Anthropic confidentially filed its draft S-1 with the SEC on 1 June 2026. OpenAI followed exactly one week later, on 7 June. The headlines treated both as “AI IPOs incoming,” and the market reaction was predictable: combined private valuations approaching $2 trillion make this the biggest capital event in tech history, and the first time public-market investors can buy pure-play AI companies directly.
Both filings are confidential under the JOBS Act. The Rule 135 announcements that hit the wires in early June tell investors almost nothing: no financials, no risk factors, no business description. Neither company will trade for months. What matters now is why the two largest frontier AI labs both ran out of alternatives at the same moment, and what that moment means for everyone watching.
What is a confidential S-1 filing and how does the SEC review process work?
A confidential S-1 is a draft registration statement submitted to the SEC under the JOBS Act. It lets Emerging Growth Companies, defined as those with revenue under $1.235 billion, begin regulatory review without publicly disclosing their financials. Both Anthropic and OpenAI are expected to qualify, which means only regulators and advisers see the sensitive information during the initial review phase.
The process moves through several stages. The company submits its draft, the SEC reviews it and issues comment letters (typically within 30 days for the initial round), and the company responds with amendments. This back-and-forth continues for multiple rounds, usually three to six months for companies of this scale. Only after the SEC is satisfied does the S-1 “flip” to public on EDGAR, followed by a 15-day cooling-off period, the roadshow, book-building, pricing, and finally first-day trading.
SpaceX’s timeline offers a useful benchmark. The company’s confidential-to-public S-1 took roughly 11 weeks, which suggests a possible public-filing window in late August or September for Anthropic and OpenAI if their reviews move at a similar pace. But analysts point to specific complications that are likely to slow things down. The biggest is Anthropic’s gross-vs-net revenue accounting. In practice, this means Anthropic books the full amount its enterprise customers spend on cloud compute as its own revenue, even when that money passes straight through to AWS or Google Cloud. The SEC may decide only Anthropic’s margin on those deals counts as revenue, which could reduce the reported run-rate by 20 to 40 percent (more on this below). Add in the Department of War litigation requiring material risk-factor disclosure, AI safety risk factors as novel disclosure territory with no settled template, and customer concentration treatment, and you are looking at four to six months rather than eleven weeks.
The Rule 135 announcement tells the market only that a filing exists. You cannot buy shares, review financials, or assess risk factors yet. Both companies’ announcements explicitly state they “do not constitute an offer to sell or the solicitation of an offer to buy any securities.” Once the S-1s become public, investors will need a framework to evaluate what they are actually buying.
Why are Anthropic and OpenAI going public now in 2026?
If the SEC process means neither company lists for months, the natural question is why file now at all. Four structural forces converged in mid-2026 to leave both frontier labs with no good alternative to the public markets.
The first is capital ceilings. Anthropic’s $65 billion Series H at a $965 billion post-money valuation and OpenAI’s $40 billion raise pushed private valuations to levels where venture and growth-equity limited partners are structurally maxed out. Mutual funds like Capital Group, Fidelity, and T. Rowe Price participated in Anthropic’s Series H not as long-term private investors but as a pre-IPO price anchor. Mutual funds do not pay $965 billion for long-term private positions. They pay to be first in line on the public listing. AI captured approximately 80% of global venture capital in Q1 2026, saturating LP allocations and making further private rounds impossible at the scale both companies now require.
The second is competitive timing. Being first to price establishes the benchmark valuation multiple for every AI IPO that follows. The lab that completes SEC review first can define the narrative for AI companies as a public-market asset class. Analysts describe it as “a race, not a parade, toward dual listings.”
The third is the market window. Equity markets in mid-2026 are receptive to high-growth tech issuance, but that window may not stay open. Beyond its value as a timeline benchmark, SpaceX also creates a competitive deadline. Neither Anthropic nor OpenAI wants to be the last mega-IPO in the wave when institutional demand may be exhausted.
The fourth is employee liquidity. Both companies have large workforces compensated substantially in equity. Without a public listing or tender offer, that equity is illiquid. IPOs unlock employee wealth and serve as a retention mechanism, particularly important when competing for AI talent. Post-IPO, employees face a 180-day lock-up before they can freely sell.
