TAIPEI, Taiwan — Nvidia CEO Jensen Huang used his latest Jensen Huang Taiwan visit to press key partners to expand capacity as demand for artificial intelligence hardware tightens supplies of chips and the high-end memory that feeds them, Feb. 1, 2026.
After hosting what local media dubbed a “trillion-dollar dinner” with senior executives from Taiwan’s tech ecosystem, Huang told reporters that the bottlenecks are no longer theoretical. He wants more wafers, more servers and, increasingly, more memory — fast.
Jensen Huang Taiwan visit spotlights a strained AI supply chain
Speaking outside a Taipei restaurant after the dinner, Huang joked that Taiwan Semiconductor Manufacturing Co. needs to “work very hard” because Nvidia needs “a lot of wafers,” underscoring how central the world’s biggest contract chipmaker is to Nvidia’s AI roadmap. Reuters described the scene — rain, cameras and fans crowding in — in its account of the late-night, impromptu press scrum outside the restaurant.
The Jensen Huang Taiwan visit also put memory on center stage. Huang said, “We need a lot of memory this year,” calling the broader supply chain “challenging” as AI demand accelerates. For Taiwan’s suppliers, that message landed alongside fresh investment expectations: TSMC has said capital spending could climb to $56 billion this year as AI demand grows.
The squeeze is showing up beyond data centers. A separate Reuters report noted that Apple has warned that rising memory chip prices are starting to pressure profitability, as major South Korean manufacturers steer more production toward high-bandwidth memory for AI servers, tightening supplies of conventional DRAM used in phones and PCs. That dynamic helps explain why the Jensen Huang Taiwan visit focused as much on memory as on compute: the broader memory crunch is spilling into consumer electronics.
OpenAI investment talk adds new urgency
Huang’s Taiwan swing was not only about hardware. In comments a day earlier, he denied reports that he was unhappy with OpenAI and said Nvidia plans a “huge” investment in the ChatGPT maker — likely Nvidia’s largest — while stressing it would not top $100 billion. He said Nvidia would be involved as CEO Sam Altman closes a funding round, with Reuters reporting that OpenAI is seeking up to $100 billion and a valuation around $830 billion. Here is Reuters’ full account of Huang’s OpenAI remarks in Taipei.
That financing backdrop matters for factories and fabs: more capital for OpenAI and its peers typically translates into more orders for GPUs, networking and — the theme Huang kept circling back to during the Jensen Huang Taiwan visit — memory-heavy systems.
The spectacle around Huang’s return trips has also become part of the story. Reuters chronicled “Jensanity” during his 2024 Computex appearance, when fans and media tracked his every move, and again in 2025 as he drew an AI-hero’s welcome, with the “trillion-dollar dinner” becoming a recurring symbol of Taiwan’s role in the boom: a familiar frenzy greeted him the following year.
In the end, the latest Jensen Huang Taiwan visit delivered a simple directive to the companies that build Nvidia’s AI machines: scale up — and do it fast — because the next constraint may not be the GPU, but the parts that surround it.
