Anthropic is considering rolling out a new AI model to counter OpenAI’s momentum since the launch of GPT-6 Astra, according to three people familiar with the matter who spoke to Reuters, a move that would land weeks after the company’s own chief executive called for the industry to slow down.
The timing is the awkward part. Anthropic CEO Dario Amodei published a 3,800-word essay on September 12 arguing that the pace of capability releases has to come down. “We must slow the pace at which we improve the capabilities of AI models,” he wrote. The essay described swarms of AI agents overwhelming the internet faster than humans can respond, and it drew public support from OpenAI CEO Sam Altman and Elon Musk. Thirteen days later, Reuters reports the company is weighing a launch of its own, with one source saying Anthropic is evaluating the safety of its next model as part of the deliberation.
The competitive pressure is concrete. OpenAI released GPT-6 Astra on September 3, pitching gains in computer use, software engineering, cybersecurity and professional work. Enterprise response has been strong enough that some Anthropic IPO investors are now asking whether OpenAI can take share from the company that has been viewed for months as the enterprise AI leader.
What the spending data shows
The numbers behind that concern come from corporate expense platform Ramp, which tracks what companies actually pay for. Astra accounted for about 13 percent of enterprise AI spending in the latest data, compared with about 8 percent for Anthropic’s Claude Fable. Astra has also pulled ahead on OpenRouter, a widely used platform that routes developer traffic across AI models.
OpenAI’s annualized revenue run rate passed $40 billion in July. Anthropic’s reached $65 billion by the end of the same month, up from about $9 billion at the end of 2025. Anthropic still leads on revenue, but the direction of the enterprise spending mix is what investors are watching, because a lead that shrinks quarter after quarter changes the story a company can tell on an IPO roadshow.
There is also a quieter threat in the background. Investors told Reuters that the rise of open-source and open-weight models, which lower token costs and let companies build their own AI infrastructure, may be a bigger challenge than the race between the two commercial leaders. Meta Platforms has been among Anthropic’s largest customers, but people familiar with the matter say Meta is looking to reduce its use of Anthropic’s models as it develops more AI capability internally. Meta did not respond to a request for comment.
The IPO math
Anthropic filed confidential IPO paperwork in June and chose Nasdaq for the listing. The schedule has slipped. Reuters previously reported that marketing was expected to begin in mid-October at the earliest, and the company could now push the offering to after the November US midterm elections, according to two people familiar with the matter. The prospectus is expected in late September, though the sources cautioned the timing remains subject to change.
The expected size is what makes the timing matter beyond Anthropic. Investors have speculated the listing could value the company at around $2 trillion and raise more than $100 billion, which would make it the largest IPO ever attempted. SpaceX went public in June at a $1.77 trillion valuation, raising $85.7 billion. Anthropic is also finalizing a $15 billion revolving credit facility as part of the process, and Morgan Stanley, Goldman Sachs, JPMorgan and Citigroup are among the underwriters. Reuters noted the company may face a tighter window than usual between analyst meetings and the prospectus release because the covering analysts already know the business well.
OpenAI, meanwhile, has taken itself out of the 2026 window. Sam Altman confirmed on Saturday that the company will not go public this year, saying concerns around AI safety make it an ill-advised time for a listing. That leaves Anthropic racing to market against a rival that just took share in enterprise spending, which is exactly the dynamic a new model launch would address.
Safety politics cut both ways
The Amodei essay has already produced legal complications. A California antitrust suit filed this week accuses Anthropic, OpenAI and Google of agreeing to slow AI development in violation of the Sherman Act, turning the safety essay into evidence in a competition-law case. OpenAI’s policy chief Chris Lehane said last week that the three companies have been coordinating on safety matters for several weeks and that no antitrust waiver is needed for that work. He framed any industry-led standards as complementing, not replacing, mandatory federal safeguards.
A new Anthropic release would also collide with the White House position. President Trump announced an “AI Force” and an AI czar this week while calling safety concerns a hoax, and the administration has pushed back against regulation calls from Congress and from the labs themselves.
None of the sources said when or whether the launch would happen. The company declined to comment to Reuters. What is clear is that Anthropic faces a choice between its stated position on pace and its market position ahead of the largest IPO in history, and the two are pulling in opposite directions. Investors get the prospectus in late September either way, and the model question will likely be answered before the roadshow begins.