Jensen Huang, CEO of Nvidia, continued his closely watched Taiwan visit with a high-profile gathering of semiconductor and electronics leaders on May 17, following a private dinner the night before with TSMC Chairman C.C. Wei and senior executives. Attendees included some of the most influential names in the region's supply chain: MediaTek CEO Rick Tsai, Quanta Chairman Barry Lam, Wistron Chairman Simon Lin, and Asus Chairman Jonney Shih.
While framed as an informal dinner, the guest list sent unmistakable signals. Most notably, Compal Electronics Chairman Ray Chen made his first appearance at such a gathering, underscoring the company's deeper push into the AI server business. Compal recently announced plans to build a manufacturing facility in Texas, aligning with U.S. reshoring policies and Nvidia's strategic manufacturing goals. In contrast, Foxconn Chairman Young Liu—despite holding sizable AI server orders—was absent, reportedly sending representatives in his place.
Beyond the usual names seen with Jensen Huang in Taiwan, Compal Chairman Ray Chen (standing directly behind Huang) appears for the first time.
Originally slated to arrive in Taiwan on May 12, Huang's trip was delayed until May 16 due to his participation in U.S. President Donald Trump's Middle East tour, where major AI infrastructure partnerships were announced. Upon landing in Taipei, Huang immediately met with TSMC leadership, an encounter unusually well-publicized by the typically discreet chipmaker—perhaps a nod to the geopolitical significance of Nvidia's Taiwan alignment.
Supply chain sources noted that such meetings are rarely documented so openly, indicating Nvidia's desire to make a statement. The Mandarin Oriental Hotel, where Huang is staying, hosted several of his meetings, with others taking place at TSMC's Hsinchu headquarters.
The May 17 dinner marks Huang's third major convening of Taiwan's tech powerhouses this year, following similar meetings around Nvidia's GTC events in January and March. Compared to January's roster, MediaTek's Rick Tsai attended in person this time, reinforcing Nvidia's expanding partnership with the fabless giant on mobile AI and edge computing.
Compal's inclusion stood out. After years on the sidelines of Nvidia's tight-knit AI supply chain ecosystem, the PC maker has invested heavily in AI server infrastructure and was finally invited into the fold. Acer, another latecomer, also participated, signaling that Nvidia's ecosystem is widening—albeit selectively.
Also present were longtime Nvidia partners in the GPU and gaming space: Asus, MSI, Gigabyte, and ASRock. Their continued presence reflects sustained demand for AI and high-performance graphics, even as the PC market fluctuates.
Industry watchers saw the unusually public nature of the dinner as symbolic: being seen with Huang now amounts to official recognition as a key node in Nvidia's AI server supply chain. It's also a reminder that the bar for entry is rising. As Nvidia's partner network deepens, the scramble for design wins, pricing power, and capacity commitments is intensifying. Amid growing U.S. scrutiny of Chinese tech exports and shifting tariff regimes, companies aligning with Nvidia must also be ready to manufacture where geopolitics allows.
Huang has used his Taiwan stop to address persistent rumors about Nvidia's H20 chips and the possibility of downgraded AI accelerators for the Chinese market. He confirmed that no new Hopper-architecture chips will be introduced, citing U.S. restrictions and design finality. While some export restrictions to the Middle East have been lifted, Nvidia is still evaluating how to respond to demand in China under existing rules.
Looking ahead, Huang is scheduled to speak at the Taipei Music Center on May 19, where he will likely reprise themes from his March GTC keynote: the rise of "agent AI" capable of interactive reasoning, and the next leap toward "embodied AI" for robotics. He has projected global data center capital spending will reach US$1 trillion by 2030, with enterprises shifting rapidly from traditional software to generative AI, fueling demand for GPUs.
TSMC Chair CC Wei. Credit:Digitimes
He is also expected to expand on Nvidia's four-year roadmap (2025–2028), including its open-source operating system for AI factories, Dynamo; silicon photonics interconnects; and the humanoid robot platform GROOT N1, aimed at making general-purpose robotics accessible to industrial developers.
According to industry sources, Huang may also use the Taiwan event to announce the site of Nvidia's future Taiwan headquarters—a closely watched move given the company's growing presence in the region.
During COMPUTEX, Huang is slated to appear at MediaTek's keynote session and visit booths of key manufacturing partners including Wiwynn and Asus. He will also participate in a global media roundtable on May 21.
Meanwhile, Intel's newly appointed CEO Lip-Bu Tan has also arrived in Taiwan and will host a separate supplier dinner on May 19, setting the stage for further strategic positioning among U.S. chip leaders on the island.
Article edited by Joseph Chen
