By: Joyce Jia

In early December 2025, the Financial Times reported that Anthropic retained the Palo Alto-based law firm Wilson Sonsini to help facilitate a public offering as early as 2026. The move may signal a strategic effort to outpace rival OpenAI, which had also been eyeing a 2026 listing. Following the report, OpenAI executives privately expressed concern about being beaten to market. Regardless of who lists first, 2026 is shaping up to be a landmark year for AI in public markets, and a long-awaited opportunity for investors and observers to lift the veil on the economics of the generative AI giants driving the Fourth Industrial Revolution.

More than an IPO: A Governance Blueprint for the AI Era

As AI systems increasingly power healthcare systems, financial markets, communications, and public services, they are functionally becoming part of the nation’s core infrastructure. Unlike pharmaceuticals subject to FDA approval, financial intermediaries registered with the SEC, or telecommunications providers licensed by the FCC, frontier AI developers operate without a comparable, unified regulatory gatekeeper. Existing oversight has been fragmented across state privacy regimes and FTC consumer protection authority, leaving the public with significant gaps in understanding how these companies operate, what their models optimize for, and how deeply they depend on cloud infrastructure, semiconductor supply chains, and energy capacity. Moreover, propelled by unprecedented growth trajectories, they have financed their expansion almost entirely through venture capital and private placements, without being subject to public disclosure obligations and public accountability that securities registration demands. 

When Anthropic or OpenAI files for its initial public offering, it will become the first frontier AI company to submit its business model to the full discipline of SEC disclosure. The filing will not merely determine one company’s market valuation. It will likely shape the disclosure template and regulatory precedent governing how artificial intelligence integrates into public capital markets, compelling for the first time a legally enforceable accounting of how frontier AI companies actually operate, what system risks they pose, and whether traditional corporate governance frameworks are adequate to contain them. 

Tip of the Iceberg: What SEC Disclosure Will Finally Force Frontier AI to Reveal

The S-1 registration statement must include all information specifically required by Form S-1, as well as any material information needed to prevent the included statements from being misleading. Current SEC disclosure requirements do not specifically address modern AI risks, though the SEC’s Division of Examinations has identified AI as a priority focus area. Under established materiality doctrine, disclosure is required for any matter to which a reasonable investor would attach importance in deciding whether to purchase a security. Furthermore, Item 105 of Regulation S-K requires a “Risk Factors” section discussing material factors that make an investment speculative or risky, with a concise explanation of how each risk affects the registrant.

Under this framework, investors will likely see for the first time the size of an LLM company’s accumulated deficit and its projected path to profitability. They will also gain visibility into the terms of related-party contracts and material agreements with cloud and semiconductor suppliers who serve as both vendors and strategic investors. Disclosure will additionally include training data liabilities such as the Bartz v. Anthropic settlement. Finally, “Business” and “Management’s Discussion and Analysis” sections may provide a first-of–its-kind monetization narrative, explaining what these companies’ algorithms are optimized for, what commercial incentives are encoded in model training, and how ecosystem allocation decisions generate indirect value. Together, these disclosures will expose the structural economics of frontier AI development in a way that no private financing round has ever required.

Disclosure priorities will also differ across companies. Anthropic’s Long-Term Benefit Trust grants outside directors veto power over decisions conflicting with the company’s safety mission, even where those decisions would maximize shareholder value. This structure directly challenges the shareholder primacy doctrine established in Dodge v. Ford, the 1919 Michigan Supreme Court decision holding that corporate directors must prioritize shareholder returns over broader social objectives. The arrangement raises a critical question: Can mission-driven governance survive activist investors and proxy battles once the company goes public? 

OpenAI, having completed its conversion to a Delaware public benefit corporation in October 2025, faces a structurally distinct set of disclosures: the terms of its nonprofit’s retained equity stake, the scope and exclusivity of its Microsoft dependency, and the resolution of Elon Musk’s active litigation challenging the conversion will each require disclosure under existing SEC frameworks in configurations not previously tested at this scale.

Materiality Standards for the AI Future

Much of what frontier AI companies should disclose is already compelled by existing SEC rules. What is needed is rigorous application of the comment process to establish materiality standards specific to the AI sector. Anthropic published a new model Constitution last month articulating how its AI systems should reason and prioritize competing values. OpenAI, for its part, has committed to a governance structure that retains a nonprofit equity stake precisely to preserve the primacy of safety and security obligations even after its corporate restructuring. Both are well positioned to lead on disclosure voluntarily, setting best practices that regulators and competitors will be compelled to follow. That makes these securities filings as consequential as any AI safety research. Markets, no less than models, need alignment.