Anthropic is continuing preparations for a potentially historic public offering this year even as CEO Dario Amodei urges the AI industry to slow the advance of frontier models long enough for safety measures and independent oversight to catch up.
By yourNEWS Media Newsroom.
Anthropic is still moving toward a 2026 initial public offering despite CEO Dario Amodei’s increasingly urgent warnings about the potential dangers of advanced artificial intelligence, creating an unusual collision between the company’s safety campaign and its ambitions on Wall Street.
Multiple people familiar with the company’s plans told Axios that the recent surge in debate over catastrophic AI risks has not changed Anthropic’s intention to become a publicly traded company this year.
Reuters separately reported that Anthropic has selected Nasdaq for its potential listing and that preparations remain underway for an offering that could begin marketing as soon as mid-October, although timing remains subject to market conditions and regulatory review. The offering has been discussed at a valuation potentially approaching $2 trillion.
That trajectory has remained intact even after Amodei publicly called on the industry to reduce the speed at which frontier AI capabilities are advancing.
In his widely circulated proposal for pacing AI development, Amodei made clear that he is not calling for companies to stop training increasingly powerful models altogether.
“To be clear, pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models, and for third party evaluators to confirm this,” Amodei wrote.
His framework calls for three escalating levels of coordination.
The first would require frontier AI developers to permit independent evaluators to operate inside their companies with access comparable to employees responsible for risk assessment. Anthropic says it intends to implement that system itself.
The second would require major AI developers in democratic countries to coordinate on common safety practices rather than racing ahead independently.
The third — and most difficult — would attempt to establish international coordination that could eventually include governments outside democratic countries, providing some means of limiting a competitive race in which every nation believes it must accelerate because its rivals are doing the same.
Anthropic’s first commitment is particularly unusual.
Amodei said outside safety evaluators should receive company laptops, building access and access to internal tools and workspaces similar to employees conducting risk assessments. Those evaluators would be permitted to publish significant findings about risks, incidents and whether Anthropic actually followed its publicly stated safety procedures.
The company would retain limited rights to protect legally privileged, commercially sensitive or customer-confidential information but would not be permitted to suppress unfavorable safety conclusions simply because they were damaging to Anthropic.
Amodei argued that such independent oversight would make safety commitments verifiable rather than leaving AI companies responsible for grading themselves.
“Regardless of what commitments we make, the public deserves to know what is going on,” he wrote.
That philosophy is now being used internally to support the case for an IPO rather than delay it.
People familiar with Anthropic’s thinking told Axios that becoming public could subject the company to additional transparency, disclosure and investor scrutiny. The argument is that those obligations could complement the company’s push for greater accountability rather than conflict with it.
Anthropic’s business prospects also have not materially changed during the safety uproar, according to the report, nor has investor demand disappeared.
Reuters reported last week that Nvidia was discussing becoming an anchor investor in Anthropic’s IPO, potentially investing as much as $10 billion in an offering that could raise up to $100 billion. The details remain under discussion and neither company has publicly confirmed the arrangement.
The contrast with OpenAI has become increasingly pronounced.
OpenAI CEO Sam Altman said Saturday that his company will not go public in 2026, telling Fortune that current concerns surrounding AI safety make this an inappropriate time for an IPO.
“I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don’t feel pressure on that,” Altman said.
Asked directly whether an IPO could occur this year, he replied, “I would say not 2026, yeah. We got a lot of stuff to do.”
Axios reported that OpenAI had already been leaning toward a 2027 listing, and current developments have not changed that direction. Reuters similarly reported that Altman is prioritizing safety and alignment work before taking the company public.
Amodei’s decision to continue toward an IPO while simultaneously advocating slower frontier development has fueled skepticism among critics who question whether the safety campaign could produce financial advantages for established AI companies.
Training frontier models requires enormous amounts of computing power and capital. A slower development cycle could therefore reduce spending pressures at companies such as Anthropic while giving established firms additional time to monetize existing models.
Critics have also raised concerns about regulatory capture — the possibility that the largest AI companies could help shape rules that smaller competitors would have difficulty satisfying.
Federal Trade Commission Chairman Andrew Ferguson expressed skepticism Tuesday about AI firms seeking government regulation while also discussing exemptions from antitrust law that might permit competing companies to coordinate development restrictions. Ferguson said proposals allowing incumbent firms to cooperate on pacing require scrutiny because they could potentially create barriers to competition.
