Major institutional investors have definitively moved artificial intelligence from the category of experimental technologies into the realm of everyday work tools. According to a fresh analysis based on a survey of 410 asset managers from North America, Europe, the Middle East, and Asia, AI has firmly established itself in analytical work but is still barely used in direct trading and trade execution. This confirms the tectonic shift in the industry that I predicted at the beginning of the year.

AI Leads in Analytics, But Does Not Replace Humans

The numbers are telling: 52% of long-term fund managers and asset owners use AI specifically for research and information retrieval. Another 44% of hedge funds use neural networks for processing market data. At the same time, the deepest level of adoption is observed precisely in hedge funds—72% of their representatives admitted to using AI daily. For comparison, among long-term fund managers, this figure is 49%, and among asset owners, it is 38%.

However, it is important to emphasize: final decisions are still made by humans. In trading and trade execution, the influence of AI remains minimal, and data security concerns are cited as the main barrier to broader adoption. Notably, only 7% of respondents expect significant staff reductions due to the transition to AI. The majority predict an increase in workload while maintaining current staffing levels.

Infrastructure Deadlock: Andreesen's Position

Marc Andreessen, co-founder of Andreessen Horowitz, recently identified the key limitation to AI development not as algorithms or data, but as energy and cooling infrastructure. His thesis organically continues the theory of the paramount importance of energy surplus for technological progress.

"In the future, in every country, the scale of AI development will directly depend on the availability of cooling systems," Andreessen wrote. And he is absolutely right. AI servers require enormous energy consumption, and cooling these facilities places additional strain on infrastructure. The International Energy Agency predicts that by 2030, demand for data center capacity will more than double, reaching approximately 945 terawatt-hours—almost comparable to the current energy consumption level of all of Japan.

The most serious infrastructure challenges are expected for the U.S. economy. According to expert estimates, in the coming years, computing centers in the U.S. will consume more resources than the entire national heavy industry. New AI clusters will concentrate in regions with cheap and stable power generation.

What This Means for Investors

The described trends paint a clear picture of the future. On one hand, commercial demand for AI technologies has already been established and continues to grow. On the other hand, the physical limitations of the market are becoming increasingly apparent. Energy efficiency and innovative cooling will become key factors determining future technological leaders.

Four giants—Microsoft, Amazon, Alphabet, and Meta—have already approved massive investment plans for 2026. Their total investment volume is planned at $725 billion, which exceeds last year's figures by a full 77%. The ability of global energy systems to cope with such a load will predetermine the outcome of the new technological race.

Expert opinion from Cryptalist: The market is clearly underestimating infrastructure risks. While all attention is focused on the competition of models and algorithms, the real "bottleneck" is energy. Investors focused on long-term growth should take a closer look at companies specializing in energy-efficient solutions and cooling systems for data centers. They could become the "second-order beneficiaries" of the AI boom.