“Whoever leads in AI, leads the world.”
At the opening keynote of Computex 2025, NVIDIA CEO Jensen Huang announced the company’s latest major initiative: the construction of a giant artificial intelligence (AI) supercomputer in Taiwan. Teaming up with key industry partners including TSMC (Taiwan Semiconductor Manufacturing Company) and Foxconn, NVIDIA aims to establish Taiwan’s first large-scale AI supercomputing system, forming the backbone of the island’s next-generation AI infrastructure and ecosystem.
Taiwan as a Strategic Hub for AI Innovation
Huang emphasized that Taiwan is not just a vital node in the global semiconductor supply chain, but also a strategic center for AI development. The planned supercomputer will be pivotal in supporting advanced AI research, large language models (LLMs), and next-generation applications such as robotics, generative design, and autonomous systems.
In parallel with the supercomputing project, NVIDIA also announced the establishment of a new regional headquarters in Taiwan. The headquarters building—dubbed "NVIDIA Constellation"—was unveiled through rendered visuals showcasing a futuristic, spacecraft-inspired architectural design. Huang noted that construction on the new facility will begin soon, providing office and research space for NVIDIA employees in the region.
NVIDIA's Expanding AI Supercomputing Ambitions
NVIDIA, already widely recognized as the global leader in AI hardware and infrastructure, continues to expand aggressively in the AI supercomputing domain.
The company’s foray into AI supercomputers began in 2016 with the launch of the first-generation DGX system, a powerful AI workstation that was famously donated to OpenAI by Huang himself. That move is now viewed as a symbolic starting point for the modern AI revolution.
Since then, NVIDIA has continuously advanced its DGX series, with systems like DGX-1, DGX A100, and the latest DGX GB200 NVL72 and DGX Cloud, integrating NVIDIA GPUs and purpose-built AI software stacks. These platforms are capable of processing massive datasets and training cutting-edge AI models at scale.
From Datacenter to Desktop: Personal AI Supercomputing
In March 2025, NVIDIA revealed its ambition to democratize AI development with the introduction of “personal AI supercomputers.” The concept is aimed at enabling researchers, developers, and scientists to train and run powerful AI models directly on desktop machines.
During his Computex keynote, Huang provided an update on this initiative, confirming that the DGX Spark, one of the flagship personal AI systems, has entered full production and will hit the market in the coming weeks.
Another product in this new category, the DGX Station, was also highlighted. Designed for office or even home use, it plugs into a standard wall outlet and is powerful enough to run AI models with over 1 trillion parameters—a milestone that underscores the growing accessibility of high-performance AI computing.
Broader Global Expansion and Continued Investment
The Taiwan supercomputer announcement follows several other key developments in NVIDIA’s global expansion strategy:
-
In April 2025, NVIDIA revealed plans to build fully domestically manufactured AI supercomputers in the United States, working closely with a network of local suppliers. This initiative marks a shift toward onshore production amid growing geopolitical sensitivities around critical technologies.
-
Just days later, Huang visited Beijing and Shanghai, reiterating the importance of the Chinese market to NVIDIA’s global business. In its most recent fiscal year (ending January 26, 2025), NVIDIA generated $17 billion in revenue from China, accounting for 13% of the company’s total sales, despite ongoing export controls and regulatory challenges.
A Race for AI Leadership
As AI continues to transform industries—from healthcare and finance to manufacturing and entertainment—NVIDIA’s strategic investments reflect a broader race to dominate AI infrastructure globally. The planned supercomputer in Taiwan further cements the island’s role as a cornerstone in the global AI supply chain, and reaffirms NVIDIA’s intention to remain at the forefront of technological innovation.
Related reading:
Xiaomi to Launch In-House 3nm Flagship Chip "Surge Xuanjie O1"
Iron-Based Flow Battery Demonstration Project Launched in Hubei, China
Ceramic Data Storage: A Solution for Millennia-Long Data Preservation
NVIDIA Is Reportedly Moving Its Global Headquarters To Taiwan And Will Announce Next Week