OpenAI seeks $30 billion at $1.4 trillion valuation before Anthropic IPO

OpenAI is rushing to close a $30 billion funding round at a valuation of more than $1.4 trillion, aiming to lock in its private valuation before rival Anthropic lists on the stock market, The News reported.

By raising the money before Anthropic’s expected public debut, OpenAI hopes to set its own valuation terms. Anthropic’s coming financial disclosures, through its confidential S-1 filing, will give Wall Street a clear look at the finances of a frontier AI lab for the first time, which could invite close comparisons with OpenAI.

What OpenAI is telling investors

• Revenue: OpenAI says it has reached $50 billion in annualised revenue, with growth speeding up after the launch of its GPT-6 Astra model, and expects $70 billion by the end of the year
• Prices: it has cut prices on its cheaper models, which it says has boosted usage, and told investors it can lower prices further
• Timing: the rapid timetable partly reflects a wish to get ahead of Anthropic, as the two labs increasingly pitch to the same investors

Demand for shares in both companies is high, but investors face the challenge of deciding how to split their money between two competing offers within days or weeks of each other. Some see OpenAI’s push as an attempt to lock in capital from funds that might otherwise put more into Anthropic’s IPO.

Altman’s three endgames

Separately, OpenAI chief executive Sam Altman has outlined three possible endgames for the arrival of superintelligence in an internal briefing to staff, The Information reported, according to The News.

1. An “IAEA” model: an international oversight body, like the International Atomic Energy Agency, that would enforce safety rules, audit models before release and independently check all the leading AI labs
2. The “Demis plan”: named after Google DeepMind chief Demis Hassabis’s idea of a CERN for AI, in which the main labs would merge or channel their research through one international consortium that supplies models to participating countries
3. The “Eliezer plan”: named after AI safety theorist Eliezer Yudkowsky, in which one developer builds AI that can improve itself, gains overwhelming economic power and uses it to “melt everyone else’s GPUs”, starving rivals of hardware

Altman reportedly favours the first path, seeing global regulatory oversight as the most balanced approach to safety. He rejected the third outright. “Obviously I think that one’s terrible,” he said, but he also reportedly told staff that OpenAI cannot afford to lose ground to its competitors.

A push for rules

The comments come as calls grow for stronger AI regulation. In the United States, a White House task force chaired by Director of National Intelligence Jay Clayton recently adopted its formal charter, focusing on rules for reporting breaches, hacks and rogue AI models after early test incidents.

Why it matters

How investors split their money between OpenAI and Anthropic will shape the race to build the most powerful AI. Read our reports on the former OpenAI engineer held without bail and on the states suing OpenAI over child safety.

This article draws on reporting by The News and The Information.