The mega-IPO summer of 2026 — SpaceX, OpenAI, and the trillion-dollar comp set
SpaceX filed its S-1 on May 20 disclosing a $52.5B compute commitment to Anthropic. OpenAI filed its draft prospectus two days later. Anthropic is "near profit" per market reports. Three of the largest tech listings in history are about to clear the disclosure threshold in the same quarter — and the IPO comp set will reprice everything below it.
The pre-disclosure private-market story has been: frontier AI labs are worth what their last funding round priced. The pre-IPO public-market story is going to be: the S-1 has to defend that valuation against actual unit economics, customer concentration, compute commitments, and the regulatory exposure under EU AI Act enforcement. Three of these S-1s are about to land in the same quarter.
SpaceX disclosed $18.7B in 2025 revenue and a $1.25B/month Anthropic compute deal through May 2029 — $52.5B in committed AI compute spend over four years, from a single customer-vendor pair. OpenAI's draft IPO prospectus filed May 22 is the largest tech listing under preparation by any metric. Anthropic's commercial side is reportedly approaching profitability, with public-listing prep less advanced but probable in 2027.
What the comp set actually says
The public-market AI-lab comp set right now is essentially Cerebras (now ~$100B mcap after the IPO), NVIDIA ($75.2B in Q1 data-center revenue alone), and a small set of Chinese-listed competitors. Frontier-lab IPOs will sit at the top of that table. The repricing is going to flow downward: every Series C/D/E AI lab valuation will be calibrated against where OpenAI and SpaceX clear public markets.
The hidden inputs on the OpenAI IPO that the analyst community will read the S-1 for: cost-per-token at current scale, the contracted Microsoft compute relationship through 2030, the actual revenue split between ChatGPT consumer and API enterprise, the Cerebras-relationship structure. Each of these has been hand-waved in private-market narratives. The S-1 disclosure regime forces specifics.
What the SpaceX disclosure already changed
The $1.25B/month Anthropic commitment is the most consequential single line item in any AI-relevant S-1 this year. It tells the market three things: (1) Anthropic has long-dated revenue visibility that's bond-investor-relevant, (2) SpaceX has free-cash-flow surplus large enough to fund AI compute at hyperscaler rates, (3) the contracted-compute model — direct customer-vendor, no infrastructure middleman — is becoming the structural pattern for the largest deals. This is the same shape as the OpenAI-Cerebras $20B / 750MW deal and the Crusoe-OpenAI Abilene operator commitment.
For investors evaluating Anthropic specifically, the SpaceX commitment is a four-year revenue floor that didn't exist in any public document before the S-1 filing. Anthropic's $30B Series G at $380B post-money (covered AM cycle) was priced before this disclosure was on the table. The post-disclosure price might be higher.
The summer ahead
Three things to watch over the next ten weeks. First: OpenAI's S-1 disclosures on customer concentration and unit economics (the analyst community is split on whether the gross margins support the valuation). Second: SpaceX's roadshow execution (this is a 50/50 outcome — the AI-compute thesis is bullish but the launch business is on track for record revenue regardless). Third: the EU AI Act enforcement cadence (the Q1 €250M fines and the August 2 transparency rules are going to land mid-roadshow, which means analyst questions about EU regulatory exposure are going to be sharper than US-listed tech companies are accustomed to).
Fortune — OpenAI trillion-dollar IPO filing → · Singh Ajit — Dev Weekly May 18-24 2026 → · IG — SpaceX OpenAI Anthropic upcoming IPOs 2026 →