The cloud unit of Taiwanese consumer hardware manufacturer Asus is building an Nvidia-powered supercomputer.

Set to be located in Tainan, southern Taiwan, the system will be developed in partnership with Taiwan AI Cloud and the National Centre for High-Performance Computing.

In an interview with South China Morning Post (SCMP), Asus CEO Peter Wu said the supercomputer will be powered by 1,700 Nvidia H200 GPUs, with an initial capacity of 80 petaflops set to go live in December. This will eventually increase to a total capacity of 250 petaflops, although Wu did not provide a timeline for this.

Once live, the system will reportedly boost Taiwan’s total computing capacity by at least 50 percent. In February 2025, the Taiwanese government’s technology council announced plans to expand the country’s overall compute capacity to around 1,200 petaflops by 2029, a significant increase on the 160 petaflops currently deployed on the island.

Asus has previously partnered with Nvidia for the development of supercomputers in Taiwan, including the nine petaflops Taiwania 2 supercomputer. In 2022, Asus again collaborated with Nvidia to build a supercomputer for use by Taiwanese medical researchers.

Founded as a subsidiary of Asus in 2021, Taiwan AI Cloud (formally Taiwan Web Service Corp) is an initiative that aims to deploy AI and HPC infrastructure for governments, enterprises, and startups.

The project has already supported a number of AI infrastructure projects in East Asia, including the construction of a data center in Singapore and an agreement with Viettel AI for a facility in Vietnam, which is currently under construction but will ultimately see 200 Nvidia GPUs deployed at the site.

However, Wu said similar projects would not be possible in mainland China due to restrictions on exporting AI semiconductors to the country.