人工智能巨头排名：前沿五强的新框架（英）2026

Giants: A NewFramework forthe Frontier Five Where the models endand the business begins Ranking the AI Giants:A New Framework forthe Frontier Five Institutional Research Group Harrison RolfesSenior Research Analyst,Private Company Coverageharrison.rolfes@pitchbook.com pbinstitutionalresearch@pitchbook.com Where the models end and the business begins Published on March 4, 2026 PitchBook is a Morningstar company providing the most comprehensive, mostaccurate, and hard-to-find data for professionals doing business in the private markets. Contents Key takeaways •The valuation-quality paradox is real and intensifying:Our AIBQ framework scoresDatabricks (with a $134 billion valuation)1an 8.7 out of 10 and Anthropic (with a$380 billion valuation)2a 7.4 out of 10, yet Databricks commands a fraction of xAI’s$250 billion valuation (with a score of 5.4)3and OpenAI’s $840 billion valuation(with a score of 4.8),4and Anthropic’s completed $380 billion round now exceedsxAI’s acquisition price despite Anthropic having a materially higher AIBQ score.OpenAI’s $110 billion funding round widened the paradox dramatically: Its valuationincreased 68%, its capital efficiency ratio collapsed from 0.31x to 0.11x, and itsoverall AIBQ score dropped 0.4 points. The market now pays $175 billion per AIBQpoint for OpenAI versus $15.4 billion for Databricks, an 11.4x premium per unitof business quality, up from 6.2x before the round. The market currently pricesnarrative and optionality, whereas AIBQ measures business fundamentals. Forinvestors seeking sustainable returns, this gap creates a significant opportunity. •Revenue quality matters more than revenue quantity:Based on estimated qualitymetrics, one dollar of Anthropic revenue (Anthropic has an NRR over 140%, an 80%enterprise revenue concentration, and multiyear contracts)5is worth roughly $2.20of OpenAI revenue (OpenAI has an estimated NRR around 115%, a roughly 60%consumer revenue concentration, and shorter contract terms).6On our quality-adjusted framework, Anthropic trades at 27.1x current ARR ($380 billion/$14 billion) Appendix102 versus OpenAI’s 42x ($840 billion/$20 billion)—a premium for OpenAI that hasinverted since its $110 billion round, with OpenAI now trading at a 55% multiplepremium to Anthropic despite a materially lower revenue quality score.7ClaudeCode’s acceleration to $2.5 billion in ARR, with enterprise users representing morethan half of its revenue and business subscriptions quadrupling since January 1,2026, further validates Anthropic’s revenue quality advantage. Recent enterprisecustomer disclosures across the cohort further support the divergence in revenue-quality metrics. •Databricks is systematically misclassified:Consensus places Databricks alongsidefrontier AI labs; however, Databricks is primarily a data infrastructure company withAI capabilities. This distinction makes it the most traditionally investable companyin the Frontier Five, having already achieved positive FCF with gross marginsover 80%, an NRR over 140%, and IPO-ready governance.8Databricks is also theonly Frontier Five company completely insulated from the Anthropic-Pentagonstandoff, with no government contract dependency, no political risk exposure, andno involvement in the broader debate over military AI use restrictions. Its capitalefficiency ratio of 0.16x now leads the cohort following OpenAI’s ratio compression,and its AIBQ lead over the second-ranked company has widened to 1.3 points (8.7versus Anthropic’s 7.4), up from 1.1 points before the crisis. How to read this report •Part I introduces the AIBQ framework—the analytical lens applied throughoutthe report.•Part II provides market context: sizing, capital formation, andcompetitive dynamics.•Part III profiles each company in order of AIBQ score (highest to lowest), applyingconsistent analysis to enable direct comparison.•Part IV synthesizes investment implications with actionable triggers. Note on methodology •PitchBook is the primary source for the data in this report. •All financial data was accessed between January 15 and February 27, 2026, unlessotherwise noted. All closed-round data is as of February 27, 2026. Secondarymarket pricing is as of February 19 or 20, 2026. •Material developments that occurred on February 27, 2026, including OpenAI’s$110 billion funding round, the presidential executive order banning federal use ofAnthropic’s technology, and xAI’s classified-systems agreement with the Pentagon,have been incorporated into the analysis and are noted throughout. •Post-money valuations as of February 2026: ‐OpenAI’s $840 billion valuation reflects its $110 billion funding round closedon February 27, 2026, with Amazon ($50 billion), NVIDIA ($30 billion), andSoftBank ($30 billion) as lead investors. The round remains open for additionalparticipants. OpenAI’s prior valuation of $500 billion, based on its October 2025secondary transaction, is used as the baseline for comparative purposes wherenoted. Addi