AI’s Next Bottleneck Isn’t Chips — It’s Electricity

AI’s Next Bottleneck Isn’t Chips — It’s Electricity ⚡🤖 The AI boom is usually described as a race for chips. But the next major bottleneck may be something much more basic: Electricity. Modern AI data centres can require enormous amounts of reliable, around-the-clock power. And as companies race to build bigger clusters of GPUs, the energy infrastructure behind them is becoming just as strategically important as the chips themselves. In this episode of Behind the Business, we explore the power race behind artificial intelligence — and why nuclear plants, gas turbines, transmission lines, substations, transformers and grid access are suddenly becoming some of the most valuable assets in the AI economy. According to the International Energy Agency, global data-centre electricity consumption could rise from around 485 TWh in 2025 to roughly 950 TWh by 2030. That would take data centres from around 1.5% of global electricity demand to roughly 3%. AI-focused data-centre power consumption alone is projected to roughly triple over that period. And the pressure is already showing up in the real world. In October 2026, Google and Constellation announced a 3,590 MW U.S. power agreement, including a 20-year deal for 890 MW of nuclear power. Amazon has signed its own long-term nuclear agreement. Microsoft is backing the planned restart of the former Three Mile Island Unit 1 — now called the Crane Clean Energy Center. And projects linked to OpenAI’s Stargate buildout have already faced commercial risks tied to potential power delays. The AI race is no longer only about semiconductors. It is becoming a race for megawatts. ⚡ In this episode: • Why AI data centres consume so much electricity • How global data-centre power demand could nearly double by 2030 • Why grid access is becoming an AI bottleneck • Google’s 3.59 GW Constellation power deal • Why Big Tech is turning to nuclear energy • Amazon and the Calvert Cliffs nuclear agreement • Microsoft and the planned Crane Clean Energy Center restart • Why gas turbines are also becoming valuable again • The PJM grid shortfall debate • Why transformers and switchgear can delay new data centres • Oracle’s Project Jupiter power-risk episode • How Japan, India and Finland are competing for AI infrastructure • Who benefits from the AI electricity boom • Why local grids may struggle even if global power supply remains sufficient • Whether the AI race will ultimately be won in chips — or megawatts The AI revolution may be built on GPUs. But none of those GPUs work without power. Copyright / Disclaimer This video is an independent educational and documentary-style production and is not affiliated with, endorsed by, or sponsored by Google, Microsoft, Amazon, Oracle, OpenAI, Constellation Energy, Meta, JERA, Reliance, PJM, the International Energy Agency, or any other company or organisation mentioned. Names, trademarks and branding referenced in this video remain the property of their respective owners. Visual material used in this production is original, generated, licensed, or sourced from stock-media libraries where applicable. This video is produced for educational, commentary and informational purposes only and does not constitute financial or investment advice. Subscribe to Behind the Business for documentaries about the companies, infrastructure, commodities and industries shaping the global economy. #AI #ArtificialIntelligence #DataCenters #Electricity #NuclearPower #Energy #BusinessDocumentary #BehindTheBusiness