The OpenAI IPO is finally real. On May 22, 2026, OpenAI confidentially filed its S-1 draft registration statement with the SEC — the same week SpaceX’s own prospectus went public — kicking off what could be the largest tech listing in history. A $1 trillion valuation target. $25 billion in annualized revenue. And a loss of $1.22 for every dollar earned in Q1 2026. The filing was confirmed simultaneously by Fortune, CNBC, Reuters, Axios, and Bloomberg within hours. Goldman Sachs and Morgan Stanley are underwriting the deal, with a public debut targeted for Q4 2026 — possibly as early as September. If you’ve been following the AI race, this is the moment it moves from private money to public markets. Here’s what the numbers actually mean, and what Wall Street is really betting on.
What the OpenAI IPO Filing Actually Says
The filing is still confidential, which means the full S-1 prospectus stays private until approximately 15 days before the public roadshow. What we know comes from CFO disclosures, reporting by The Information, Bloomberg, and CNBC, and analyst estimates from firms including Sacra and FutureSearch.
The revenue story is genuinely extraordinary. OpenAI went from $2 billion in annualized revenue in 2023 to $13 billion in 2025 to $25 billion by March 2026. Sam Altman has publicly targeted $100 billion by 2027. That would represent one of the fastest revenue ramp-ups in corporate history. The company counts 50 million consumer subscribers, 9 million business users, and processes 15 billion API tokens per minute. Enterprise contracts now drive more than 40% of revenue.
• S-1 filed: May 22, 2026 (confidential)
• Private valuation: $852 billion (March 2026 funding round)
• IPO target valuation: $852B–$1 trillion+
• Monthly revenue: ~$2 billion ($25B annualized)
• Q1 2026 operating margin: -122%
• Projected 2026 loss: $14B (non-GAAP) / $25–26B (GAAP est.)
• Underwriters: Goldman Sachs, Morgan Stanley, JPMorgan
• Target listing: Q4 2026, possibly September
Monthly Run Rate
Operating Loss
Infrastructure Spend
Through 2030
The Loss Number Nobody Is Leading With
Every headline focuses on the $1 trillion valuation. The number that should get equal billing is this: OpenAI lost $1.22 for every $1 of revenue in Q1 2026. That’s a negative 122% operating margin. For context, in full-year 2025 the company generated $13.1 billion in revenue and burned through approximately $22 billion to do it.
The company doesn’t project profitability until 2029 or 2030. Internal documents suggest a $14 billion operating loss for 2026 on a non-GAAP basis — but FutureSearch, which has modeled OpenAI’s financials since 2024, estimates the actual GAAP loss lands closer to $25–26 billion, roughly 80% higher than the headline figure most outlets are citing.
The Compute Problem Is Structural
The losses aren’t accidental — they’re a deliberate bet on compute dominance. OpenAI plans to spend $50 billion on computing infrastructure in 2026 alone. Co-founder Greg Brockman confirmed the figure publicly. To put that in context, the entire U.S. semiconductor industry spent $56 billion on R&D in all of 2023. The company has signed infrastructure deals that read like sovereign investment treaties: a $38 billion, seven-year agreement with AWS, and a reported $300 billion total commitment with Oracle through 2031.
• Amazon Web Services: $38 billion over 7 years (2025–2031)
• Oracle: $60 billion per year, 5 years (2027–2031) = $300 billion total
• Total capital needed through 2030: ~$207 billion
→ Public markets are effectively the only pool deep enough to cover this
Microsoft Dependency Is a Double-Edged Sword
Microsoft’s multibillion-dollar investment made OpenAI’s growth possible. It also created a dependency that the SEC will scrutinize closely. OpenAI still runs substantially on Microsoft Azure infrastructure. If that relationship faces commercial or regulatory pressure, OpenAI’s cost structure changes sharply — and so does the investment thesis. The renegotiated revenue-share deal reduced Microsoft’s long-term cut, but the infrastructure reliance remains a line item every institutional investor will flag.
The Competition Is Closing the Gap Fast
Sacra’s April 2026 analysis shows OpenAI’s developer market share declined from roughly 60% to 51% year-on-year, with Claude taking meaningful share in AI coding. ChatGPT’s consumer web traffic share dropped from 76.5% in February 2025 to 54.7% by June 2026, while Google Gemini surged to 27.4% — up 104% in six months. Claude’s web visits grew 306% in a single quarter. The moat is narrowing, and the IPO roadshow will need to explain why it won’t narrow further.
• ChatGPT web share: 76.5% → 54.7%
• Developer share: ~60% → 51%
• Free Gemini/Llama alternatives
• Claude Code taking coding share
• No profitability until 2029–30
• $25B annualized revenue
• 50M consumer + 9M business users
• 15B API tokens/min processed
• Enterprise now 40%+ of revenue
• Most recognized AI brand globally
OpenAI IPO vs. The Biggest Tech Listings Ever
To understand just how unprecedented this IPO is, here’s how OpenAI’s targeted valuation stacks up against the largest public market debuts in tech history.
What Happens Between Now and the Listing
The confidential filing process has a fairly predictable cadence. SEC review of a confidential S-1 typically takes 60 to 90 days with multiple rounds of comments. The full public S-1 becomes available approximately 15 days before the roadshow. That puts the earliest realistic window around late August for the public filing and mid-September for the actual listing.
The CFO Is Pumping the Brakes
CFO Sarah Friar has reportedly told colleagues internally that the company should delay until 2027 to ensure public company reporting readiness. Sam Altman’s own statement at the filing announcement was unusually cautious: “We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company.” That’s not the language of a company sprinting to market. The September window is possible — but don’t bet on it.
The Legal Overhang Is Finally Cleared
One of the last major institutional obstacles was removed in May 2026 when a California jury dismissed Elon Musk’s lawsuit against OpenAI and its leadership, which had alleged the company moved away from its founding nonprofit mission. The dismissal eliminated one of the clearest risk factors that would have appeared prominently in any S-1 filing, giving the company a cleaner path to public markets from a legal standpoint.
🔗 More on the AI IPO wave
▶ Anthropic Just Filed for IPO. Now What? ▶ SpaceX Just Filed for Its IPO. Here’s What’s Inside. ▶ Nvidia Q1 2027 Earnings: $81.6B Revenue and What $91B Guidance Really Means✅ OpenAI IPO — What You Need to Know
Filed May 22, 2026 — Confidential S-1 submitted to SEC. Public prospectus expected ~15 days before roadshow. Q4 2026 listing targeted, possibly September.
$25B revenue, $14B loss — Growing at extraordinary speed, burning capital at extraordinary rate. Loses $1.22 for every $1 earned. No profit expected until 2029–30.
$207B needed through 2030 — Compute commitments to AWS and Oracle require capital that only public markets can provide at this scale. This IPO is a funding necessity.
Competition is real — ChatGPT web share dropped from 76.5% to 54.7% in 16 months. Claude and Gemini are closing the gap faster than most IPO bulls are pricing in.
This sets the benchmark — With Anthropic and SpaceX also targeting Q4 2026 listings, OpenAI’s debut pricing will define whether AI stocks get a premium or a discount from public markets.