As the AI Funding Tracker notes, a clean first-day pop for SpaceX opens the window for every name that follows. A stumble resets 2026 expectations entirely.
How do Anthropic and OpenAI’s business models and valuations compare ahead of their IPOs?
The same structural pressures produced the same timing response for both companies. Beyond that, they are making different bets about where AI value accrues, and those bets will appeal to different types of investors.
Anthropic is a Public Benefit Corporation. Its board is legally required to balance shareholder returns with a stated public benefit: developing AI systems that are safe, interpretable, and aligned with human values. This affects fiduciary duties in ways no large-scale tech IPO has tested before. The board can reject acquisition offers that compromise the stated benefit even at a premium price, and shareholders have limited legal standing to challenge decisions made in pursuit of that benefit.
OpenAI operates under a capped-profit structure with the OpenAI Foundation, a nonprofit, retaining ultimate control. Microsoft owns approximately 27%, and its revenue-share agreement reportedly grants Microsoft 20% of revenue through 2030. Investors face return limits, with profits above a negotiated cap flowing back to the foundation. The S-1 will need to explain this structure in detail to public-market investors who are not used to capped upside.
The revenue profiles are equally divergent. Anthropic’s run-rate crossed $47 billion as of mid-May 2026, driven heavily by Claude Code. That terminal tool alone grew from $500 million to $8 billion ARR between September 2025 and May 2026. Anthropic holds 42 to 54% of enterprise coding market share versus OpenAI’s 21%, and coding is 51% of all generative AI enterprise usage. The revenue is concentrated, sticky, and enterprise-heavy. But the gross-vs-net accounting question means headline ARR could be reduced by 20 to 40% if the SEC forces a restatement to net revenue.
OpenAI reports roughly $2 billion in monthly revenue, about a $24 billion annualised run-rate, with 900 million weekly active ChatGPT users and 50 million paying subscribers. Enterprise revenue makes up more than 40% of that total and is on track to reach parity with consumer revenue by end of 2026. The diversification is broader, but consumer-revenue volatility is a reality. OpenAI lost nearly $9 billion in 2025 and does not expect break-even until 2029 or 2030.
Valuation-wise, Anthropic’s Series H implies roughly 21 times run-rate ARR, subject to the accounting question. OpenAI’s most recent round valued it at $852 billion. Both are well above traditional tech IPO multiples (the Magnificent Seven trade at 3 to 13 times forward sales). The comparison is not perfectly aligned (Anthropic’s multiple uses current run-rate rather than forward estimates), but the gap is wide enough that the direction is clear. The valuations have drawn dot-com-era comparisons from academics, though the revenue growth at both companies is real in a way that late-nineties internet startups rarely delivered.
What matters is that the two IPOs represent two incompatible theories of AI value. Anthropic is betting that enterprise agentic AI, sold through APIs and terminal tools into coding workflows, is where durable moats form. OpenAI is betting that platform-scale distribution, with 900 million users and a brand that has become synonymous with AI, creates competitive advantages that compound over time. The public market will have to choose which theory it believes.
The confidential filing process means every investor is operating on incomplete information. The real analysis begins when the S-1s flip public on EDGAR, likely between late August and December 2026. The one-week gap in June 2026 marked the moment structural pressure overwhelmed private-market capacity for both companies simultaneously. When the S-1s flip public, the investor community will see two distinct bets: a safety-governed enterprise API company whose revenue maths turns on cloud-reseller pass-through accounting, and a consumer-platform company whose subscriber conversion rate will be tested by quarterly earnings. Whichever company prices first will define the valuation multiples public-market investors use to judge every AI listing that follows. The race is not just about which stock trades higher. It is about which theory of AI value the public market validates.
Frequently Asked Questions
How can retail investors buy shares in Anthropic or OpenAI when they go public?
Retail investors can purchase shares through any standard brokerage account on the first day of trading, the same way they would buy any newly listed stock. Most major brokers (CommBank, Stake, Interactive Brokers) will make shares available at the opening auction. Some platforms also offer conditional access to IPO allocations, though retail allocations for IPOs of this size are typically small and heavily oversubscribed.
