Paper: arXiv 2606.30583

Authors: Nicola Borri, Yukun Liu, Aleh Tsyvinski

Abstract

We study 380 trillion tokens of realized AI consumption across more than four hundred LLMs. We build a high-frequency AI factor and show that a long-short strategy based on firms’ AI exposure earns significantly positive returns. The average strategy return is larger based on intensive, frontier-oriented AI consumption but smaller based on casual or open-weight usage. Internationally, the return spread is significant in developed countries but insignificant in emerging markets. Examining occupational AI exposure, we find more positive exposure in occupations intensive in nonroutine interactive tasks and more negative exposure in those intensive in nonroutine analytical tasks.

Complexity vs Empirical Score

  • Math Complexity: 7.0/10
  • Empirical Rigor: 6.0/10 (capped from 9.0: transaction costs not modelled; see penalty flags)
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: This paper demonstrates high empirical rigor through its use of a massive, proprietary dataset and robust statistical methodologies. Its novelty lies in constructing an AI factor from real-time consumption data and linking it to stock returns and occupational exposure, offering fresh insights into a rapidly evolving area.

Research Flowchart

  flowchart TD
    A[Research Goal: Cross-Section of Stock Returns & AI Exposure] --> B{Key Methodology: AI Factor Construction & Long-Short Strategy};
    B --> C[Data/Inputs: 380T Tokens Realized AI Consumption, >400 LLMs, Firm Financials];
    C --> D(Computational Processes: Factor Building, Portfolio Sorting, Regression Analysis);
    D --> E{Key Findings: Significant Positive Returns from AI Factor};
    E --> F[Outcomes: Stronger Returns from Frontier AI Consumption; Returns Significant in Developed Markets; Occupational AI Exposure Varies];

Cited References (115)