Those concerns do not establish that Anthropic’s safety warnings are financially motivated. Amodei has repeatedly argued that increasingly autonomous systems present genuine risks that require action before capabilities advance beyond the ability of companies and governments to control them.
Other leading AI executives have recently moved closer to his position.
Altman, Elon Musk and Google DeepMind CEO Demis Hassabis have publicly supported slowing or more carefully pacing frontier development, producing an unusual degree of agreement among companies that normally compete intensely for researchers, computing capacity, customers and technical leadership.
Amodei has warned that future systems could potentially coordinate large numbers of autonomous AI agents, exploit cybersecurity vulnerabilities and improve their own capabilities faster than safeguards can be developed.
Those warnings gained additional attention after researchers inside the industry publicly expressed fears that AI could escape meaningful human control. Reuters reported Tuesday that former Google DeepMind researcher Bilal Chughtai joined former Anthropic researcher Jacob Coxon and others in warning of potentially catastrophic consequences if AI development proceeds without adequate safeguards.
Investors have reacted more cautiously than some of the rhetoric might suggest.
Technology and semiconductor shares fell after Amodei, Altman and Musk backed slower development, as investors considered whether a reduction in AI spending could affect chipmakers, data centers and the enormous infrastructure buildout supporting the industry.
AI-related capital spending is expected to approach $800 billion this year and exceed $1 trillion in 2027, according to estimates cited by Reuters, meaning even a moderate reduction in the pace of frontier development could have consequences extending far beyond AI laboratories themselves.
Yet the prospect of Anthropic becoming public has not disappeared.
The successful June IPO of SpaceX, which also includes exposure to Musk’s AI operations, demonstrated that investors remain willing to place enormous valuations on companies pursuing highly capital-intensive technologies. SpaceX raised a record $75 billion at a $1.77 trillion valuation and surged to roughly $2.1 trillion in market capitalization after its Nasdaq debut.
Alphabet, whose Google DeepMind unit is one of Anthropic’s leading AI competitors, has operated for years under public-company reporting requirements.
Those examples do not resolve whether an AI developer becomes safer by entering the public markets. Public ownership can increase financial disclosure and shareholder scrutiny, but it can also create additional pressure for revenue growth and returns.
That tension is becoming increasingly important as Anthropic prepares to ask public investors to finance a company whose CEO simultaneously argues that the industry may need to deliberately slow its most advanced technological work.
President Donald Trump has taken the opposite position on the underlying safety debate.
Trump dismissed warnings about catastrophic AI risks Monday and questioned why executives would support regulations that he said could damage the companies they run.
“AI taking over the World, destroying Humanity, and all other things bad, is a HOAX,” Trump wrote. “No different from RUSSIA, RUSSIA, RUSSIA — UKRAINE, UKRAINE, UKRAINE — IMPEACHMENT HOAX #1 — IMPEACHMENT HOAX #2 — and all of the other HOAXES and SCAMS that America was forced to endure through the Destructionists’ and Deviants’ foul play and illegal conduct.”
Trump has argued that slowing U.S. AI companies would benefit China and has repeatedly emphasized maintaining American leadership in the technology.
The administration’s stance increasingly differs from that of several executives building the most advanced systems. Amodei, Altman and Musk have all acknowledged significant risks from continued rapid development, even while their companies remain competitors in what could become one of the largest commercial markets in history.
That disagreement is now moving from technical discussions into financial markets and federal policy.
Members of Congress are considering proposals that could require frontier developers to demonstrate safeguards against catastrophic risks, permit government testing of advanced systems and potentially block models deemed unacceptably dangerous. Those proposals remain under debate and face opposition from Trump and others who argue existing government authority is sufficient.
Anthropic, meanwhile, appears determined to pursue both objectives at once.
The company is asking the industry to give safety mechanisms more time to catch up with increasingly powerful models while simultaneously preparing to enter the public markets at a valuation that could rank among the largest IPOs ever attempted.
For Anthropic, the argument is that those positions do not conflict: slowing the most dangerous aspects of frontier development does not require slowing the company’s growth, and becoming public could force the kind of transparency Amodei says the AI industry needs.
Whether investors, regulators and the public ultimately accept that distinction may become one of the defining questions surrounding Anthropic’s expected 2026 market debut.
Original article
Comments