European companies with artificial intelligence exposure earned about 1.3 percentage points more in quarterly stock returns than comparable firms without such exposure, research published Thursday by the European Central Bank found.
The advantage extended beyond technology companies supplying AI infrastructure to businesses using the technology in products and operations, the authors said. The central bank published “Artificial intelligence and European firms: the cost of market-based financing” on Nov. 12 in its Economic Bulletin.
The authors also found that AI exposure corresponded to corporate bond spreads about 5 basis points, or 0.05 percentage point, lower in quarterly data covering 2019-26. Bond spreads measure the extra return investors demand to hold company debt against a benchmark, with lower spreads indicating cheaper financing.
The findings depend on investors’ expectations about returns from the technology, the authors said. Disappointing returns could trigger abrupt changes in stock valuations and bond spreads across businesses beyond those driving AI investment, they said. They called for further analysis to establish the possible scale and effect on euro area financing conditions.
Irene Ferrari, Sofia Gori, Johannes Groß, Fabian Schupp and Sigurd Steffensen prepared the research. They examined companies in the STOXX Europe 600 index using quarterly earnings calls from the first quarter of 2017 through the second quarter of 2026. They classified firms as AI-exposed when calls linked the technology to infrastructure, products or improvements in operations and productivity.
About 75% of calls mentioning AI discussed product innovation or growth opportunities, while 50% discussed productivity gains, the central bank found. Some calls covered both. The authors found that technology firms discussed AI product innovation as early as 2017, while other sectors mainly began discussing productivity improvements after ChatGPT launched in late 2022.
The stock-return analysis compared AI-exposed companies with firms in the same sector and quarter, accounting for company characteristics, earnings surprises and sector trends. The authors found that exposed firms also had higher median annual stock returns in most years and tended to command higher prices relative to earnings.
In the bond analysis, the authors accounted for characteristics of individual bonds and issuers. They found that AI-exposed companies issued a larger share of bonds maturing in more than 10 years. The authors said longer maturities would typically require higher spreads, making their combination with lower spreads evidence of more favourable financing conditions.
The authors said AI could improve expected company earnings and reduce investors’ perceptions of credit risk if they believed businesses were making real progress. But adopting AI could also increase financing costs because it often requires large upfront investments in assets with limited value as collateral, they said.
