Anthropic confidentially files draft S-1, putting OpenAI on a public-listing clock

Anthropic confidentially submitted a draft S-1 registration statement to the SEC on June 1, 2026, five days after closing the $65 billion Series H that took it to a $965 billion valuation. The filing makes Anthropic the first of the frontier AI labs to clear the regulatory bar for a U.S. listing, ahead of OpenAI’s expected filing in the same quarter and roughly two weeks behind SpaceX’s much larger public prospectus. The earliest realistic listing window is the fall of 2026, contingent on SEC review and market conditions that Anthropic flagged in its own Rule 135 notice.

What is on the public record

The S-1 itself is not public. Confidential filings under the JOBS Act let emerging growth companies work through SEC review without disclosing financials until shortly before the road show. Anthropic’s statement confirmed only the filing itself, that share count and pricing remain unset, and that the listing depends on market conditions. Everything else in the public record is reporting around the filing rather than from it.

The recent operating disclosures do most of the position-setting work. Run-rate revenue crossed $47 billion in May, up from roughly $10 billion in annual revenue at the end of 2025. Anthropic has told investors that the run-rate would clear $50 billion by the end of June. Q2 2026 revenue is expected to come in at about $10.9 billion, more than double the prior quarter, and Anthropic is reportedly on track for its first profitable quarter. Fintech firm Ramp reported that more businesses paid for Anthropic services than for OpenAI services for the first time in May. That single data point is the one most likely to anchor the S-1 narrative when the filing eventually goes public.

The compute side is unusually concrete. Anthropic’s agreement with SpaceX to use Colossus 1 GPU capacity runs at roughly $1.25 billion per month through May 2029 (per SpaceX’s own prospectus), with a 90-day mutual termination clause. The Series H announcement disclosed up to 5 GW of new AWS capacity, 5 GW of next-generation TPUs through Google and Broadcom, and additional Colossus 2 GPU access. Those numbers translate into a multi-tens-of-billions-per-year compute line the IPO will need to underwrite directly.

Where this lands in the AI capital sequence

The competitive frame is the one Bloomberg’s headline pointed at. Anthropic and OpenAI are now on overlapping IPO calendars. OpenAI is reportedly preparing its own confidential filing within weeks. SpaceX has already filed publicly and is targeting roughly $75 billion at a $1.75 trillion valuation. Three AI-or-AI-adjacent mega-listings concentrated in the same six-month window is the actual market story. Whoever prices last absorbs whatever fatigue the prior tape sets up.

The Anthropic-versus-OpenAI ordering matters for a softer reason. OpenAI sits at an $852 billion private valuation from March, and its public listing will be priced against Anthropic’s tape if Anthropic gets there first. That is the kind of sequencing detail investors usually shrug at, but the gap between the two companies’ private marks has narrowed enough that order of listing affects price discovery meaningfully. Sam Altman publicly disputed the “race to IPO” framing this week, saying the timing is “when we think it makes sense” rather than a competitive deadline. The filings themselves will adjudicate that interpretation.

Two operating wildcards sit on top of the IPO narrative. Anthropic remains in active dispute with the Pentagon over a supply-chain-risk designation that led defense contractors to drop its models. The financial impact Anthropic itself flagged is “billions of dollars in revenue” at risk, though private-sector growth has more than offset the loss so far. Separately, Claude Mythos Preview, the cybersecurity-focused model Anthropic released to a limited set of partners under Project Glasswing, is the model Anthropic has been telling Washington about. A wider rollout before pricing would meaningfully expand the addressable-market story; a delayed rollout leaves the cybersecurity line as forward-looking commentary in the S-1.

What’s worth watching

1. OpenAI’s confidential filing. Reuters and Bloomberg both reported a filing is imminent. If OpenAI files within the same quarter, the two listings move onto a parallel SEC-review track and the pricing of the second one becomes the cleaner read on private-AI valuation discipline.

2. Public S-1 reveal timing. Confidential filings can stay confidential until shortly before the road show. The earliest plausible public reveal is the second half of August. The interesting unknowns are gross-margin disclosure, customer concentration (Amazon, Bedrock, AWS-related), and how Anthropic accounts for the Series H strategic commitments inside revenue versus deferred-revenue lines.

3. Pentagon resolution. A negotiated settlement or a successful challenge to the supply-chain-risk designation would close out one of the largest open risk factors before pricing. A continued standoff pushes the risk straight into the prospectus and frames the listing’s investor conversation accordingly.

Stackmaven’s follow-up on the IPO progression lands on or around September 3, 2026.

Sources cited

1. Anthropic: Anthropic confidentially submits draft S-1 to the SEC www.anthropic.com
2. CNBC: Anthropic confidentially files IPO prospectus with SEC www.cnbc.com
3. TechCrunch: Anthropic files to go public techcrunch.com
4. Bloomberg: Anthropic Files Confidentially for IPO in Race With OpenAI www.bloomberg.com
5. TechCrunch: Ahead of its IPO, Anthropic's Daniela Amodei shrugs off doubts about AI's returns techcrunch.com