What is the realistic timeline for these IPOs to actually complete?
The most likely window for public S-1 filings is late August to September 2026, assuming the SEC review follows a pace similar to SpaceX’s 11-week confidential-to-public timeline. From the public S-1 filing, a further 15-day cooling-off period plus a 1-2 week roadshow means first-day trading would fall between late September and mid-October 2026. However, complex reviews involving novel governance structures could extend this timeline by several months.
Is it true that both Anthropic and OpenAI are unprofitable?
Neither company has disclosed GAAP profitability figures, and their confidential S-1s remain private, so the market does not yet know. What is clear is that frontier AI training runs cost billions of dollars, and neither company operates with the cost discipline of a mature public company. The gross-vs-net revenue accounting question at Anthropic further complicates any profitability estimate. The public S-1 flip will reveal whether either company is approaching break-even or burning cash at venture scale.
How does a Public Benefit Corporation differ from a regular company for shareholders?
A Public Benefit Corporation like Anthropic legally requires its board to balance shareholder returns against a stated public benefit (AI safety), rather than prioritising shareholder value alone. This affects fiduciary duties in two practical ways: the board can reject acquisition offers that compromise the stated benefit even at a premium price, and shareholders have limited legal standing to challenge decisions made in pursuit of the public benefit. No large-scale tech IPO has tested this structure at public-market scale before.
What does “capped-profit” mean for OpenAI’s corporate structure?
OpenAI’s capped-profit model limits the return that investors can earn, with profits above a negotiated cap flowing back to the non-profit entity that governs OpenAI. The specific cap multiples have not been publicly disclosed for the most recent funding rounds. This structure means OpenAI is not a conventional for-profit company: Microsoft’s approximately 27 percent stake is subject to return limits, and the non-profit board retains ultimate control regardless of ownership percentage. The S-1 will need to explain this structure in detail for public-market investors.
What happens to existing investors’ shares when these companies go public?
Existing private investors (venture funds, mutual funds, strategic investors) will typically have their shares converted to common stock at the IPO. Standard IPO lock-up agreements prevent these shareholders from selling for 180 days after the listing, protecting the stock from immediate insider selling pressure. Some large pre-IPO investors like Capital Group and Fidelity may have negotiated different lock-up terms. When the lock-up expires, the market will watch closely for any large block sales.
Why does the gross-vs-net revenue accounting question matter for Anthropic?
If Anthropic books cloud-computing costs that pass through to enterprise customers as gross revenue, its headline ARR of $47 billion could overstate the company’s true economic scale by 20 to 40 percent. The SEC will scrutinise this treatment during review, and a restatement to net revenue would reduce the reported run-rate and raise Anthropic’s effective valuation multiple (currently approximately 21x ARR). This is not an abstract accounting debate; it directly affects how investors value the company against peers.
Could either IPO be delayed or cancelled?
Yes. Both filings remain confidential, and the SEC can issue multiple rounds of comment letters requiring material amendments. If the SEC raises significant issues around governance structure (particularly Anthropic’s PBC model or OpenAI’s capped-profit structure), revenue recognition, or risk disclosure, the process could extend into 2027. External factors also matter: a market downturn, a poorly received SpaceX listing, or a macroeconomic shock could cause either company to delay pricing or withdraw entirely.
Is it a coincidence that both companies filed in the same week?
No. The simultaneous June 2026 filings are the result of structural convergence, not coincidence. Both companies hit private-capital ceilings at roughly the same time, both face employee equity liquidity pressure, and both recognise that the first to price sets the AI valuation benchmark for public markets. The one-week gap between filings likely reflects competitive intelligence rather than accident: each company wanted to establish its own narrative without appearing to react to the other.
How much will Anthropic and OpenAI shares cost on their first day of trading?
The final IPO price per share is set during the book-building process in the final days before listing, and it depends on the total offering size divided by the number of shares offered. Neither company has disclosed these figures. For context, a company valued at $100 billion might price shares in the $30 to $50 range, but companies can also execute reverse stock splits to achieve a desired nominal price. No reliable estimate exists until the public S-1s disclose the proposed offering amount and share count.
