This expansion, fueled by increasing global demand for the most advanced AI chips, is putting even more pressure on the island’s grid. Taipower chairperson Wen-sheng Tseng notes that each wafer fab needs around 200MW to operate, suggesting that the multiple facilities planned for and being built around the island will demand 5.3 to 5.4 gigawatts of electricity. This does not count the other plants and factories needed to support the operation of these factories, so it’s likely that the power demand will even be higher than projected. Nevertheless, four new electricity-generating gas units are set to come online this year, with an additional 5.2GW expected to enter trials soon. The company is also working to beef up Taiwan’s grid and enhance resilience plans, ensuring that the island’s power supply can keep up with the demand, particularly from science parks and industrial zones.
Despite that, Taipower is still warning that demand is also increasing outside of these areas. Some fabs and servers are setting up facilities beyond these designated sectors, sometimes even requesting up to 20% of the power consumption of a science park for just a single structure. Because of this, Tseng says that both fabs and tech companies should be careful when selecting sites for their future infrastructure, ensuring that they’ll only be built in places where there is ample power supply to avoid bottlenecks in the future.
U.S. electricity grid stretches thin as data centers rush to turn on onsite generators
Anthropic promises to pay for electricity price increases due to it's AI data centers
Nvidia-backed trial shows AI data centers can flexibly adjust power use in near real time, with global implications for energy consumption
Even though Taiwan’s power utility company is warning about power consumption, it seems to be on track to cover the future growth of the AI and semiconductor industry, provided that the current projections hold. This is a far cry from what the U.S. is experiencing, with the White House resorting to forcing big tech companies to promise that they will pay for the grid and power source upgrades needed to supply the massive amounts of electricity that AI data centers need.
Follow Tom's Hardware on Google News , or add us as a preferred source , to get our latest news, analysis, & reviews in your feeds.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Key considerations
- Investor positioning can change fast
- Volatility remains possible near catalysts
- Macro rates and liquidity can dominate flows
Reference reading
- https://www.tomshardware.com/tech-industry/SPONSORED_LINK_URL
- https://www.tomshardware.com/tech-industry/taiwan-expects-power-demand-to-increase-by-more-than-5gw-by-2030-enough-to-power-nearly-4-million-homes-rise-in-electricity-consumption-driven-by-semiconductor-manufacturing-and-ai-data-center-deployments#main
- https://www.tomshardware.com
- Labubu sues 3D printer maker Bambu Lab for items made by its users — MakerWorld design repository in hot water over IP theft by its users
- OpenAI building GitHub alternative after frequent platform outages and disruptions — a public OpenAI code repository would directly compete with one of its bigg
- M4 MacBook Air with 24GB of RAM plunges to lowest-ever price following M5 announcement — $300 off is a steal on Apple's highest-spec RAM configuration
- New SemiAnalysis InferenceX Data Shows NVIDIA Blackwell Ultra Delivers up to 50x Better Performance and 35x Lower Costs for Agentic AI
- DDR5 RAM kits skyrocket to an astonishing $4,000 on Newegg with wild price listings on multiple G.Skill and Corsair products — various speeds and capacities all
Informational only. No financial advice. Do your own research.