SpaceX comparison and Anthropic’s mitigation message frame the disclosure
Anthropic’s IPO filing places a broad set of operational and financial risks before prospective shareholders. Of the 335 pages, 93 are devoted to “risk factors,” the things that could hurt potential shareholders’ investments. For comparison, rival SpaceX’s listed risk factors took up 38 of more than 250 pages. The San Francisco company’s document also covers its finances, executive compensation, dividend policy and similar material, while identifying risks involving computing resources, data-center development, major customers and distribution channels.
Compute—the computing power or resources needed to build, test and improve AI models—is expensive and in high demand, the filing says. Anthropic warns that if it cannot secure adequate compute, or fails to direct compute effectively across its workloads, it may be unable to train models that remain competitive with competitors’ models or adequately serve customers.
Anthropic also identifies community opposition to data centers as a business risk. It says development has prompted public controversy and regulatory scrutiny regarding environmental impact, land and water use, and allocation of limited power resources. If regulators impose new requirements or charges as a result, the filing says Anthropic’s ability to do business could suffer.
The company relies on a small number of customers, including SpaceX and Microsoft, and on distribution channels for a large portion of revenue. Anthropic said that if just one major customer scales back its business or drops Anthropic altogether, the company’s prospects and finances could suffer.
Anthropic presents the IPO as a step toward mitigating the AI risks it believes the technology poses. In a letter to prospective shareholders, the company’s founders wrote, “To build AI is to harness cognition itself, and anything so powerful can be used for ill as well as good.” They added: “Anthropic is a deeply values-driven organization, founded in large part to help the world realize AI’s benefits and avoid its dangers.”
The filing also addresses the possibility that models could develop so quickly that Anthropic cannot implement adequate safety measures. Anthropic said the gap could “increase the likelihood of safety incidents, regulatory intervention, or reputational harm, and could adversely affect our business.” Because AI models now contribute to Anthropic’s own research and development, the company said AI might progress “beyond our ability to evaluate and safeguard new models,” which could adversely affect its business.
In September, CEO Dario Amodei warned that within six to 12 months AI could be capable of leading a swarm of agents that could take over the entire internet. The warning came as worries grew inside and outside the industry about systems that solve problems beyond human capacity but might also act on their own to carry out other, more harmful tasks. The filing separately said that as AI systems “become more autonomous and capable of taking consequential actions with less human oversight,” they could “take harmful actions independently and at scale without human intervention.”
The filing’s broader risk discussion included the prospect of bad actors using AI models to build deadly weapons. Its more specific catastrophic-risk discussion was conditional on safeguards and oversight and included the risk from systems allowed to operate with a high degree of autonomy. Anthropic said advanced AI systems, if not developed and operated with enough safeguards and oversight, “pose catastrophic or existential risks to humanity, particularly if such systems were misused, replicated, weaponized, or allowed to operate with a high degree of autonomy.” Anthropic also said it would be harmed if use by bad actors caused damage or if its model guardrails or safeguards failed to prevent such misuse.
The filing described the prospect of bad actors using AI models to build chemical weapons and authoritarian governments using the technology to suppress citizens. In September, the company disclosed it had blocked attempts to use its models for malicious activity such as cyberattacks, surveillance and research that could have led to biological weapon development.
“Even if our own models do not cause such harm, a significant safety incident involving any frontier AI could trigger broad reputational damage to the AI industry, public backlash, and sweeping regulatory responses that could adversely affect our business,” Anthropic said.
The filing also presents a wider mix of risks. They range from the possibility that Anthropic cannot retain key staff or navigate regulatory roadblocks to risks involving cyberattacks and the spread of dangerous misinformation.
Associated Press characterized the combination of AI’s enormous risk and enormous potential as Anthropic’s main selling point. Associated Press reported that Anthropic has tried to position itself as a responsible AI company and courts business customers by emphasizing safety-mindedness. In the filing, Anthropic states, “We believe that frontier AI companies must invest in safety and alignment to help ensure the responsible diffusion of this technology.”
Brian Mulberry, chief market strategist at Zacks Investment Research, said IPO deals have to display “very clearly” what risks people face when they invest their money in a newly public stock. He called the U.S. capital markets’ disclosure requirement a strength, saying a company listing on a U.S. exchange must outline those risks clearly so people can decide whether to participate by offering their hard-earned money and help finance the deal.
AP Business Writer Bernard Condon and AP Technology Writer Kaitlyn Huamani contributed to this story.