Anthropic confidentially filed its draft S-1 with the US Securities and Exchange Commission on Monday, June 1. OpenAI filed its own a week later, on June 8. Two frontier AI labs, the only two that have ever been valued above half a trillion dollars, are now racing to be the first trillion-dollar AI company to list on a public exchange. Anthropic’s last private valuation is $965 billion. OpenAI’s is $852 billion. Both are short of a trillion. The roadshow will probably fix that.

The filings were not headline news when they happened — both are confidential, under the JOBS Act’s draft-registration rules — but the implications for the AI industry, for public markets, and for the upcoming G7 summit in France (where both CEOs will be sitting down with heads of state on Monday) are considerable. Two companies with no audited public financials, no profitability on the horizon before 2028, and a combined valuation north of $1.8 trillion are about to ask retail investors to take the other side of every position the late-stage VCs have been holding.

🔍 THE BOTTOM LINE

Two AI labs, both unprofitable, both private, both built on the same bet that foundation models will eat enterprise software, have filed to go public within seven days of each other. Anthropic got to the SEC first because its cap table is cleaner and its valuation is higher. OpenAI followed because Sam Altman cannot let Anthropic list first. The order matters for the story both CEOs will tell on the roadshow. The substance is the same: both are asking the public markets to value a future that has not arrived yet.

The Filing Order Matters

Per filings and reporting, Anthropic filed on June 1, four days after closing a $65 billion Series H at a $965 billion post-money valuation. OpenAI filed on June 8, seven days later, on the heels of a $122 billion round at $852 billion in March.

The order is a narrative tell. Anthropic, the younger company, less famous, more enterprise-focused, will go to the public markets as the more valuable of the two. That is a first in the AI race, which OpenAI has led on every front until now. For Anthropic, the timing locks in the “we passed OpenAI” story before the public markets get to see OpenAI’s audited financials, which will look very different. OpenAI’s reported annualized revenue is roughly $24 billion, double Anthropic’s. OpenAI’s consumer footprint (ChatGPT 800M+ weekly active users) dwarfs Anthropic’s (claude.ai ~150M MAU). But the market is currently pricing the future, not the present, and on the future Anthropic has a higher multiple.

A confidential S-1 under the JOBS Act lets both companies lock in SEC feedback, fix disclosure issues, and adjust timing without making their numbers public. The filings are real, but they are not yet visible to retail investors. The 15-day public-filing window before roadshow is the next milestone.

The Numbers Side By Side

The consensus from Goldman, Morgan Stanley, and the prediction markets is that Anthropic will list in October and OpenAI will follow two to four months later. If the public markets accept the same multiples the private rounds commanded, both listings will be the largest AI IPOs ever. Combined with SpaceX’s $1.77T listing earlier this week, 2026 is on track to be the largest IPO year in Wall Street history.

Why Both Are Going Now

The real reason is the same reason SpaceX went first: liquidity. The late-stage VC funds that backed these companies at $50B, $100B, $300B valuations need to return capital to their own LPs. They cannot do that while the companies stay private. Both Anthropic and OpenAI are now sufficiently large, sufficiently known, and sufficiently stable that the public markets are willing to absorb them — even at a discount to the last private round. The cost of waiting is that the public markets might decide they are not willing to absorb them at any price. The cost of going now is the discount. Both CEOs have decided the discount is smaller than the risk of a market re-pricing.

A second reason: the competitive gap is closing. Anthropic’s Claude Fable 5, launched the day after OpenAI’s S-1 filing, was the first publicly available Mythos-class model. It scored 80.3% on SWE-Bench Pro. The two companies’ frontier models are now close enough that “who’s better” is a marketing question, not a technology one. The public market will reward whichever company tells the more compelling business-model story on the roadshow. That is a different kind of competition than the one the labs have been running for the last three years.

The SpaceX Distraction

The third reason, less talked about, is SpaceX. Elon Musk’s other company began trading on Nasdaq on June 12 at a $135 share price and a $1.77 trillion valuation. That is, on its own, the largest IPO in history. It also means the public markets are about to be flooded with Musk-issued paper. Anthropic and OpenAI cannot afford to compete with SpaceX for the same retail-investor liquidity. The faster Anthropic and OpenAI can list, the more they can capture retail enthusiasm before it gets absorbed by the Musk phenomenon. Goldman’s latest note projects total 2026 US IPO proceeds north of $160 billion, more than quadruple 2025’s level. The flood is real.

