OpenAI and Anthropic File for IPO: What the S-1 Filings Mean

The two leading AI labs are moving toward the public markets in near lockstep. Anthropic submitted a confidential draft Form S-1 to the SEC on June 1, 2026, and OpenAI followed with its own confidential filing days later. Both companies framed the step in standard language: the filing creates the option to go public after SEC review, with any offering depending on market conditions. No share counts, prices, or valuations were set.

What changed

A confidential S-1 is the formal on-ramp to an IPO. It lets a company begin the SEC review process privately before deciding whether to proceed. Anthropic's announcement was explicit that the number of shares and the price are not yet set, and stressed the filing is not an offer to sell securities. OpenAI's filing carried the same procedural character.

What makes this notable is the synchronization. Two competitors that have raised enormous private capital are now both signaling they want access to public markets. Reporting around the filings, including by Axios, places them alongside other large 2026 listings and frames a wave of high-profile tech IPOs. Public commentary has attached large forecast valuations to both companies, in the rough vicinity of a trillion dollars each, though those figures come from analysts and market chatter rather than the confidential filings themselves.

Why it matters and for whom

Going public changes the rules. Private AI labs have operated with patient capital, heavy losses, and limited disclosure. Public markets demand transparency, recurring scrutiny, and a credible path to profitability. The S-1 process forces these companies to articulate their economics in a way they have largely avoided.

For investors, the central tension is cost versus growth. Frontier AI runs on extraordinary compute spending and long-term infrastructure commitments. Revenue has grown fast, but so has burn. The public-market question is whether the growth narrative survives contact with disclosed numbers on margins, capital expenditure, and the timeline to free cash flow. Hype that worked in private rounds gets tested against an S-1's financial statements.

For partners and enterprise customers, an IPO means more predictable, regulated reporting and a clearer view of each company's financial footing, which matters when you are betting critical workflows on a vendor. For the broader sector, two trillion-dollar-scale AI listings would tie a large slice of public-market value to the AI thesis, raising the stakes if growth disappoints.

What to watch

When the filings become public, focus on a short list. First, revenue quality and run-rate: how fast is it growing and how durable is it. Second, burn rate and the path to profitability: how much cash is consumed and when, if ever, the company expects positive free cash flow. Third, compute commitments: the size and length of infrastructure and chip obligations, which are the dominant cost and the biggest fixed risk. Fourth, enterprise adoption, which is the most defensible revenue line. Fifth, regulatory risk, given that frontier models now attract direct government attention.

The timing also matters. Both companies preserved optionality by filing confidentially, so an actual listing depends on market conditions and could slip. The filing is a signal of intent, not a calendar commitment.

There is a strategic reason both labs are moving now rather than waiting. Going public unlocks a deeper, more liquid pool of capital than private rounds can sustain indefinitely, and the compute bills these companies face are growing faster than almost any private market can comfortably fund. A public listing also gives early employees and investors a path to liquidity and hands the company a tradable currency for acquisitions. Filing first carries a competitive edge too: the lab that lists earlier sets the reference point investors use to value the other, which is part of why the near-simultaneous filings read as a race as much as a coincidence. For observers, the contest is no longer only about who has the better model but about who can convince public markets that the economics of frontier AI eventually work.

Bottom line

• Anthropic and OpenAI both filed confidential draft S-1s in June 2026, the first formal step toward IPOs, with no prices or valuations set.
• The move forces frontier AI economics into the open: investors will judge revenue durability, burn rate, compute commitments, and the path to profit rather than benchmark scores.
• Confidential filing preserves optionality, so timing remains conditional on market conditions.

The coming public filings will be the first real test of whether the market can price intelligence as a business, not just a breakthrough.