Jensen Huang Arrives in Taiwan: NVIDIA Most Powerful "Nuke" New Product Set for Mass Production

On the afternoon of May 23, NVIDIA CEO Jensen Huang arrived early at Taipei Songshan Airport via private jet, a full four days ahead of the originally expected date of May 27.

His trip will not only kick off GTC Taipei and COMPUTEX in June but also see him personally coordinate NVIDIA's largest-ever product introduction plan, marking the official entry of the global AI supply chain into the Vera Rubin era.

Huang said in an interview that he has many things to do on this trip, including visiting customers, partners, and employees, as well as holding company meetings.

Speaking at the airport, Huang clearly stated that the next-generation AI superchip platform, codenamed Vera Rubin, will be the most successful product generation in the history of NVIDIA and the entire computer industry.

Unlike the Hopper era, when only one or two cutting-edge AI companies collaborated with NVIDIA, all major cloud service providers, computer manufacturers, and AI model companies worldwide have now joined the Vera Rubin ecosystem.

As a blockbuster next-generation product, Vera Rubin is NVIDIA's new AI superchip platform, named after astronomer Vera Rubin, a pioneer in dark matter research. The platform consists of the Rubin GPU and Vera CPU, manufactured using TSMC's third-generation 3nm process and CoWoS-L packaging technology, and is the first to support 8-layer HBM4 high-bandwidth memory.

In October 2025, Huang first showcased the chip at GTC. In March 2026, the Vera Rubin AI platform was officially unveiled at GTC 2026, with mass production planned for Q3 or Q4 2026.

Huang previously stated during an earnings call that Vera Rubin is expected to begin production and shipment in the second half of this year, delivering 35 times the inference throughput of the current Blackwell architecture.

NVIDIA stated that Vera Rubin is already generating strong pre-launch demand and is expected to face supply constraints throughout its entire lifecycle.

Remarkably, a single Vera Rubin system contains nearly 2 million components and requires approximately 150 ecosystem partners in Taiwan to build, covering the entire supply chain including wafer foundry, advanced packaging, servers, cooling, and power systems.

Huang emphasized that NVIDIA typically plans 2-3 years ahead. The company's revenue is currently growing at nearly 100% year-over-year and is expected to maintain this momentum next year. Taiwan's supply chain will face an unprecedented period of activity in the second half of the year.