In partnership with
Dear Readers,
Artificial intelligence is considered the great promise of our time – but its rapid spread raises an uncomfortable question: What actually powers AI? While most of the attention is focused on models, algorithms, and use cases, a quiet but profound revolution is taking place in the background: the energy hunger of AI systems is growing exponentially – and with it the pressure on power grids, infrastructure, and entire economies.
This DeepDive examines how the race for computing power is increasingly turning into a race for electricity – and why energy policy is suddenly becoming a key strategy in the global AI race. Who will be able to meet the growing demand reliably, cleanly, and affordably – and what geopolitical power shifts will result?
All the best,
Power supply in the age of AI
The TLDR
The explosive growth of AI is causing an unprecedented surge in global electricity demand, with data center consumption expected to more than double by 2030, straining power grids and posing significant environmental challenges. The US is tackling this with a mix of massive renewable energy deals and a renewed interest in natural gas and nuclear power. In contrast, China is geographically relocating its data centers to renewable-rich regions, driven by strong government mandates for green energy.
Demand, bottlenecks, and strategies in the US, China, and Europe
An inconspicuous input into a chatbot—and in the background, thousands of graphics processors start up, drawing megawatts of power and transforming raw energy into statistical probabilities. What seems easy in the digital world is physically difficult: By 2030 alone, data centers could account for almost 1% of the global CO₂ budget if current trends continue. The International Energy Agency (IEA) expects their electricity consumption to rise from 460 TWh in 2024 to around 945 TWh in 2030 – more than the entire country of Japan consumes today.
The reason is clear: the growth of artificial intelligence is shifting data processing from traditional servers to high-density GPU clusters with power consumption of 30 kW per rack and more. This raises a fundamental question:
Who will be able to meet the rapidly growing demand for clean, reliable energy for AI data centers—and which region will come out on top?
Ad
The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.
Unlock a focused set of AI strategies built to streamline your work and maximize impact. This guide delivers the practical tactics and tools marketers need to start seeing results right away:
7 high-impact AI strategies to accelerate your marketing performance
Practical use cases for content creation, lead gen, and personalization
Expert insights into how top marketers are using AI today
A framework to evaluate and implement AI tools efficiently
Stay ahead of the curve with these top strategies AI helped develop for marketers, built for real-world results.
An explosion in demand in figures
“The U.S. is set for "explosive" demand growth over the next five years, Chris McKissack, CEO of battery energy storage systems (BESS) developer Fullmark Energy, told Reuters Events. The U.S. Department of Energy forecasts 20 GW of new data center load by 2030 and predicts data centers will consume 6.7%-12% of total U.S. power production by 2028, up from 4.4% in 2023. This represents a huge margin of error for just three years in the future.” — Reuters
USA: According to the Energy Information Administration's Annual Energy Outlook 2025, computing and storage systems already account for 8% of commercial electricity consumption; by 2050, this figure could rise to 20%. The US Department of Energy estimates that data centers will account for 6.7–12% of the country's electricity demand by 2028.
China: The IEA estimates that Chinese data centers will consume around 100 TWh in 2024; this figure is likely to double by 2027. A government target requires new centers to run on 80% “green” energy by 2025.
Europe: Data centers currently account for just under 3% of the EU's electricity consumption; Brussels expects this figure to rise to 4.5% by 2030. In Ireland, the share was already 22% in 2024.
These figures translate into enormous capacities: Goldman Sachs forecasts a 165% increase in global data center capacity by 2030 – driven almost exclusively by AI workloads.
Scarce resources and growing risks
Demand is colliding with an infrastructure that is already operating at full capacity in many places.
Power grids under pressure (US). The Washington Post warned in 2024 that even large grids such as PJM and ERCOT are reaching their limits. At the same time, connection requests are piling up: Five to ten times more data centers are being planned than are actually being built, which makes planning and investment difficult.
Volatile load profiles. “AI data centers can increase their load tenfold within seconds,” emphasizes Hitachi Energy CEO Andreas Schierenbeck—a dynamic he compares to steel furnaces and believes requires regulation.
Costs and emissions (US). A Business Insider analysis shows that data centers could require almost as much electricity as Florida by 2025. Due to a lack of renewable capacity, utilities are turning back to fossil fuels – with health costs of up to US$9.2 billion annually.
Europe is also struggling with grid bottlenecks. The think tank Ember warns that without accelerated grid expansion, one in five projects could be delayed by 2030.
Comparison of strategies – US, China, Europe
US: from wind PPAs to small modular reactors
Renewable power deals. US data centers have already secured contracts for over 50 GW of clean electricity (mainly solar and wind energy) by the end of 2024 – more than four times the European procurement volume.
Gas as a bridge technology. S&P Global expects additional demand of up to 3 Bcf of natural gas per day for backup turbines and peak loads.
Nuclear option. Microsoft and Constellation are looking into reactivating decommissioned reactors, while the DOE is providing $900 million for modular reactors.