  1. Generative AI and firm values (2023)
  2. Returns to Intelligence: Worker and Firm Adaptation to Advances in AI (2026) — link
  3. The Uneven Impact of Generative AI on Entrepreneurial Performance (2024)
  4. Understanding Firms’ AI Efforts and Their Economic Impact (2026) — link
  5. Artificial Intelligence, Firm Growth, and Product Innovation (2024) — link
  6. Artificial Intelligence and Firms’ Systematic Risk (2024) — link
  7. Do Markets Believe in Transformative AI ? (2025)
  8. Introducing Gemini 2.0: Our New AI Model for the Agentic Era (2024)
and 107 more
  1. OpenRouter Now Processes More Than a Quadrillion Tokens a Year (2026)
  2. Goldman Sachs Says the AI Boom Is Bigger Than Investors Think (2026)
  3. Anthropic Economic Index Report: Uneven Geographic and Enterprise AI Adoption (2025) — link
  4. WildChat: 1M ChatGPT Interaction Logs in the Wild (2024) — link
  5. How People Use ChatGPT (2025) — link
  6. State of AI : An Empirical 100 Trillion Token Study with OpenRouter (2025)
  7. Who Captures AI Deflation? Quality-Adjusted Prices, Markups, and Pass-Through in AI Services (2026) — link
  8. Estimating Consumer Preferences for LLMs : Evidence from LMArena (2025) — link
  9. The Welfare Impact of Large Language Models (2026)
  10. Market Beliefs about Open vs. Closed AI (2026) — link
  11. Menu Pricing of Large Language Models (2026)
  12. The Emerging Market for Intelligence: Pricing, Supply, and Demand for LLMs (2025) — link
  13. Risk, Return, and Equilibrium: Empirical Tests (1973)
  14. The Cross-Section of Expected Stock Returns (1992)
  15. Common Risk Factors in the Returns on Stocks and Bonds (1993)
  16. Multifactor Explanations of Asset Pricing Anomalies (1996)
  17. A Five-Factor Asset Pricing Model (2015)
  18. Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency (1993)
  19. On Persistence in Mutual Fund Performance (1997)
  20. The Other Side of Value: The Gross Profitability Premium (2013)
  21. Investment and the Cross-Section of Equity Returns (2019)
  22. A Skeptical Appraisal of Asset Pricing Tests (2010)
  23. … and the Cross-Section of Expected Returns (2016)
  24. Does Academic Research Destroy Stock Return Predictability? (2016)
  25. The Characteristics That Provide Independent Information about Average U.S. Monthly Stock Returns (2017)
  26. Characteristics Are Covariances: A Unified Model of Risk and Return (2019) — link
  27. Taming the Factor Zoo: A Test of New Factors (2020) — link
  28. Shrinking the Cross-Section (2020)
  29. Empirical Asset Pricing via Machine Learning (2020)
  30. Asset Pricing with Omitted Factors (2021)
  31. Open Source Cross-Sectional Asset Pricing (2022)
  32. Principal Portfolios (2023)
  33. Technological Revolutions and Stock Prices (2009)
  34. Investment Shocks and Asset Prices (2011) — link
  35. Growth Opportunities, Technology Shocks, and Asset Prices (2014) — link
  36. Displacement Risk and Asset Returns (2012) — link
  37. Automation and the Displacement of Labor by Capital: Asset Pricing Theory and Empirical Evidence (2023) — link
  38. In Search of Attention (2011)
  39. Investor Sentiment and the Cross-Section of Stock Returns (2006)
  40. Giving Content to Investor Sentiment: The Role of Media in the Stock Market (2007)
  41. Is there a replication crisis in finance? (2023)
  42. Robots and Jobs: Evidence from US Labor Markets (2020)
  43. Tasks, Automation, and the Rise in U.S. Wage Inequality (2022)
  44. The Productivity J -Curve: How Intangibles Complement General Purpose Technologies (2021)
  45. Occupational, Industry, and Geographic Exposure to Artificial Intelligence: A Novel Dataset and Its Potential Uses (2021)
  46. The Impact of Artificial Intelligence on the Labor Market (2020)
  47. GPTs are GPTs : An Early Look at the Labor Market Impact Potential of Large Language Models (2023)
  48. Hybrid Corn: An Exploration in the Economics of Technological Change (1957)
  49. Technical Change and the Rate of Imitation (1961)
  50. Adoption of New Technology (2003)
  51. An Exploration of Technology Diffusion (2010)
  52. We’d Better Watch Out (1987)
  53. The Lever of Riches: Technological Creativity and Economic Progress (1990)
  54. The Use of Knowledge in Society (1945)
  55. Efficient Capital Markets: A Review of Theory and Empirical Work (1970) — link
  56. The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox (1990)
  57. General Purpose Technologies: `Engines of Growth’? (1995) — link
  58. Endogenous Technological Change (1990) — link
  59. A Model of Growth Through Creative Destruction (1992) — link
  60. Quality Ladders in the Theory of Growth (1991) — link
  61. The Information-Technology Revolution and the Stock Market (1999) — link
  62. The Information-Technology Revolution and the Stock Market: Evidence (2001) — link
  63. Optimal Investment, Growth Options, and Security Returns (1999) — link
  64. Equilibrium Cross Section of Returns (2003) — link
  65. Corporate Investment and Asset Price Dynamics: Implications for the Cross-Section of Returns (2004) — link
  66. Stock Valuation and Learning about Profitability (2003) — link
  67. Was There a Nasdaq Bubble in the Late 1990s? (2006) — link
  68. Technological Innovations and Aggregate Risk Premiums (2009) — link
  69. Innovation, Growth, and Asset Prices (2015) — link
  70. Technological Innovation, Resource Allocation, and Growth (2017) — link
  71. An Essay in Dynamic Theory (1939)
  72. Directed Technical Change (2002)
  73. The Skill Content of Recent Technological Change: An Empirical Exploration (2003)
  74. Skills, Tasks and Technologies: Implications for Employment and Earnings (2011)
  75. The Growth of Low-Skill Service Jobs and the Polarization of the US Labor Market (2013)
  76. Why Are There Still So Many Jobs? The History and Future of Workplace Automation (2015)
  77. The Growing Importance of Social Skills in the Labor Market (2017)
  78. Automation and New Tasks: How Technology Displaces and Reinstates Labor (2019)
  79. Artificial Intelligence, Automation, and Work (2019)
  80. New Frontiers: The Origins and Content of New Work, 1940–2018 (2024)
  81. Steam as a General Purpose Technology: A Growth Accounting Perspective (2004)
  82. Diffusion of General Purpose Technologies (1998)
  83. Was Prometheus Unbound by Chance? R isk, Diversification, and Growth (1997) — link
  84. Distance to Frontier, Selection, and Economic Growth (2006) — link
  85. Labor-Technology Substitution: Implications for Asset Pricing (2019) — link
  86. Labor-Force Heterogeneity and Asset Prices: The Importance of Skilled Labor (2017) — link
  87. Labor Mobility: Implications for Asset Pricing (2014) — link
  88. Occupation-Level Income Shocks and Asset Returns: Their Covariance and Implications for Portfolio Choice (2013) — link
  89. Industry-Specific Human Capital, Idiosyncratic Risk, and the Cross-Section of Expected Stock Returns (2013) — link
  90. Agentic Design Patterns Part 1: Four AI Agent Strategies That Improve GPT-4 and GPT-3.5 Performance (2024)
  91. Agentic Design Patterns Part 3: Tool Use (2024)
  92. AI and Our Economic Future (2026)
  93. Past Automation and Future A.I. : How Weak Links Tame the Growth Explosion (2026)
  94. The Adoption and Usage of AI Agents: Early Evidence from Perplexity (2025)
  95. Anthropic Economic Index Report: Learning Curves (2026)
  96. Google I/O 2026: Sundar Pichai ’s opening keynote (2026)
  97. Technology and labor displacement: Evidence from linking patents with worker-level data (2023)
  98. Technology and Labor Markets: Past, Present, and Future (2025) — link
  99. Beyond Exposure: Predicting AI Adoption Based on Comparative Advantage (2026) — link
  100. The Macroeconomic Effect of AI : Sizing the Software Engineering Channel (2026)
  101. AI Agents and Higher-Order Work (2026) — link
  102. Left behind: Creative destruction, inequality, and the stock market (2020)
  103. Artificial intelligence, productivity, and the workforce: Evidence from corporate executives (2026)
  104. Mind the Gap: AI Adoption in Europe and the US (2026)
  105. Durability of Output and Expected Stock Returns (2009)
  106. Risks for the Long Run: A Potential Resolution of Asset Pricing Puzzles (2004)
  107. Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models (2026) — link

0 of 115 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.