Attacks on data centres in Ukraine, Russia and the Middle East are raising concerns over the security of artificial intelligence infrastructure, prompting calls for European governments to rethink where and how they build major computing facilities.

Mehdi Paryavi, an adviser to the United Nations and founder of the International Data Center Authority, said recent strikes demonstrated how vulnerable data centres had become as military targets and warned that damage to critical facilities could disrupt entire economies.

“When you hit enough important targets, you can take down the entire economy,” Paryavi told Euronews Next.

He said attacks on data centres could affect essential services, including banking, healthcare and military operations, as countries become increasingly dependent on digital systems.

Russia has struck data centres in Kyiv, while Ukrainian drones have targeted multiple facilities in Russia in recent weeks. Kyiv has described some of its attacks as targeting infrastructure that generates revenue for Moscow’s war effort.

Data centres house servers and store the information needed to run digital services. They are also essential to the development of AI because they provide the computing capacity and energy required to train and operate advanced systems.

Their strategic importance has increased as Ukraine uses artificial intelligence for drone targeting and other military applications.

Paryavi said traditional approaches to choosing data centre locations and keeping them discreet might no longer provide sufficient protection. He suggested that countries may need to consider underground facilities, reinforced bunkers and air defence systems.

“The days that you would just build data centres in random locations are over,” he said.

However, he acknowledged that governments and the technology industry had yet to find a clear solution that would improve security without slowing the expansion of AI computing capacity.

European AI expansion faces security questions

Asked whether the attacks would prompt European governments to reconsider plans for large-scale AI facilities, known as AI gigafactories, Paryavi said they would.

The European Commission told Euronews Next that a tender had been launched to establish up to seven AI gigafactories across Europe. The application process was scheduled to close on November 12, after which proposals would be evaluated and decisions made.

The Commission did not specifically address whether attacks on data centres would affect its plans.

Paryavi warned that the financial consequences of a major attack could be severe. Individual data centres can cost as much as $25 billion to build and take up to three years to complete, making destroyed facilities expensive and difficult to replace.

“Europe cannot afford to finance hundreds of billions of dollars of assets and then somebody takes them out,” he said, adding that the pace of construction might need to slow.

He urged European countries to coordinate their approach rather than develop infrastructure independently. Shared planning, he argued, could help address security threats alongside energy shortages and a lack of skilled workers.

Paryavi also suggested concentrating data centres in designated digital hubs instead of scattering facilities across multiple locations. Such clusters could make it easier to establish coordinated protection systems, including air defence.

The debate highlights a growing challenge for governments seeking to expand AI capacity while protecting the infrastructure on which digital economies increasingly depend.

Europe is accelerating plans to establish its own satellite communications network as restrictions affecting Starlink services in Ukraine highlight the risks of relying on privately operated systems for military operations.

The Ukrainian army reported restrictions to Starlink on Wednesday, saying connections dropped when receivers travelled at speeds exceeding 160 kilometres per hour, including during drone operations. Although the service was partially restored, a speed limit remained in place. It was unclear who ordered the restrictions.

SpaceX chief executive Elon Musk denied that his company had ordered the shutdown, saying on social media platform X that Starshield was under the direct control of the US government.

Starlink provides satellite internet services to consumers and businesses, while Starshield is designed for government and military customers. The distinction has renewed concerns about how much control governments and armed forces have over the communications systems they depend on.

Polish Foreign Minister Radosław Sikorski criticised Musk publicly, warning that European customers were closely monitoring the company’s decisions. Poland has financed a significant share of the Starlink terminals used by Ukraine.

The European Union has been developing its own alternative, known as IRIS², short for Infrastructure for Resilience, Interconnectivity and Security by Satellite. The network is intended to provide secure communications for governments, businesses and citizens, while improving coverage in areas of Europe and Africa where connectivity remains limited.

Under the project, the network will comprise 348 satellites, including 330 in low Earth orbit and 18 in medium Earth orbit. A further 66 satellites are planned to strengthen defence, security and emergency services. The European Commission estimates that the additional capacity will increase secure communications capacity for public-sector users by 60% within the EU and 54% worldwide.

