Featured image by Josie Doan/PitchBook News
Harrison Rolfes is a senior research analyst at PitchBook, where he covers private companies.
Valuation doesn’t necessarily reflect business quality when it comes to the world’s biggest AI companies, according to a new analysis by PitchBook that scored OpenAI ($840 billion) lower than Databricks ($134 billion).
For institutional allocators and GPs, the risk is clear: As LPs increasingly scrutinize mark-to-market valuations and secondary liquidity tightens, the spread between a company’s stated valuation and its underlying business quality is likely to emerge as a meaningful source of downside exposure. With the five top AI companies having raised a combined $202 billion in primary capital, how much of that is priced on proof will define the next phase of this market.
The contrast is starkest at the extremes. Databricks, valued at $134 billion, scored the highest at 8.7 on PitchBook’s new AI Business Quality (AIBQ) framework, a model evaluating frontier AI companies on capital efficiency, revenue quality, computing independence, governance optionality and moat durability. The company’s business quality is driven by enterprise-grade revenue cohorts, a multi-cloud architecture that limits infrastructure concentration risk, and the distinction of being the only one to achieve positive free cash flow in 2025
OpenAI presents the inverse. Despite announcing in February a record $110 billion in funding commitments—anchored by Amazon, Nvidia and SoftBank—with an $840 billion post-money valuation, it scored very low at 4.8. OpenAI’s go-to-market skews heavily toward consumer and developer channels, with about 85% of its 800 to 900 million weekly active users on free tiers. Compounding infrastructure concentration and a cash-flow-positive timeline that extends to 2029 further weighed on its score. Microsoft‘s absence from this latest funding round also raises questions about a partnership that has been central to OpenAI’s capital and distribution strategy.
Anthropic scored 7.4, thanks to net revenue retention above 140% and an 80% enterprise mix, though it has a dependence on third-party cloud infrastructure. A late-February Pentagon standoff over AI safeguards validated its governance optionality score, with the resulting ban not impairing its core enterprise business.
Meanwhile, Safe Superintelligence has no product, no revenue and no disclosed go-to-market timeline—its $32 billion valuation is a pure research bet on the founders’ credibility. It scored the lowest at 2.3.
Rounding out the group, xAI—now a SpaceX subsidiary—scores 5.4, reflecting meaningful computing independence that’s offset by revenue that’s still nascent and unique governance constraints.
The conclusion is unambiguous: Investors are effectively betting on mission and future optionality rather than the metrics that determine whether these companies will generate durable returns.
Learn about our editorial standards.