AI is turning energy into a strategic bottleneck for data centers

May 17, 2026 3 min read

Rising AI workloads are increasing pressure on electricity consumption, cooling systems and water usage, with AI projected to account for nearly 70% of total data center energy demand in the coming years.

Table of Contents

The expansion of artificial intelligence is turning energy into one of the most strategic resources for data center operations.

The sector is facing growing pressure around electricity consumption, cooling and sustainability as high-demand AI models continue to expand.

The rapid growth of artificial intelligence is increasing pressure on the energy and environmental infrastructure of data centers. Every query made to platforms such as OpenAI’s ChatGPT or Microsoft Copilot requires electricity and water resources in data centers operating under rising computational demands.

In that environment, energy has evolved from a technical component into a strategic and competitive resource within the AI-driven digital economy.

The International Energy Agency projects that electricity consumption from data centers could double by 2026. At the same time, estimates from Deloitte indicate that the sector’s energy demand could grow by as much as 300% by 2035, driven primarily by AI-related workloads.

Related Posts:

The firm also estimates that AI could account for nearly 70% of total energy consumption across these technology complexes in the coming years.

Luis Santamaría, Cloud and Service Provider Segment Leader at Schneider Electric, said:

“The industry has been changing very quickly: we now have greater density, higher energy dependence, more pressure for speed and less tolerance for error.”

According to the executive, energy management can no longer be addressed solely from a technical perspective and has become part of broader business strategy decisions.

High-density infrastructure and new cooling systems.

Traditional rack designs have evolved toward configurations exceeding 40 kilowatts per unit, increasing demands on cooling systems and operational efficiency.

Cooling systems can account for up to 40% of a data center’s total energy consumption, driving new strategies focused on energy optimization.

Among the sector’s main trends are:

Liquid cooling and direct-to-chip cooling;
free cooling systems;
modular and scalable architectures;
optimization of Power Usage Effectiveness (PUE).

These technologies are designed to reduce energy consumption, improve thermal efficiency and enable progressive capacity expansion without oversizing infrastructure.

However, the challenge goes beyond simply adding more electrical capacity. Energy planning is increasingly becoming a strategic factor in responding to simultaneous demand spikes, power failures and the rapid expansion of AI-driven workloads.

Poor energy design can increase operating costs, create infrastructure overprovisioning or compromise availability during periods of accelerated demand growth.

The International Energy Agency warns that artificial intelligence could represent more than 4% of global electricity demand before 2030.

Water consumption and sustainability amid regional expansion.

In Latin America, the data center market reached an estimated value between US$5 billion and US$6 billion in 2023 and could double by 2029, according to estimates from the United Nations Development Programme.

However, regional expansion will depend on factors such as competitive energy availability, regulation and environmental sustainability.

Water consumption is also emerging as one of the main challenges facing digital infrastructure. According to the United Nations Development Programme, a data center can consume between 10 and 50 times more electricity than a standard commercial building and use up to 25.5 million liters of water per year for cooling.

That figure is equivalent to the approximate annual consumption of 300,000 people, making it a particularly sensitive issue in regions experiencing water stress.

Santamaría said the structural shift occurs when energy, software and operations function as an integrated system. He added:

“The organizations moving forward are the ones that understood that the data center is no longer just infrastructure, but a critical digital platform that must be managed collaboratively to operate with greater efficiency, resilience and predictive capacity.”

Globally, the industry is exploring growing investments in renewable energy, energy storage, treated water reuse and emerging technologies such as small modular reactors.

The challenge for the sector is twofold: sustaining AI-driven computing growth without compromising energy stability while simultaneously reducing environmental pressure on critical resources such as electricity and water.