Three AI labs, one queue for the public markets — and the bill hasn't come due yet
OpenAI confirmed on 8 June 2026 that it has confidentially filed paperwork for a US initial public offering, putting the ChatGPT maker in line behind a small but growing queue of frontier-AI labs that have decided their next act is a public listing rather than another private round. The filing came a little over a week after rival Anthropic filed its own confidential IPO documents, according to the company and to reporting carried by Reuters and the BBC, and it lands on a market that has spent two years repricing what an AI company is actually worth.
Three things are now true at the same time. The two most capital-intensive AI labs in the West have told the Securities and Exchange Commission they intend to sell shares to the public. A third — Perplexity, the AI-search upstart — has told CNBC, as relayed by Reuters, that it plans to go public in 2028 "regardless of what happens" to either of those listings. The financial press is reading the cluster as a referendum on the sector; the labs are reading it as a financing event. The difference matters.
The filing
OpenAI's announcement, carried on its own corporate blog and reported by TechCrunch on 8 June 2026 and by BBC News the same day, did not disclose a target valuation, a share count, or a tentative exchange. A confidential submission lets a company circulate a draft registration statement to the SEC's reviewers without putting the document on the public file, which is the standard route for issuers who want to test regulator and investor temperature before committing to a roadshow. Anthropic used the same mechanism when it filed, by OpenAI's own description, "a little more than a week" earlier. The two filings mean that, for the first time, the two firms most often described as the US frontier-AI duopoly are running the same SEC gauntlet in the same calendar quarter.
That is the part of the story the wires are correctly emphasising. Less remarked is the timing: the filings arrive into a US equity market that has, over the last twelve months, shown a marked preference for AI-adjacent revenue over AI-adjacent promise. The reception that actually greets these papers — the book-building spread, the anchor allocations, the post-IPO drift — will be the first hard data point on whether the public markets agree with the private marks that have propped up the sector since 2023.
The third queue
Perplexity's claim, reported by CNBC and surfaced on 8 June via Polymarket's news feed, is the more revealing of the three. The company's stated 2028 timeline is conditional on nothing — not on OpenAI pricing, not on Anthropic clearing the SEC, not on the Federal Reserve's rate path. That posture tells you something specific about how a category-defining challenger is reading the moment. If Perplexity can hold its narrative of an "AI-native search engine" with measurable query share through 2027, a 2028 listing on its own terms is plausible; if it cannot, no amount of waiting for the leaders to misstep will save the story.
The interesting structural point is that the queue itself is now a competitive instrument. Investors who want AI exposure are being told they have at least three near-term vehicles, and that the third is willing to come to market last and on a separate cadence. That changes the negotiating dynamic in every private round between now and those dates. Capital that was previously locked in by scarcity — "if you want frontier, this is the only ticket in town" — is now visibly fungible.
What the filings don't say
A confidential submission is, by design, an opaque document. Neither OpenAI nor Anthropic has published the contents of its draft registration statement, and the financial press has not yet reported a target valuation, a primary-versus-secondary split, or a lock-up structure for either filing. The public cannot yet see how each company intends to characterise its compute commitments, its revenue concentration, its dependency on a small number of cloud providers, or the structure of its existing investor agreements. Those are exactly the items a public-market investor will eventually price, and they are exactly the items the labs currently have an incentive to keep out of the headlines.
The most plausible alternative read of the cluster is that the filings are defensive, not opportunistic. A confidential submission in a soft tape is, among other things, an option: it preserves the right to go public when conditions improve, without paying the reputational cost of being seen to have retreated from a prior IPO timetable. The bullish read — that the AI sector has cleared its 2024–2025 scepticism phase and is now ready to be marked in continuous trading — cannot be tested until at least one of these companies actually prices.
Stakes
If even one of these listings clears at a valuation the market holds, the second-order effect is a wave of follow-on issuance from every well-funded AI startup that has been told, for two years, that private liquidity is a substitute for a public listing. If both clear and drift, the sector re-enters a venture-funded winter whose length is set by the next decent secondary market. The compute, the model weights, the research staff — none of that is determined by the listing. What is determined is who, from late 2026 onward, gets to mark the sector's price on a lit exchange rather than in a venture-capital term sheet. That is a small, technical, and genuinely consequential change.
Desk note: Monexus framed this as a financing-cluster story rather than a technology story. The wires are leading on the "AI race" angle; the more durable fact is the timing and sequencing of the three filings.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- http://reut.rs/3PQhu57
- https://x.com/Reuters/status/0