The first satellites are scheduled to launch in 2029, with initial services expected to become available gradually from 2029 or 2030. The network is also expected to use quantum communication technology through the European quantum communication infrastructure, EuroQCI.

Germany will play a significant role in the programme through Bremen-based satellite manufacturer Orbitale Hochtechnologie Bremen (OHB). On August 31, OHB signed a contract worth almost €1 billion with satellite operator SES to supply 18 medium-orbit satellites.

Each satellite will weigh around 2.6 tonnes at launch. OHB chief executive Marco Fuchs has described the project as an important step towards greater European independence in space communications.

Belgian company Aerospacelab, founded in 2018 and based in Charleroi, is building 264 low-orbit satellites under a contract valued at €2.4 billion, according to the company. Production is expected to take place at a new factory designed to manufacture up to 500 satellites annually from 2027. Airbus is building at least 66 additional satellites for military communications, completing the allocation of the low-orbit segment.

Germany is also planning a separate satellite network for its armed forces, known as SATCOMBw 4. The project is intended to operate under national control and eventually include hundreds of low-orbit satellites. An initial constellation of up to 40 satellites is planned to become operational in 2029, with full operations targeted for 2030. The estimated cost is between €8 billion and €10 billion.

Defence Minister Boris Pistorius announced in September 2025 that Germany would invest €35 billion in military space projects by 2030. A joint venture between OHB and Rheinmetall is bidding for the SATCOMBw 4 contract, but no award has been announced.

Meeting the timetable for Europe’s satellite programmes remains uncertain. Experts have identified rocket-launch capacity as a major obstacle, with Europe relying on Ariane 6, which has a limited launch schedule. The European Commission has identified securing launch services as a key next step for IRIS².

The projects are intended to give Europe greater control over secure communications, reducing its dependence on foreign providers at a time when satellite connectivity has become increasingly important to military operations and national security.

OpenAI is donating $5 million to the Simons Foundation to help mathematicians examine and build on results generated by artificial intelligence, days after criticism from a mathematics organisation over the company’s claims about solving a longstanding mathematical problem.

The donation follows an announcement by OpenAI in September that an unreleased AI model had produced a major result involving the Navier-Stokes equations, which describe the motion of fluids and are based on principles of classical physics. The equations have challenged mathematicians for almost a century.

The Association for Human Mathematics (AHM) criticised OpenAI’s work, arguing that the company had disregarded established scientific research practices. The organisation said mathematicians had not requested the work and accused the company of showing a “total disregard for the norms of scientific research”.

According to TechCrunch, two human mathematicians said they had been working towards a similar result and questioned whether OpenAI’s research had drawn on their work. OpenAI has denied seeing their research.

One of the mathematicians, Levent Alpöge, works at rival AI company Anthropic and was pursuing the research in a private capacity.

On October 6, OpenAI said its model had produced more than 700 manuscripts and nearly 400 additional mathematical results. The company said the work aimed to address some of the most difficult questions in mathematics and computer science.

The controversy has raised questions about how AI-generated mathematical research should be evaluated, particularly when its findings overlap with work being conducted by human researchers.

OpenAI announced the donation on Thursday in a post on social media platform X, saying the funding would support a new Simons Foundation initiative designed to help researchers assess AI-generated results and use them to advance mathematical understanding.

“We want to support the mathematics community in turning progress in AI into deeper human understanding and new discoveries,” OpenAI said.

The company said the money would fund workshops, conferences and special programmes. It also pledged to provide access to its AI models and computing resources for researchers supported by the Simons Foundation.

The initiative will bring together $15 million in combined funding and resources from the Simons Foundation, OpenAI and Anthropic.

The Simons Foundation focuses on supporting scientific research, including mathematics. The new programme is intended to help researchers scrutinise AI-generated findings, understand their significance and develop further work based on them.

OpenAI’s contribution comes as the use of AI systems in scientific research expands, prompting debate over research transparency, attribution and the role of human experts in validating machine-generated discoveries. The dispute over the Navier-Stokes equations has brought those questions into focus within the mathematics community.