"Blackouts, Transition and the Nuclear Taboo," by Tiago Aires Mateus.

This is a full English translation of the original Portuguese text, published in February 2026 print edition of Advocatus magazine.

‍

Blackouts, Taboos, and the Digital Transition

Global electricity generation to power data centers is expected to grow by more than 280% by 2035. Natural gas and coal are projected to supply over 40% of the additional electricity demand from data centers through 2030. From that point onwards, small modular reactors (SMRs) are expected to enter the scene, accelerating the decline of coal-fired generation for this purpose.

Major technology companies know that energy is the key bottleneck for investment in artificial intelligence. That is why hyperscalers are among the leading supporters of nuclear power, particularly SMRs. Nuclear fusion — still at an experimental stage — has also attracted significant investment, as illustrated by the agreement between Trump Media & Technology Group and TAE Technologies (which counts Google among its investors). These companies have made substantial commitments to purchase “clean energy” and to invest in SMRs. Breakthroughs in these and other technologies (such as molten-salt reactors) could create even greater opportunities.

On the other side of the world, China is close to becoming the largest generator of nuclear energy globally, while also beginning to export reactors, technology, and talent. Even Japan — who can forget Fukushima? — is preparing to restart Kashiwazaki-Kariwa, the most powerful nuclear power plant in the world.

In the European Union, both the AI Continent Action Plan and the recent Apply AI Strategy affirm Europe’s ambition to achieve global leadership in artificial intelligence, while acknowledging that doing so will require a major increase in energy consumption and its sustainable integration into electricity systems. Electricity demand from data centers is expected to at least double by 2030 and will account for a significant share of overall demand growth. At the same time, major constraints on grid access are already emerging in hubs such as Dublin, Frankfurt, and Amsterdam.

Against this backdrop, attention now turns to the lessons of the Iberian blackout. Although the final report is only expected on 26 March 2026, many have argued that the blackout resulted from a system with high levels of photovoltaic generation and insufficient inertia: a fault at a Spanish substation may have been enough to trigger the collapse.

In Portugal, the government has argued that many of Spain’s structural weaknesses do not apply domestically. The market, however, is not waiting: applications for grid connections from large data centers already exceed currently installed capacity, with an investment pipeline estimated at around €13 billion by 2030. In November, legislation was published revising the legal framework governing the allocation of grid connection capacity in “high-demand zones.” To date, only Sines has been designated as such a zone, home to the Start Campus project.

At a time when data centers and AI gigafactories require clean, stable, and secure electricity, the vulnerability of the grid to large-scale outages presents an additional challenge.

Energy policy decisions are increasingly carrying electoral consequences, as seen in the debate surrounding the Almaraz nuclear power plant in Spain’s recent elections.

Investment in strengthening and modernizing the grid will be essential, but policymakers must also decide how to respond to rising electricity demand.

The realization of AI-related investments will depend on an energy policy capable of supporting the digital transition without compromising security of supply. Just as mass can be converted into energy, access to stable, secure, low-carbon electricity can be converted into investment. In the aftermath of the blackout, it is time to reassess the role of nuclear energy — with critical thinking, and without irrational fears or ideological taboos.

Tiago Aires Mateus

‍