OpenAI is going public as the most expensive artificial intelligence company in its peer group—not by market cap, but by what investors are paying for each unit of business quality, according to PitchBook’s latest research. The company’s upcoming IPO disclosures will test whether that premium is justified or just momentum.

OpenAI’s path to profitability rests on a contract with an expiration date. In April, it renegotiated its revenue-sharing agreement with Microsoft MSFT, capping payments at $38 billion through 2030 and saving an estimated $70 billion-$97 billion. Without that cap, positive free cash flow is not possible. Investors must price the listing based on an agreement whose most consequential provisions haven’t been written.

The company’s revenue is real, but the economics are broken. First-quarter revenue hit $5.7 billion, but at a negative 122% adjusted operating margin, OpenAI spent $2.22 for every dollar earned. To justify an $852 billion valuation, the company would need to generate $95 billion-$105 billion in free cash flow by 2030. Based on first-quarter numbers, it is on track to lose $10 billion-$30 billion that year instead.

PitchBook’s AI business quality scorecard puts OpenAI last among its peers, at 4.8 out of 10.0. At an $852 billion valuation, that works out to $177.5 billion per AIBQ point—11.8 times what investors pay for Databricks. The firm’s main competitor, Anthropic, is pursuing a parallel listing with higher run-rate ARR (estimated at $47 billion vs. $25 billion-$33 billion for OpenAI), a faster path to profitability, and the enterprise market share lead (40% vs. 27%). The full report breaks down the scorecard in detail, including OpenAI’s capital efficiency, compute obligations, and governance risks.