Nvidia CEO to keynote Computex Taipei

Nvidia CEO Jensen Huang will return to Taiwan in June to attend Computex Taipei, where the company plans to make new announcements and deliver a keynote speech, according to his statements in Taipei.

Huang, visiting Taiwan for the Lunar New Year, met with local suppliers and praised Taiwan’s supply chain, citing companies like TSMC and Foxconn.

He mentioned he may attend a signing ceremony for Nvidia’s new Taiwan headquarters, scheduled between February 10-15.

Huang also acknowledged the competitive nature of the AI chip market, noting that Nvidia is active across various sectors, including cloud, enterprise data centers, robotics, and automotive.

🔗 Source: Focus Taiwan

🧠 Food for thought

Implications, context, and why it matters.

Taipei headquarters plan moves ahead during a tough negotiation

Nvidia plans a Taiwan headquarters called Nvidia Constellation at Beitou Shilin Technology Park in Taipei 1.
The plan follows long talks between Nvidia, Taipei City, and Shin Kong Life Insurance Co., since Shin Kong Life already held the site’s development rights 2.
Shin Kong Life said it should receive NT$10.7 billion to end the contract, based on 50 years of expected profits, according to Taipei Deputy Mayor Lee Shu-chuan 2.
In mid-October 2025, Taipei City and Shin Kong Life still disagreed on compensation, while Nvidia asked for a mutual termination so the land could return to the city and then be offered to Nvidia 2.

Nvidia reaches past chips into wider AI infrastructure

Nvidia laid out a GPU and interconnect roadmap through 2027, which suggests a fast platform cycle that can squeeze rivals and shape customer buying timelines 3.
Nvidia says GPU performance grew about 1,000x from Pascal P100 to Blackwell B100 over about eight years, helped by moves to lower numerical precision; this can cut the cost of training and running models, though hardware prices and power limits still matter 3.
Through NVLink Fusion, Nvidia is widening NVLink (a high-speed chip-to-chip interconnect) so partners can build semi-custom systems using NVLink with third-party CPUs (central processing units) and accelerators (specialized chips for AI workloads); Nvidia named Qualcomm and Fujitsu for CPUs plus Marvell among accelerator partners 4.
This could make Nvidia’s interconnect and full-stack platform (its combined hardware and software offerings) a common foundation for “AI factory” data-center builds, alongside its GPUs and software stack 4.