Anthropic targets $900B+ valuation in $50B round, set to surpass OpenAI ahead of IPO

Anthropic is asking investors to submit allocation commitments within 48 hours for a fundraise expected to total roughly $50 billion, targeting a valuation of about $900 billion — more than doubling its February valuation of $380 billion. Sources indicate the round could close within two weeks, with final valuation potentially exceeding $900B due to intense investor demand. The company’s annual revenue run rate has surpassed $30 billion, with some sources placing it closer to $40 billion. Some early backers from 2024 or earlier are sitting out, preferring to cash out during Anthropic’s anticipated IPO later this year.

Why it matters: This round signals the apex of the capital-compression arc in foundation-model scaling, where the leading labs must raise staggering sums to fund compute infrastructure before going public. At $900B, Anthropic would surpass OpenAI’s $852B post-money valuation, reshaping the competitive hierarchy among the top-tier labs. The fact that early investors are skipping the round to wait for the IPO suggests confidence in a near-term public listing, which would mark the first major foundation-model IPO and test whether public markets can sustain the astronomical private valuations that hyperscaler-backed capital cycles have created.

Expert take: The round crystallizes two structural forces: first, the capital-cycle dynamics of frontier AI, where the cost of training and inference compute demands that labs raise at ever-larger scales — $50B is no longer shocking in this context. Second, the hyperscaler-distribution pattern is inverted here: Anthropic is raising from a broad investor base rather than deepening reliance on a single cloud partner, which may give it more strategic flexibility ahead of an IPO. The real open question is whether public markets will accept a $900B+ valuation for a company that, despite $30B+ in revenue, still spends heavily on compute and has yet to demonstrate durable profitability at scale.