is quietly plotting a New York listing as early as October, vaulting from lab bench to trading floor in a bid to bankroll the next phase of the AI arms race, according to people familiar with the talks.

The San Francisco start-up, valued at $38 billion in February, has opened preliminary discussions with Goldman Sachs, JPMorgan and Morgan Stanley about underwriting what could be a $60 billion-plus float, the same trio now circling OpenAI’s long-mooted debut. The parallel roadshows would turn autumn into a cage match between the two most powerful large-language-model builders on Earth.

The pentagon calls anthropic a supply-chain threat

While investors salivate over growth curves, national-security mandarins are tightening the screws. The Department of Defense in March branded Anthropic a “substantial risk” to America’s supply chain, citing its refusal to let the military weaponise Claude. The label, normally reserved for Chinese telecoms firms, triggered a funding freeze that Anthropic only loosened last week with a court injunction. The ruling, unsealed in Delaware, warned that a blackout could wipe “billions” off future revenue.

The clash is more than bureaucratic noise; it is a live stress test of Silicon Valley’s vow to build “safe” AI. Anthropic’s constitutional AI stack—the guardrails baked into Claude—has become its marquee pitch to banks and hospitals. Yet those same red lines now draw red tape across federal contracts. The irony: the ethical stance that lures enterprise customers may cap its single biggest sales channel.

Behind the scenes, the cap-table reads like a tech-titan family tree. Google, Amazon, Microsoft and Nvidia have poured in chips, cloud credits and raw cash, collectively locking up preferred shares worth tens of billions. Their incentives diverge: cloud vendors want usage surges; Nvidia wants compute lock-in; founders Dario and Daniela Amodei want governance autonomy. An IPO forces those agendas into daylight quarterly.

Market timing is delicate. Venture funding for generative start-ups contracted 27 % year-on-year, yet public markets have rewarded anything with an AI label—see Nvidia’s 180 % YTD surge. Missing the window could mean a valuation haircut later, but rushing risks the optics of a cash-burn story still years from profit. The Information first flagged the October timeline; sources now say a final board vote is slated for September, conditional on SEC feedback.

Meanwhile, OpenAI’s Sam Altman is courting the same banks for a 2025 listing, dangling a revenue run-rate already above $3.4 billion. Anthropic counters with a smaller top line but faster enterprise uptake—Claude now powers back-office workflows at Pfizer and Bridgewater. Investors must pick between scale today versus safety credentials tomorrow.

Whichever lists first will set the pricing compass for the entire cohort. A frothy debut could unlock a wave of AI secondaries, from Cohere to Hugging Face. A dud, and the sector’s risk premium widens overnight. Either way, the October clock is ticking louder every day.