Taiwan to spend $3 billion turning nation into 'AI island', targets top five global compute power — new goals threatened by energy shortfalls
Government-backed initiative aims for global top five in compute.
Taiwan has formally committed over NT$100 billion (US$3.2 billion) to a new national initiative to turn the island into a global hub for artificial intelligence. The funding, reported by Nikkei Asia on Tuesday, November 18, will support a ten-point strategy focused on next-generation hardware, with government officials naming silicon photonics, quantum computing, and AI robotics as priority areas for research and development.
The AI push is being modelled after the "10 Major Construction Projects" of the 1970s, a campaign widely credited with modernising Taiwan’s infrastructure. This time, the government wants AI to serve as the new industrial core, capable of generating NT$7 trillion in added value by 2028 and NT$15 trillion by 2040. The draft 2026 budget alone sets aside more than NT$30 billion for early-stage deployment.
Premier Cho Jung-tai and President Lai Ching-te have both repeated the goal of placing Taiwan among the world’s top five countries in computing power. That includes a new national AI data center to be built in Tainan, backed by public funding, alongside a growing list of private deployments in Kaohsiung and other cities. Foxconn and Nvidia are already scaling a joint facility in Kaohsiung that is expected to hit 100 megawatts at full capacity, using Nvidia’s Blackwell platform.
However, sustained growth at this scale may run into energy constraints. Taiwan’s last nuclear power plant went offline in May, and the island’s renewable energy capacity remains below target. Various reports have confirmed that shortfalls exist in Taiwan’s offshore wind generation, and transmission expansion remains slow in southern regions where new data centers are being concentrated.
To compensate, Taiwanese operators could begin shifting to newer high-efficiency infrastructure. Nvidia has championed an 800-volt DC busbar system for data center power delivery, and the company could use the upcoming Nvidia-Foxconn Kaohsiung facility to promote those specifications to local partners. GMI Cloud, another early adopter, recently announced that it will deploy 7,000 Blackwell GPUs in Taiwan as part of a 16 megawatt build optimised for dense inference workloads.
Even with government support, Taiwan’s ambition to become a global compute power will depend on whether new racks can be energised and cooled at a pace that matches the roadmap. Private-sector investment is growing, but the scale of planned deployments will test the island’s grid far beyond historic chipmaking demand.
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.

Luke James is a freelance writer and journalist. Although his background is in legal, he has a personal interest in all things tech, especially hardware and microelectronics, and anything regulatory.
-
pug_s Good luck with that. US is having AI GPU's collecting dust because of insufficient electricity. China is the only country is adding electricity capacity.Reply -
thisisaname Reply
Quite a lot of that increase is down to then burning coal, not all but a significant part of it.pug_s said:Good luck with that. US is having AI GPU's collecting dust because of insufficient electricity. China is the only country is adding electricity capacity. -
A Stoner There must be a mistake here. $3 billion is quite a bit of money to people, but with respect to AI spending it is not even part of a rounding error anymore.Reply