Deutsch
English
Español
Français
Nederlands
Português
The artificial intelligence revolution is colliding with a stark physical reality: the global electrical grid cannot keep up. By 2026, data center electricity demand is projected to more than double, surpassing 1,000 terawatt-hours (TWh) annually — roughly equivalent to Japan's entire power consumption. This AI-driven energy hunger is causing unprecedented strain on aging infrastructure, with up to 50% of planned data center projects facing delays due to grid capacity shortages and transformer lead times stretching to 36–48 months. The bottleneck has shifted from GPU supply to electrical infrastructure, reshaping energy markets, industrial policy, and the geopolitics of computing power.
The Scale of the Crisis
According to the International Energy Agency (IEA), global data center electricity consumption is on track to double from 460 TWh in 2024 to over 1,000 TWh by 2026. Morgan Stanley warns of a 126 GW surge in data center power demand through 2028 — nearly matching Canada's entire generating capacity — with a 49 GW shortfall expected in the United States alone by 2028. The PJM Interconnection, which operates the grid for 65 million people across 13 states, saw its capacity auction fall 6.6 GW below reliability requirements for the first time in history. Capacity prices have spiked tenfold, with data centers driving $9.33 billion in additional capacity costs.
The global energy crisis is no longer speculative. Communities in Ohio, Oregon, Georgia, and Virginia are pushing back, demanding tech companies fund their own infrastructure rather than shifting costs to ratepayers. AEP Ohio has frozen new data center interconnections, while Dominion Energy in Virginia — home to the world's largest data center market — has warned that demand could outpace supply by 2027.
Bring-Your-Own-Power: Big Tech's Nuclear Pivot
Faced with grid interconnection queues totaling 2,600 GW and delays of up to seven years, hyperscalers are adopting a 'bring-your-own-power' strategy. In 2026, tech giants are investing directly in power generation at an unprecedented scale. Microsoft has signed a deal to restart Three Mile Island Unit 1 (835 MW) by 2027, marking the first-ever restart of a retired U.S. nuclear plant. Google has inked a 500 MW agreement with Kairos Power for small modular reactors (SMRs), while Amazon has committed $700 million to X-energy SMRs and is spending $20 billion on the Susquehanna nuclear plant.
In total, 13 announced nuclear-powered data center projects represent over 9.8 GW of capacity. However, no commercial SMRs are yet operational in the United States, leaving natural gas as a temporary bridge — a move that directly contradicts national climate targets. NextEra Energy has noted that hyperscalers are leaning heavily on gas-fired capacity to meet surging demand, with the company's CEO stating: 'The data center developers are financing their own generation, and they are overwhelmingly choosing gas for firm power.'
Grid Infrastructure: The Weakest Link
Transformer lead times have ballooned to 36–48 months, with some utilities reporting waits of up to five years. The supply chain bottleneck is compounded by a shortage of skilled labor and raw materials. Gartner predicts that power shortages will restrict 40% of AI data centers by 2027. The gap between AI capital expenditure — which exceeds $650 billion in 2026 — and actual energized megawatts is now the widest on record.
Wholesale electricity prices near data center hubs have risen up to 267%, with U.S. households potentially facing $15–25 monthly increases. The rising electricity costs are sparking political backlash, with regulators in multiple states considering new tariffs that force data centers to pay for grid upgrades.
Impact on Climate Targets
The AI power crunch is directly undermining national climate commitments. The United States, which aims for a carbon-free grid by 2035, is now seeing utilities delay coal plant retirements and extend natural gas operations to meet data center demand. The EU's Carbon Border Adjustment Mechanism (CBAM), effective January 2026, adds further compliance costs for data center operators importing power. In Asia, countries like Malaysia and Singapore are grappling with similar tensions between AI growth and sustainability goals.
The climate policy conflicts are intensifying as tech companies' renewable energy purchases fail to keep pace with demand. While Google, Microsoft, and Amazon remain the world's largest corporate buyers of renewables, the sheer scale of AI's energy consumption means that even record-breaking solar and wind installations cannot fully offset the carbon footprint of new gas plants.
Expert Perspectives
Dr. Fatih Birol, Executive Director of the IEA, has warned: 'The growth in data center electricity demand is one of the most significant developments in global energy markets this decade. Without urgent investment in grid infrastructure and clean power generation, we risk a scenario where AI progress comes at the expense of climate progress.'
Similarly, a senior analyst at Goldman Sachs noted: 'The bottleneck has moved from chips to power. We are now in a race between AI deployment and grid expansion, and the grid is losing.'
FAQ
How much electricity will AI data centers consume by 2026?
Global data center electricity demand is projected to exceed 1,000 TWh annually by 2026, more than double the 460 TWh consumed in 2024.
Why are data center projects being delayed?
Up to 50% of planned projects face delays due to grid capacity shortages, transformer lead times of 36–48 months, and interconnection queues totaling 2,600 GW.
What is the 'bring-your-own-power' strategy?
Tech giants like Microsoft, Amazon, and Google are directly investing in power generation — including nuclear plant restarts and natural gas facilities — to bypass grid constraints.
How does this affect climate goals?
The reliance on natural gas as a bridge fuel and delays in coal plant retirements are directly contradicting national climate targets, raising concerns about carbon emissions.
Will electricity prices increase for households?
Yes. Wholesale electricity prices near data center hubs have risen up to 267%, and U.S. households may face monthly increases of $15–25 as utilities pass on infrastructure costs.
Conclusion: The Geopolitics of Computing Power
The AI energy crisis of 2026 marks a fundamental shift in the relationship between technology and energy. The geopolitics of computing power are now defined not by chip manufacturing capacity but by access to reliable electricity. Countries with robust grid infrastructure and nuclear capabilities — such as France, Canada, and the United Arab Emirates — are positioning themselves as AI hubs, while others risk being left behind. The coming years will test whether the world can reconcile AI's insatiable energy appetite with the urgent need for decarbonization.
Follow Discussion