🇳🇿 NZ Angle

New Zealand retail investors cannot buy Anthropic or OpenAI shares directly at IPO. We do not have the kind of retail brokerage infrastructure that gets US-listed stocks on day one, and our regulators are more conservative about local investors buying into foreign listings. That means Kiwis who want exposure to the AI public-market boom will probably do it the indirect way: through the local funds that hold US tech, through companies that supply into the AI infrastructure chain (the NZ companies selling data centre cooling, fibre, or specialised engineering services to the hyperscalers), or through the handful of KiwiSaver providers who are already overweight US tech. The trillion-dollar AI IPOs will, for most Kiwis, be a thing they read about in the business pages, not a thing they participate in. That asymmetry — being economically exposed to the AI bubble without being able to invest in it directly — is one of the more under-reported facts about being a small open economy in 2026. The companies that will benefit most from Anthropic and OpenAI going public are not the AI labs themselves. They are the chipmakers, the cloud providers, and the energy utilities that the labs have to keep buying from to justify their valuations. Some of those are listed. Some of those are NZ-adjacent.

⚖️ The Other Side

It is also fair to note that the “trillion-dollar AI bubble” framing is not the only reading. Both companies have real revenue, real customers, and real products. Anthropic’s $47 billion run rate is a 4× jump from the start of the year. OpenAI’s $24 billion annualised is the fastest ramp in enterprise software history. The public markets are not being asked to buy a science project. They are being asked to buy two companies that are already generating tens of billions of dollars a year, growing at 10× annually, with the kind of margin profile that, if it holds, will produce some of the most profitable companies of the 21st century. The “bubble” is a real risk, but the “category-defining business” is also a real possibility. Both CEOs are betting the second story will dominate the first once the S-1 numbers are public. That is the bet worth watching.

❓ FAQ

Q: What is a confidential S-1, and why are both companies filing one?

A: A confidential S-1 is a draft registration statement filed with the SEC under the JOBS Act. It lets a company engage with SEC examiners privately, fix disclosure issues, and adjust timing without making its financials public. It is standard practice for high-profile tech IPOs. Both Anthropic and OpenAI used it because they want SEC feedback locked in before the public filing window, which opens 15 days before the roadshow.

Q: Which company will list first?

A: The consensus is Anthropic in October 2026 and OpenAI in late Q4 2026 or Q1 2027. Polymarket has Anthropic at 92.5% to list by November 1, OpenAI at around 30%. Bank guidance from Goldman and Morgan Stanley points to a two-to-four month gap between the two listings.

Q: How do the valuations compare?

A: Anthropic’s last private valuation is $965 billion (May 2026 Series H). OpenAI’s is $852 billion (March 2026). Both are expected to cross the $1 trillion mark in the IPO roadshow. By comparison, SpaceX listed on June 12 at $1.77 trillion, the largest IPO in history.

Q: Which company is more profitable?

A: Neither is profitable yet. Anthropic’s break-even target is 2028, OpenAI’s is 2030. But Anthropic generates roughly $0.23 in annualised recurring revenue for every dollar it has raised — about double OpenAI’s ratio. OpenAI has higher absolute revenue ($24B annualised vs Anthropic’s $10–12B) but Anthropic has better capital efficiency.

Q: What does this mean for smaller AI companies?

A: Both listings, if they succeed, validate the foundation-model business model at trillion-dollar scale. That is good news for Mistral, Cohere, and the Chinese labs (Moonshot, Qwen), all of whom are now in a position to articulate clearer IPO paths. The bad news: the public markets will now be the comparison point for everyone, and the bar will be very high. Labs that cannot match Anthropic’s $47B run rate or OpenAI’s 800M+ WAU will have a hard time getting the same multiples.

Q: How does this connect to the G7 summit starting Monday?

A: Both Dario Amodei (Anthropic) and Sam Altman (OpenAI) are flying to Évian-les-Bains for the G7 summit, where they will sit down with heads of state to discuss AI governance. The timing of the IPO filings — both in the last two weeks — is not a coincidence. The regulatory environment both companies will operate under as public companies is being written in the rooms they are about to enter. The order, the substance, and the photo opportunities in France will all be read by public-market investors when the roadshow begins in the autumn.

🔍 THE BOTTOM LINE (Synthesis)

Two AI labs have filed to go public within a week of each other. Both are unprofitable, both have crossed the trillion-dollar mark in late private rounds, and both are betting that the public markets will accept a future in which foundation models eat most of enterprise software. The filings are confidential, but the implications are not. Whoever lists first will set the narrative. Whoever lists second will have to live with that narrative. The G7 summit starting on Monday will write the rules both companies will operate under once they are public. The next six months will determine whether the AI public-market boom is the start of a category or the top of a bubble. Both companies are betting it is the start. They have to be.

📰 Sources

• Fortune — Anthropic confidentially files for IPO after a $965 billion valuation
• Forge Global — Anthropic IPO: Investment Opportunities & Pre-IPO Valuations
• Andrew.ooo — OpenAI S-1 Filing vs Anthropic IPO: The 2026 AI IPO Race
• Enterprise DNA — Anthropic Files Confidential S-1 for Near-$1 Trillion IPO
• How2Shout — Anthropic Files for IPO After $965B Valuation Surpasses OpenAI
• Decode the Future — Anthropic S-1 Filing: IPO Signal Explained