Capital markets have already begun rewarding companies that investors believe will be the main beneficiaries of artificial intelligence adoption — and this process is happening long before these companies actively start using AI in their operations. According to a new large-scale analysis, the stocks of such "potential leaders" grow on average 0.64% per week faster than the stocks of other market participants. Experts have dubbed this persistent yield gap the "AI premium."
The research is based on a unique dataset — 380 trillion tokens collected by the OpenRouter platform from January 2024 to April 2026. This dataset covers over 400 models, including GPT, Claude, and Deepseek, and reflects about 2% of the global monthly AI consumption. This volume of data made it possible to identify a direct correlation between the growing popularity of neural networks and stock market dynamics.
To measure this effect, economists developed a special indicator — the "AI factor" — which weekly records changes in global artificial intelligence consumption. Companies were then divided into two groups: those whose stocks are sensitive to the growing popularity of neural networks, and those whose value barely reacts to such changes.
The result was telling: the difference in yield between these groups was approximately 0.64% per week. At first glance, this may seem like a negligible amount, but in the long term, the compound interest effect turns this premium into a colossal gap. Essentially, the market is front-loading future AI benefits, revaluing assets based on expectations rather than current financial performance.
The key conclusion of the study is that the status of "AI beneficiary" is determined solely by investor expectations. A company may not use the technology at all today, but if the market believes in its potential, its stocks will rise. As one of the co-authors, Yukun Liu, emphasized, "markets now reward the very proximity of companies to the most advanced models in the US and Europe."
Who benefits?
The most unexpected finding was that the AI premium has extended far beyond the technology sector. Not only IT giants but also retailers, consumer goods manufacturers, and even representatives of heavy industry are receiving increased returns. Investors are betting on a massive increase in labor productivity across all areas of business. As another co-author, Aleh Tsyvinski, noted, "the AI story is not just a technological narrative, but a much broader story that affects companies and workers in all parts of the economy."
The second important factor was the geographic distribution of profits. The bulk of the premium is concentrated in the US and Europe — regions closely tied to building infrastructure and constructing modern data centers. In China and emerging markets, this effect is significantly weaker. Market mechanisms encourage proximity to cutting-edge developments, which are currently available to only a few.
Finally, the structure of AI users played a decisive role. The financial premium is formed exclusively by the professional segment, which uses complex, long queries and paid subscriptions. As co-author Nicola Borri emphasized, "the premium is set precisely by professionals working with advanced AI, not by those who occasionally try free or open models." Investors value the depth of technology integration, not mass reach.
Analytical commentary from Cryptalist: This study confirms a long-standing thesis that the stock market is a machine for discounting the future. A premium of 0.64% per week is not just a statistical anomaly, but a signal that institutional investors are already pricing in the impending transformation of the economy into asset values. For the crypto community, this is also an important lesson: the value of digital assets related to AI infrastructure (decentralized computing tokens, protocols for model training) could also receive a similar boost. The key is to distinguish real projects from hype, because the AI premium only works for those whom the market considers true beneficiaries.