The promise of AI can only be realised if the world learns to expand this technology without exceeding the limits of the energy, water and infrastructure it ultimately relies on.
In just five years, artificial intelligence has gone from being a technological promise to becoming an economic and social phenomenon on a global scale. But its accelerated growth is revealing a less visible face: an unprecedented demand for energy and resources, which challenges the ability of electricity, water and environmental systems to sustain the massive digitalisation of the planet.
The AI revolution is not abstract. It materialises in thousands of data centers around the world, where millions of servers process huge amounts of information. Every generative model, every predictive search, or personalised recommendation requires enormous computing power. This consumption translates into electricity and water, two goods whose availability is already limited in large regions of the planet.
According to estimates by the International Energy Agency (IEA), the electricity consumption of data centers could double before 2030, driven by the expansion of artificial intelligence and cloud computing. Other forecasts are even higher, with scenarios pointing to growth of more than 165% or more if efficiency does not advance at the same pace. The uncertainty is significant, but the message is clear: the pace of expansion of AI far exceeds the current capacity of energy networks to adapt.
Added to this tension is a temporary mismatch in the allocation of capital. Cycles of technological innovation last just two or three years, while electrical infrastructures are designed for periods of twenty or forty years. The risk is obvious: investments that are born modern can become obsolete long before they are amortised.
Governments in the face of a delicate balance
In many countries, the pressure is already being felt. China combines the expansion of its renewables with centralised efficiency mandates; Europe is moving along the regulatory path, demanding transparency and sustainability in new projects; Japan prioritises “green” operators, and Singapore conditions data center licenses on strict consumption standards.
Governments are increasingly aware of what is at stake. In April 2025, a massive blackout affected Spain and Portugal, temporarily paralysing industrial activity and reminding us of the fragility of electricity grids that operate at the limit. Incidents of this type act as a warning of what could happen if digital demand grows faster than the capacity of the system. In a context of accelerated electrification, even advanced economies are at risk of overloading their critical infrastructures.
The challenge is not limited to energy. Water, an indispensable resource for cooling servers, is becoming a critical variable. An average data center uses about 1.14 million liters of water per day, which is equivalent to the consumption of a thousand homes. In regions with water scarcity, this factor has become a source of tension with local communities, especially when the labor benefits of these facilities are scarce. In the United States, neighbourhood opposition has stalled $64 billion worth of projects, reflecting growing social scrutiny over technological expansion.
The Business Answer: Efficiency, Transparency, and Resilience
Faced with this scenario, companies in the technology and energy sectors are beginning to recognise that sustainability is not only a reputational issue, but a condition of business survival. Energy efficiency has become a strategic factor of competitiveness, capable of reducing costs and mitigating regulatory risks.
In our conversations with power and technology companies, we see a convergence of priorities. Large technology groups are reviewing their supply chains and energy consumption with the aim of strengthening security of supply and operational resilience, and could increase integrating efficiency metrics into their tenders and contracts. Power companies are incorporating flexible contracts and planning tools to manage demand shifts, avoid risk of stranded assets and ensure access to affordable energy.
In the traditional energy sector, together with infrastructure operators, companies are beginning to perceive methane leakage reduction and capital discipline as direct levers of profitability and reputation. In Asia, even financial players are conditioning their technological expansion plans on the availability of clean and stable energy.
But just as important as technological innovation is corporate governance, supervision from boards of directors must integrate these risks into investment decisions, avoiding stranded assets or financially unsustainable projects. In turn, transparency towards local communities and regulators will be key to sustaining the social license to operate in a context of growing environmental sensitivity.
A global call to action
The debate on AI and energy is not a conflict between progress and sustainability, but an invitation to rethink the infrastructure of the future. Technology can and should be part of the solution. More efficient chips, less intensive algorithms and circular cooling systems are advances that are already beginning to transform the energy equation of the sector.
However, closer cooperation between companies, regulators and investors will be needed to anticipate the impacts of the digital revolution. The energy transition and the digital transition can no longer move in parallel: they must be integrated.
Spain, despite the spring power failure, also offers a lesson in resilience: with more than half of its electricity in 2024 coming from renewable sources, according to data from the national electricity system operator[1] REE (part of the Redeia group), it shows that energy security can be compatible with sustainability. But even in that favourable context, the challenge remains the same as in any other part of the world: to ensure that the growth of artificial intelligence does not compromise the stability of the system or the well-being of communities.
Artificial intelligence promises a new era of productivity, knowledge, and well-being. But its expansion must be supported on a sustainable basis. Only if we learn to nurture this new intelligence without exhausting our resources can we say that, this time, progress is truly under human control.
Velika Talyarkhan is director for engagement, EOS at Federated Hermes Limited
