Understanding AI Tokens and Their Impact on Financial Markets

The concept of AI tokens has gained prominence alongside the vision shared by industry leaders like OpenAI CEO Sam Altman. He envisioned a future where intelligence would become a utility, comparable to electricity or water. In this scenario, AI tokens might enter mainstream use, similar to how kilowatt-hours measure electricity consumption.

What Are AI Tokens?

Tokens are integral to AI models. They function as units of measurement reflecting the work performed by the AI. Each prompt read or response generated by AI is measured in these tokens. Companies in the technology sector have utilized AI more extensively, discovering that token-based pricing can become costly. This has led businesses such as Uber and Amazon to regulate AI usage and control token-related expenses.

Tracking AI Usage and Economic Impact

AI tokens not only measure usage but provide data for researchers studying economic impacts. Economists Nicola Borri, Aleh Tsyvinski, and Yukun Liu used these tokens to analyze AI’s influence on financial markets. Their study focused on how variations in AI consumption affected stock prices of different companies.

Although they didn’t track individual usage, they assessed the overall rise in AI utilization and its effects on company stocks. They found that as AI usage increased, companies perceived as major AI beneficiaries experienced higher stock returns, termed as “AI premium.” This concept extends beyond purely tech companies, indicating the broader economic implications AI may have.

The New Wave of Data-Driven Research

This analysis uses data from OpenRouter, a platform offering access to multiple AI models via a single interface. As token costs accumulated, businesses turned to such platforms to manage expenses and compare rates across models. The data collected from OpenRouter contributed significantly to the economists’ research, representing around 2% of global AI usage between January 2024 and April 2026.

Stock Market Observations

The study found that companies with stock prices most sensitive to changes in AI consumption enjoyed higher returns — the so-called “AI Premium.” This premium was particularly pronounced in companies within the U.S. and Europe but less apparent in emerging markets such as China.

Among S&P 500 companies identified with high AI premiums were AppLovin, Carvana, Lumentum, Expand Energy, and Baker Hughes. Contrarily, stocks of companies like Moderna, Estée Lauder Companies, ON Semiconductor, Skyworks Solutions, and Aptiv were deemed least favorable in terms of AI opportunities.

Caveats and Implications

The paper’s findings have limitations and have not yet undergone peer review. Additionally, OpenRouter users may not represent the average consumer, leading to skewed data. The study’s biggest contribution may lie in introducing a novel method for economists to track AI’s economic effects using token data.

This research illuminates a path toward better understanding AI’s role in transforming industries economically. As AI continues to integrate into various sectors, data-driven insights will remain crucial in navigating this technological evolution.