Yazılar

Belgium Considers Power Limits for Data Centres Amid AI-Driven Energy Surge

/in /tarafından

Belgium’s electricity grid operator Elia is weighing plans to introduce energy allocation limits for data centres, as a wave of AI-fueled demand threatens to strain the country’s power network and crowd out other industries.

Under the proposal, Elia would place data centres in a separate consumption category, giving them a fixed share of grid capacity. The move aims to prevent high-energy facilities from monopolising the grid while still allowing flexible connections that could be curtailed during peak demand or congestion.

The proposal comes as the global race to build AI data infrastructure drives electricity demand to unprecedented levels. In Belgium alone, requests from data centre operators have surged ninefold since 2022, Elia told Reuters. Reserved capacity for 2034 already exceeds twice the 8 terawatt-hours projected in national grid development plans.

“These volumes were not anticipated when Belgium’s grid scenarios were designed,” Elia said, warning that speculative projects risk blocking capacity for other sectors if left unchecked.

The issue will be addressed in Belgium’s next federal grid development plan (2028–2038), Energy Minister Mathieu Bihet told parliament this week. “I will pay particular attention to this during the plan’s approval,” he said.

Belgium’s debate reflects a broader European challenge: balancing energy-intensive AI operations with industrial and environmental goals. Data centres—essential for AI model training and cloud computing—are rapidly becoming one of Europe’s largest sources of new electricity demand.

Tech giants such as Google are already ramping up investment. The U.S. company plans to spend €5 billion ($5.8 billion) expanding its Belgian data centre campuses as part of its global AI strategy.

If approved, Elia’s proposal could make Belgium one of the first European nations to formally cap grid access for AI infrastructure—signalling a shift toward tighter energy governance in the digital age.

Talen Energy Weighs New Data Center Power Plans After Amazon Deal Hits Regulatory Snag

/in /tarafından

Talen Energy is exploring alternative methods to supply electricity to data centers after U.S. energy regulators restricted a groundbreaking co-location deal with Amazon, executives said Thursday.

The move follows a Federal Energy Regulatory Commission (FERC) ruling that capped electricity delivery from Talen’s Susquehanna nuclear power plant in Pennsylvania to Amazon’s data center at 300 megawatts, down from the 960 megawatts originally proposed. FERC cited potential risks to grid reliability and public electricity costs.

Context:

  • Co-located arrangements—where data centers are built near power plants to bypass grid delays—have become attractive to Big Tech and power producers amid surging data demands.

  • Talen’s Amazon deal, announced in early 2024, was a first-of-its-kind setup that aimed to rapidly scale power delivery without waiting for traditional grid connection queues.

Key Developments:

  • Talen CEO Mac McFarland said on Thursday’s earnings call that the company is now considering traditional grid-connected contracts and other commercial agreements to power future data center partnerships.

  • Despite the FERC setback, Talen is still supplying Amazon, with expectations to deliver 120 MW by year-end.

  • The company is appealing the FERC decision and expects a court schedule to be set soon.

Broader Industry Shift:

Talen isn’t alone. Constellation Energy, another major nuclear operator, also announced this week it is shifting focus to more conventional data center energy deals after facing similar challenges with co-location concepts.

Why It Matters:

The FERC ruling marks a pivotal moment in how power for AI and cloud-driven data centers is allocated—especially when nuclear plants are seen as low-carbon, high-capacity sources suitable for tech’s skyrocketing energy needs. The outcome of Talen’s appeal may reshape how future tech-power partnerships are structured.

AI Leaders Urge U.S. Senate to Accelerate Power Permitting, Unlock Government Data for AI Training

/in /tarafından

Top executives from Microsoft, OpenAI, AMD, and CoreWeave will testify before the U.S. Senate Commerce Committee on Thursday, pressing lawmakers to modernize power infrastructure and expand access to federal data to meet the soaring demands of artificial intelligence.

Key Points from Testimonies:

🔹 Brad Smith (Microsoft President)

  • Warns U.S. AI development is hampered by “50-year-old infrastructure”.

  • Calls for streamlined permitting for new energy sources and transmission lines.

  • Urges Congress to unlock federal government data for AI training to stay competitive with China and the U.K.

“The federal government remains one of the largest untapped sources of high-quality data.”

🔹 Sam Altman (OpenAI CEO)

  • Emphasizes growing global reliance on AI:

“We want to build a brain for the world and make it super easy for people to use it.”

  • Says increased AI adoption requires more chips, energy, supercomputers, and training data.

  • Advocates for “common-sense restrictions” to mitigate potential AI harms.

🔹 Michael Intrator (CoreWeave CEO)

  • Highlights the massive energy appetite of AI:

“An insatiable hunger for compute and energy that borders on exponential.”

  • Points to DOE projections: Data centers could consume 12% of U.S. electricity by 2028 (up from 4.4% in 2023).

  • Urges faster approval of generation and transmission projects.

🔹 Lisa Su (AMD CEO)

  • Argues leadership in AI means rapid data center expansion powered by reliable, clean, affordable energy.

  • Stresses the need to extend AI beyond the cloud, integrating it into everyday consumer devices.

“AI must be as accessible and dependable as electricity.”

Context & Urgency:

  • The Senate hearing, titled “Winning the AI Race”, comes as AI’s power and data demands grow exponentially.

  • Leaders argue that regulatory inertia threatens U.S. competitiveness in AI against global rivals.

By linking national competitiveness with infrastructure and data reform, the tech leaders hope to align federal policy with AI’s exponential growth trajectory.