The short version: OpenAI publicly confirmed on 8 June 2026 that it has filed a confidential S-1 registration statement with the SEC — the first step in a public listing process. This follows Anthropic's own confidential IPO filing the week before. Both companies are now on the public market path, with Goldman Sachs and Morgan Stanley underwriting OpenAI's offering at a targeted valuation of around £800 billion to £1 trillion. A listing window of September to November 2026 is widely cited, though OpenAI has stressed timing remains undecided.
What happened
A confidential filing means the S-1 documents are submitted to the SEC for review but not yet made public. OpenAI will publish its prospectus — including revenue figures, cost structure, risks, and business model — at least 15 days before any roadshow. That is when the full picture of its financial model becomes visible.
The known figures are already substantial. OpenAI reported approximately $25 billion in annualised revenue as of early 2026. Its most recent private valuation was $852 billion. The targeted IPO valuation of around $1 trillion would make it one of the largest tech IPOs in US market history.
Anthropic's numbers are smaller but no less significant. The company closed a financing round at a $965 billion valuation in late 2025 — briefly surpassing OpenAI in private market terms before OpenAI's latest round re-established the gap. Anthropic is also in its own IPO preparation process.
Why this matters for businesses that use AI tools
The IPO filings are a pricing signal as much as a corporate finance event. Here is why:
Both OpenAI and Anthropic have been heavily subsidised by investor capital for years. ChatGPT Pro at £17 per month and Claude Pro at the same price are not priced to generate substantial profit at current AI infrastructure costs. Public markets will change that calculation. Investors buying IPO shares expect a path to profitability — and the only path runs through pricing, product bundling, or both.
Anthropic's billing change on 15 June — separating automated AI workflows into a credit-based model — is an early example of what this structural shift looks like in practice. The flat-rate model served the growth phase. The usage-based model serves the profitability phase. Expect more of this across both platforms.
What might change for business customers
- Subscription pricing: Expect gradual increases for Pro and Team plans over the next 12 to 18 months as both companies demonstrate margin improvement to public market investors.
- Free tier restrictions: The free version of ChatGPT has already been substantially restricted over the past 12 months. This is likely to continue as subscriber conversion becomes a key metric for investor reporting.
- Enterprise packaging: Both companies are likely to push harder on enterprise contracts — annual commitments, volume pricing, and feature gating — to generate predictable recurring revenue that public market investors value highly.
- Feature access: Some capabilities currently included in standard plans may move to higher tiers or usage-based add-ons.
What does not change
The underlying AI capabilities are not going backwards. Competition between OpenAI, Anthropic, Google, and Microsoft keeps the capability floor rising regardless of what happens to pricing. The tools available to small businesses in 2027 will be more capable than those available today, even if they cost more.
The businesses that will benefit most from this environment are those that have built clear systems for AI use — specific workflows, measurable outcomes, defined cost budgets — rather than those using AI tools ad hoc. When pricing changes, systematic users adapt; casual users stop.
What to do now
Audit your AI tool spend: List every AI subscription your business pays for and what you use each one for. Know the number before pricing changes make you look it up.
Identify your most valuable AI use: Which two or three AI-powered tasks in your business produce the clearest return? Protect those from being disrupted by pricing changes.
Consider annual commitments carefully: Locking in current pricing has appeal — but only if you know you will use the tool at that level for 12 months. Don't commit to something you haven't yet built.
Watch the prospectus: When OpenAI's S-1 is made public — likely August or September — its revenue breakdown will tell you which customer segments it is optimising for.