OpenAI confidentially filed IPO paperwork with the SEC in June 2026, targeting a public listing at roughly $1 trillion. Internal projecti...
The Financial Times published its analysis on July 5 under the headline "Why OpenAI and Anthropic may struggle to float." The core argument isn't that these companies are failing it's that the economics of frontier AI are genuinely difficult to reconcile with what public market investors expect from companies at these valuations. That tension is real, and it's getting sharper as enterprise customers start questioning whether paying per token at frontier prices is actually worth it.
The Token Payment Problem Nobody Wants to Talk About
Palantir CEO Alex Karp said something unusually direct last week on CNBC: "Something has gone completely wrong" with the token-payment business model. He was talking about the broader AI industry, but the comment landed specifically on OpenAI and Anthropic, whose revenue is almost entirely built on pay-per-use token consumption. The aggressive adoption of agentic AI this year with agents running long, multi-step tasks has led to what some enterprise teams are calling "tokenmaxxing": workflows that consume tokens at a rate that makes the economics of frontier AI feel unsustainable at scale.
Enterprise customers are starting to respond. Some US and international organizations have already reported switching to cheaper Chinese models like DeepSeek and cutting back on payments to OpenAI and Anthropic. Beijing startup Z.ai's GLM-5.2 is now ranked among the top 10 large language models by Artificial Analysis and sits second for web development by Code Arena placing it alongside Anthropic, OpenAI, and Google at four to six times lower cost. That competitive pressure from open-weight Chinese models is exactly what the FT is pointing to as a structural threat to the frontier AI business model at IPO valuations.
Anthropic's situation is meaningfully different from OpenAI's here. Its $47 billion run-rate revenue as of mid-May 2026 disclosed with its Series H announcement represents rapid enterprise expansion. Claude Code and agentic tooling have driven adoption in ways that are reducing Anthropic's operating losses faster than analysts expected. The $965 billion valuation is aggressive at 21 times current ARR, but the growth trajectory is real. OpenAI's $20 billion ARR against a $14 billion projected loss this year is a harder story to tell to public market investors who care about the path to profitability.
What the SpaceX Debut Actually Taught Investors
SpaceX went public on June 12, 2026 the same day the Anthropic export control ban landed raising over $85 billion. At its peak, it was valued at $2.66 trillion, or 142 times trailing sales. It was already profitable going into the listing; its Starlink division's profits offset losses in the space division. Then it acquired xAI in an all-stock deal before IPO, recasting its 2025 financials and becoming deeply unprofitable in the process. The stock has since fallen from above $225 to around $153.
The SpaceX lesson for OpenAI and Anthropic investors is uncomfortable. Day-one euphoria at extreme revenue multiples creates its own momentum, which then fades when the market focuses on the actual numbers. OpenAI and Anthropic plan to float only 5% to 10% of shares, creating limited supply that will drive initial price volatility the same dynamic that made SpaceX's debut look like a success before the pullback. Motley Fool's analysis notes that if either company mimics SpaceX's retail-investor-heavy allocation strategy, it should be read as a signal that institutional demand at the offered price isn't sufficient.
OpenAI is reportedly reconsidering its timeline entirely after the SpaceX experience. Reports in late June indicated Altman was pushing for a $1 trillion valuation with a raise near $60 billion, but advisers are now weighing a delay until 2027. Anthropic is still tracking toward a late-2026 debut potentially as early as October which would make it the first of the two major AI labs to test public market appetite for frontier AI at megacap valuations.
The Regulatory Overhang Is Now a Financial Risk
June 2026 added a new variable that wasn't priced into either company's private valuations: regulatory risk that directly impacts revenue. Fable 5 and Mythos 5 were offline for 18 days under export controls. During that period, Anthropic struggled to meet demand and instituted usage limits during peak hours a demand-rationing situation that directly constrains the revenue growth story prospective IPO investors are being asked to believe in. The access has been restored, but the precedent is set.
OpenAI's GPT-5.6 rollout was delayed by government request in the same week. A coalition of 42 state attorneys general has opened a probe into OpenAI. Copyright lawsuits involving AI training data, data privacy investigations, and antitrust concerns tied to large technology partnerships are all live risks that will appear prominently in any S-1 filing. Public market investors reading those risk sections will see a regulatory environment that is actively intervening in the business, not observing from the sidelines.
The government stake proposal Altman floated last week giving Washington 5% of OpenAI and ideally every other major AI lab is partly a response to this regulatory overhang. A government that owns equity has a financial incentive for the company's success. Whether that trade-off is worth ceding $42.6 billion in equity is the calculation OpenAI is making, and it's a calculation that will be visible to every public market investor reading the prospectus.
What Enterprise Teams Should Take From This
For enterprise teams making procurement decisions, the IPO dynamics matter more than they might seem. A company preparing for a public listing is under pressure to demonstrate revenue stability and reduce costs. That pressure tends to manifest in pricing changes, product prioritization shifts, and enterprise contract terms that favor the company's metrics story over customer flexibility.
The token pricing environment is already being questioned. Anthropic's Mythos-class tier launched at $50 per million output tokens double the $25 Opus anchor that held for four consecutive generations. If that price holds post-IPO, organizations building token-intensive agentic workflows face ongoing cost pressure. If it drops, Anthropic's revenue story weakens. Neither outcome is straightforward for enterprise planning.
The competitive threat from cheaper open-weight Chinese models is also enterprise-relevant independently of the IPO story. If GLM-5.2 and DeepSeek continue closing the capability gap at four to six times lower cost, the pressure on frontier AI pricing will come from below as well as from public market scrutiny above. Organizations that have already been evaluating model redundancy and cost optimization are better positioned for whichever direction this resolves.
Frequently Asked Questions
Q: When are OpenAI and Anthropic going public?
Anthropic filed confidentially June 1, 2026 and is targeting a late-2026 debut, potentially October. OpenAI filed in June but is reportedly weighing a delay to 2027 after SpaceX's post-IPO stock decline. Neither company has confirmed a firm date or price.
Q: What are OpenAI's projected losses going into its IPO?
Internal projections show losses of around $14 billion in 2026, with profitability not expected until 2029-2030. At its $852 billion valuation and $20 billion ARR, OpenAI would debut at roughly 50 times trailing revenue.
Q: Why did SpaceX's IPO matter for OpenAI and Anthropic?
SpaceX peaked above $225 after its June 12 debut and has fallen to around $153 showing that extreme valuation multiples face correction once institutional investors focus on fundamentals. Both AI labs plan the same 5-10% limited float that amplified SpaceX's day-one volatility.
Q: What is tokenmaxxing and why does it matter?
Tokenmaxxing describes agentic AI workflows that consume tokens at rates making frontier AI uneconomical at scale. Enterprise customers are switching to cheaper Chinese models like DeepSeek and Z.ai's GLM-5.2 which offer comparable capability at four to six times lower cost directly threatening frontier AI revenue growth.
References
- The Daily Upside. OpenAI, Anthropic Speed Toward IPOs Amid Growing Scrutiny of Token Payments. July 5, 2026. thedailyupside.com
- The Motley Fool. 4 Lessons for Future OpenAI and Anthropic Investors, Following SpaceX's Historic IPO. July 4, 2026. fool.com
