Taiwan's rapid expansion in artificial intelligence and advanced semiconductor manufacturing is placing new pressure on an energy policy built around a nuclear phase-out. As computational power becomes increasingly central to economic competitiveness, the island faces a fundamental question: can its existing electricity strategy sustain the scale of demand the AI era now requires?
Around the world, energy is emerging as a defining constraint on AI development. In a recentpodcast interview with Dwarkesh Patel, Tesla CEO Elon Musk suggested that grid capacity and transformer shortages may soon eclipse chip supply as the primary bottleneck for artificial intelligence. While some of Musk's broader projections remain speculative, concerns about electricity availability are widely shared across the technology sector.
According tothe International Energy Agency, global data center electricity consumption accounted for roughly 1.5% of total power use in 2024 and is projected to exceed 3% by 2030—effectively doubling within six years. In the United States, electricity demand from data processing could rival or surpass that of several heavy industries combined, including steel, aluminum, cement and chemicals. Some technology firms have responded by securing long-term nuclear supply agreements or investing directly in generation assets to ensure reliable, dispatchable power.












































