NVIDIA's Jensen Huang refers to Taiwan as the 'epicenter' of the AI revolution, noting that spending has reached $150 billion annually.

      NVIDIA's CEO Jensen Huang addressed an audience in Taipei during Computex 2026 on Wednesday, stating that Taiwan is the “epicenter” of the AI revolution, with the company's annual expenditure on the island projected to be around $150 billion. This figure, which Huang has revealed as the highest specific spending amount related to Taiwan, surpasses the GDP of most EU member countries.

      Huang's keynote largely detailed the supply chain behind this figure, with the Vera Rubin platform being a prime example. He described this next-generation AI platform as “probably the largest product launch in the history of Taiwan.” Each Vera Rubin system comprises nearly 2 million components, constructed in collaboration with 150 ecosystem partners, predominantly within Taiwanese enterprises. TSMC is responsible for the foundational logic, while assembly is managed by firms like Foxconn, Quanta, and Wistron. Additionally, SK Hynix, although based in Seoul, has a significant presence in Taiwan and provides the HBM4 memory required for the platform to achieve 22 TB/s bandwidth.

      The implications of Huang's statements are significant politically as well as technically. The Trump administration's second-term tariff policies have exerted pressure on NVIDIA and other U.S. chip manufacturers to relocate more of their production capabilities to the United States. By disclosing the Taiwan spending, Huang signals to Washington the substantial costs associated with that transition while reinforcing to Taipei NVIDIA’s steadfast commitment to the island.

      His recent comments about the potential use of Huawei chips for DeepSeek, which he deemed a “horrible outcome” for the U.S., indicate NVIDIA’s clear alignment with the U.S. in the ongoing tech cold war.

      Notably, Huang did not disclose any specifics regarding new fabrication plants, packaging facilities, or supply assurances. The $150 billion represents an aggregate figure flowing through the existing Taiwanese supply chain rather than a distinct capital investment plan. The ramp-up for Rubin has already been putting pressure on that ecosystem, with TSMC reportedly working overtime to fulfill NVIDIA’s Rubin orders, and the $150 billion-a-year projection further intensifies that pressure.

      For Taiwan, the political implications are significant as the Lai Ching-te administration has been advocating in Washington to highlight the island's critical role in the U.S. AI development as a strategic advantage rather than a burden. Huang’s “epicenter” reference bolsters that narrative.

      In contrast, for Beijing, which has been tightening its AI policies in 2026, Huang's keynote serves as additional evidence of the geographical concentration of the global AI supply chain rather than its dispersion.

      Computex 2026, with the theme “AI Together,” continues throughout the week. Huang’s keynote was the main highlight, with AMD's Lisa Su also scheduled to present later, both likely to outline their respective strategies ahead of 2027, when NVIDIA’s Rubin Ultra and AMD’s MI400-series are set to be released in significant volumes.