AI’s round-the-clock power needs are colliding with Taiwan’s slow green build-out, forcing a fresh look at its nuclear-free pledge.
The computational demands of the AI era don’t stop at data centers. Advanced fabs now draw power on the scale of large data centers, and two new hyperscale sites can equal a full thermal plant.
Unlike traditional industrial loads, AI servers run 24/7 and demand exceptionally stable, “clean” electricity. As grid stability slips and supply tightens, Taiwan risks losing ground as a global hub if it cannot deliver reliable power.
Renewables were supposed to fill the gap. But progress has lagged targets, leaving a carbon-heavy mix: in 2024, coal was roughly 39%, gas 42%, and renewables about 12%. Projections for 2025 put renewables near 13% — well shy of the 20% goal.
Missed milestones and moving goalposts have compounded the problem. An early aim to “add 2 GW annually” gave way to a 31.2 GW-by-2030 target, underscoring the gap between policy and execution.
After Fukushima, Taiwan doubled down on a “Nuclear-Free Homeland,” letting licenses expire and shelving new plants. That stance became a signature achievement for the ruling party and a legal and political commitment.
The AI surge is now testing that commitment. Public opinion on life-extensions is split, and environmental groups warn that reversing course would betray both law and social consensus. They argue the government cannot “quietly” restart reactors without eroding legitimacy.
Yet practical pressures are mounting. The 2025 energy mix goal — 50% gas, 30% coal, 20% renewables — remains unmet. Nuclear has been phased out on schedule, but low-carbon supply has not backfilled the loss, leaving coal and gas dominant.
Against that backdrop, the Maanshan plant went offline on schedule: Unit 1 in July 2024, Unit 2 in May 2025. A referendum effort to restart the second unit put the issue squarely back on the table.
Proponents make a short-to-medium-term case. Extending Maanshan would stabilize the grid while renewables and storage scale, and it would cut gas reliance at a time of geopolitical risk. Taipower’s own cost figures in early 2025 put nuclear generation near NT$1.87/kWh (US$0.06) — well below offshore wind and utility solar — helping contain system costs and losses. In a crisis, stored nuclear fuel offers months of on-site energy security; advocates argue an operating Maanshan could supply on the order of 5–6% of nationwide load in wartime.
Opponents see the opposite risk calculus. Life-extending a 40-year-old plant means large safety retrofits, inspections, and replacements that can run into the tens of billions of Taiwan dollars. They note that the true “cost of nuclear” must include decommissioning and waste — liabilities already underfunded in the backend fund.
They also stress downside tail risk. If a major incident occurred, cleanup costs could be orders of magnitude larger than any savings, citing Fukushima-scale burdens that no budget could bear.
Safety isn’t just about costs. Taiwan’s plants sit in a seismically active zone, and critics argue that aging components, proximity to faults, and grid-dependency for cooling make large central stations vulnerable in conflict or disaster. They prefer a distributed pathway — rooftop solar, offshore wind, storage, and hardened grids that are harder to knock out.
The unresolved issue of nuclear waste disposal is another major concern. Taiwan lacks a permanent final repository, and waste is currently stored temporarily at the power plants. Extending nuclear operations would only exacerbate this problem, making future disposal even more difficult.
Regardless of the outcome of the nuclear debate, Taiwan must accelerate its energy transition. The government must address public concerns about nuclear power by evaluating life extensions in a scientific and transparent manner, fully disclosing all safety and economic risks.
Simultaneously, it must invest more aggressively in the development of renewable energy and its supporting infrastructure, establishing a clear roadmap and timeline to meet renewable energy targets within the next decade. Improving energy efficiency, developing energy storage technologies, and strengthening grid resilience are also urgent priorities.
Only through a balanced development and intelligent application of a diverse energy portfolio can Taiwan secure its industrial power needs in the age of AI while progressively realizing a low-carbon, sustainable energy future.