China: West-East pipeline and solar microgrids
Geographical decoupling. The “East-West Computing Corridor” is relocating energy-intensive clusters to wind- and solar-rich provinces such as Inner Mongolia, where a 6,000-petaflops campus is being built that will run on 100% renewable energy.
Political momentum. By 2025, Beijing aims to have 45 EFLOPS of AI capacity, powered by “sustainable, energy-efficient infrastructure.”
Corporate initiatives. Tencent has been operating its own solar microgrid since 2024, which supplies enough electricity for 6,000 households.
Europe: Efficiency package and nuclear comeback light
“Data centres account for 3% of EU electricity demand, but their consumption is expected to increase rapidly this decade due to the expansion of artificial intelligence.” (Reuters)
Regulatory pressure. The EU Commission is working on a package for data center efficiency; operators will have to publish detailed energy performance indicators starting in 2026.
Nordic model. Microsoft is investing US$3.2 billion in Sweden, coupled with almost 1 GW of new wind and hydro power.
SMR alliance. Since February 2024, Brussels has been promoting an industry alliance for small modular reactors; Denmark is even considering ending its 40-year ban on nuclear power.
Network hotspots. In the FLAP-D markets (Frankfurt, London, Amsterdam, Paris, Dublin), capacity is expected to grow by only 55% by 2035, while southern and northern Europe could see an increase of 110% – this is where “data centers are following the wind.”
Who is leading the race?
The US scores points with financial clout and early PPAs, but runs the risk of failing due to grid and transformer bottlenecks. China impresses with state coordination and its west-east approach, but remains heavily dependent on coal. Europe is focusing on efficiency and renewable regions, but is struggling with regulatory delays and a fragmented electricity market.
Conclusion
The analysis shows that no region currently has a perfect model.
The US could win the race if it alleviates grid bottlenecks through SMRs and accelerated line approvals.
China must drastically reduce its coal quota in order to credibly underpin its green ambitions.
Europe, on the other hand, urgently needs grid expansion – otherwise it faces an “AI brain drain” to northern countries and non-European locations.
The path to a climate-neutral AI era does not lie in a single magic bullet, but in a trio of energy efficiency, comprehensive grid expansion, and firm, CO₂-free baseload electricity. Whoever harmonizes these three components first will be able to serve the data demand of the future – and secure social acceptance at the same time.
“AI is one of the biggest stories in the energy world – but only now are policymakers and markets beginning to understand its scope.” – Fatih Birol, IEA
Whether Birol's warning falls on deaf ears or becomes a blueprint will be decided in the next decade. So perhaps the real question is: Can the energy transition learn as fast as the algorithms that drive it?
Sources:
🔗 International Energy Agency (IEA): Electricity 2024 Report – https://www.iea.org/reports/electricity-2024
🔗 U.S. Energy Information Administration (EIA): Annual Energy Outlook 2025 – https://www.eia.gov/outlooks/aeo
🔗 U.S. Department of Energy (DOE): Data Center Energy Use Impact Study – https://www.energy.gov/articles/data-center-energy-use-impact
🔗 S&P Global Commodity Insights: Natural Gas Demand from AI Data Centers – https://www.spglobal.com/commodityinsights
🔗 Business Insider: How Data Centers Are Straining U.S. Power Grids – https://www.businessinsider.com/data-centers-energy-grid-impact-2025
🔗 Hitachi Energy CEO Interview: Transformers and AI Grid Stress – https://www.reuters.com/technology/hitachi-energy-transformers-ai-demand
🔗 McKinsey & Company: The AI Data Center Energy Challenge in Europe – https://www.mckinsey.com/industries/electric-power-and-natural-gas
🔗 European Commission: EU Data Centre Energy Efficiency Directive – https://ec.europa.eu/energy/topics/energy-efficiency
🔗 Tencent Solar Microgrid Announcement – https://www.tencent.com/en-us/articles/solar-microgrid-ai
🔗 Microsoft Press Release: AI Data Center Expansion in Sweden – https://blogs.microsoft.com/blog/2024/06/12/ai-data-center-investment-sweden
Sponsored By Vireel.com
Vireel is the easiest way to get thousands or even millions of eyeballs on your product. Generate 100's of ads from proven formulas in minutes. It’s like having an army of influencers in your pocket, starting at just $3 per viral video.
Chubby’s Opinion corner
The next big battle will not be fought over computing power or algorithms – it will be fought over electricity. After all, intelligence, whether biological or artificial, is ultimately nothing more than a process of energy conversion. The more powerful our AI models become, the clearer it becomes that without a stable, clean, and extremely scalable energy supply, even the most brilliant technology remains a paper tiger.
Whoever solves the energy question will control not only the infrastructure of the future, but also the rules of the game for the economy, climate, and power politics in the 21st century. In this sense, energy policy is becoming a new form of AI strategy – and whoever first achieves the perfect symbiosis of green energy, storage technologies, and grid infrastructure will set the pace of global transformation.
Perhaps we need to accept that the ultimate bottleneck is not data, but the kilowatt hours that move it.