European companies are gearing up to deliver their strongest quarterly earnings in more than three years. According to LSEG I/B/E/S data compiled by analyst Tajinder Dhillon and published on July 9, constituents of the STOXX 600 index are expected to post a 15.3% year-on-year profit growth for the second quarter, totaling an estimated €156.8 billion. It’s a figure many markets would envy, but one that deflates expectations of an AI-driven boost: analysts emphasize that almost none of this growth stems from artificial intelligence.

The headline chosen by The Next Web leaves no room for misinterpretation: this is an energy story, not an algorithmic one. Traditional sectors—oil, gas, utilities—are propelling European profits, not the revolution of Large Language Models or GPU acceleration. While Wall Street treats each Nvidia quarter as proof that generative AI is reshaping business, Europe’s corporate finance reality tells a different tale: the digital transition, at least in revenue terms, hasn’t yet made a tangible mark on the balance sheets of large firms.

This disconnect between the U.S. tech narrative and the European accounting truth has deep implications for anyone planning or evaluating on-premise AI infrastructure. If profits aren’t coming from AI, companies have less incentive to invest in specialized hardware, from NVIDIA H100 GPU servers to bare-metal clusters for inference. It’s no coincidence that the European data center market is growing more cautiously, often constrained by data sovereignty requirements and uncertainty about real demand. For those focused on Total Cost of Ownership, the message is clear: before committing capital to self-hosted LLM environments, you need to see concrete economic returns, not just a promise.

The energy sector’s strength, moreover, serves as a reminder that Europe values stability over disruption. In a continent where GDPR and data residency rules mandate localization-aware architectures, the fact that profits come from regulated, mature industries might paradoxically encourage a more measured approach to AI. Companies can afford to build pipelines and fine-tuning environments without the pressure to monetize quickly, concentrating efforts on use cases where control and compliance matter more than speed. It’s a scenario that scales back the hype but doesn’t extinguish interest in on-premise AI—it simply brings it down to a more sustainable scale, in line with the continent’s economic reality.