Schneider Electric: AI can not only consume energy, but also save it

Artificial intelligence devours electricity on an unprecedented scale — and simultaneously can become the most powerful energy-saving tool in history. This is the thesis put forward by Schneider Electric CEO Olivier Blum, and in this paradox lies, perhaps, the principal energy intrigue of the coming decade.

Schneider Electric is not simply a large company. It is a French industrial giant with revenues exceeding €35 billion, holding dominant positions in the electrical power management and industrial automation markets. When the head of such a corporation makes a statement about AI's energy potential, it merits attention: Schneider Electric supplies equipment and software for data centers, industrial enterprises, and commercial buildings worldwide. The company sees the situation from the inside — and from both sides of the equation.

The context of Blum's statement cannot be overstated. Over the past two years, global data center energy consumption has grown many times over, spurred by the generative AI boom. By various estimates, by 2028 global electricity consumption by data centers could reach 1000–1500 terawatt-hours per year — comparable to the energy consumption of an entire Japan. Hyperscalers like Microsoft, Google, and Amazon are snapping up nuclear power plant capacity, investing in thermonuclear fusion, and building solar farms across thousands of hectares. The problem of powering AI has shifted from an engineering task to a geopolitical factor.

Against this backdrop, Blum proposes viewing the situation from another angle. Yes, AI consumes enormous amounts of energy. But these same technologies are capable of radically optimizing energy consumption in buildings, industrial facilities, and entire power grids. We're not talking about theoretical models, but about quite concrete applications: intelligent microclimate management systems in buildings, predictive maintenance of power grids, real-time transformer substation load optimization, and demand-supply balancing in energy systems with high renewable energy penetration. Schneider Electric is already implementing such solutions, and the results are impressive: in certain projects, electricity savings reach 20–40 percent.

It is important to understand the scale. Buildings consume approximately 30 percent of all global electricity, industry another 40 percent or so. If AI can reduce this consumption by even 10–15 percent, the savings would multiply exceed all the electricity consumed by data centers. In essence, Blum paints a picture in which AI doesn't simply "offset" its own energy costs, but creates a net positive effect for global energy systems. This is a fundamentally different narrative compared to the alarmist forecasts that have dominated media discourse over the past year.

That said, skeptics rightly note that the Schneider Electric CEO has an obvious commercial interest in promoting this narrative. The company actively sells "smart" energy management solutions, and growing demand for AI optimization directly increases its revenue. Nevertheless, Blum's position is confirmed by independent research. The International Energy Agency in its latest report indicated that digitalization and intelligent management could reduce global energy consumption by 10 percent by 2030 — provided there are targeted investments and proper regulation.

For the Russian market, this topic is particularly relevant. Energy intensity of the Russian economy remains among the highest in developed countries, and the potential for AI optimization in industry and housing utilities is virtually untapped. Meanwhile, development of its own AI infrastructure already creates additional strain on the power system. Schneider Electric's experience shows that these two processes need not contradict each other — with the right approach, AI can develop and simultaneously help save resources.

Blum's statement marks an important shift in the discussion about artificial intelligence's energy footprint. The conversation is gradually moving from simple accounting of megawatts consumed by data centers to more complex and honest calculation — how much energy AI helps save on the other end of the chain. And if this arithmetic proves correct, the paradox of energy-hungry AI-as-energy-saver could become one of the defining stories of the technological decade.