Foxconn Chairman Young Liu announces AI data center with NVIDIA

Young Liu, Chairman of Hon Hai Technology Group (Foxconn), delivered a keynote speech on May 20 at Taipei Nangang Exhibition Center Hall 2, announcing that it would build an artificial intelligence (AI) center with Nvidia targeted to have 100 megawatts of power in Taiwan.

Young Liu, Chairman of Hon Hai Technology Group (Foxconn), delivered a keynote speech on May 20 at Taipei Nangang Exhibition Center Hall 2, announcing that it would build an artificial intelligence (AI) center with NVIDIA targeted to have 100 megawatts of power in Taiwan.

At the event, Liu said the AI data center would be built in phases and would be located in Kaohsiung and some other regions of the country, depending on power availability.

“Power is a very critical resource in Taiwan. I don’t want to use the word ‘shortage’. It will take a few steps to reach 100 megawatts, starting with an initial phase of 20 megawatts,” he said.

NVIDIA’s CEO Jensen Huang, appearing as Liu’s special guest, reiterated that the AI factory to be built in Taiwan, announced earlier this week, was intended for use by Foxconn, NVIDIA and Taiwan’s entire ecosystem.

Huang also highlighted the two companies’ collaboration spanning from AI agents for manufacturing inspection to a wide spectrum of robots, such robots orchestrating other robots and even robots building other robots.

Before Huang’s surprise appearance, Liu announced Foxconn’s partnership with TechOrange to establish Taiwan’s leading robotics community, aiming to drive innovation and address the challenges posed by the GDP paradigm shift.

Elaborating on the paradigm shift, Liu said developed countries tend to have insufficient workers for low-pay jobs, and that would be where AI and robotics came in to fill the void, urging leaders of developed countries to pay close attention to the rapid advancement in AI when facing such demographic challenge.

Apart from hardware, Liu also announced FoxBrain, the company’s developing foundation model that would be built on top of NVIDIA’s software stack, would support the company’s new initiative.

The model would be specialized in reasoning, with deep understanding in specific numeric readings that most general AI models were not capable of, Liu said, adding that it would be made open-source to encourage the wider AI community to contribute to make it more powerful and useful.

The model’s capability in reasoning would be required for the company’s smart platforms applications, namely smart manufacturing, smart electric vehicles and smart city, that were part of the company’s new initiative, Liu said.

The three intelligent platforms would be powered by AI factories, Liu said.

Envisioning future factories, he anticipated that the Omniverse digital twin factory would first be created, then came the AI factory to process the data from the former to create AI models, and finally the optimized physical factory that deployed and adapted the AI models would be constructed.

Liu said he believed that the demand for compute power would only continue to grow exponentially due to the proliferation of AI applications and the ongoing evolution of AI models.

Hide
Show

Discover more from COMPUTEX Daily - COMPUTEX 2025 Official Publication Website

Subscribe now to keep reading and get access to the full archive.

Continue reading