Nvidia's Jensen Huang at Computex 2026: Vera Rubin Ships, a Plea for More Memory, and a TSMC Summit

Nvidia CEO Jensen Huang kicked off Computex 2026 in Taipei with a keynote that set the agenda for the week and the future of AI infrastructure. He confirmed that the company's next-generation Vera Rubin platform—a six-chip, extreme-codesigned architecture—is now in volume manufacturing, a milestone that shifts Nvidia's AI ambitions from product launch to global factory deployment . But the news from the keynote was quickly overshadowed by a spontaneous moment on the show floor that revealed the immense pressure on the AI supply chain: Huang's personal, handwritten plea to a critical memory partner—"Please Make More."

Vera Rubin Enters Full-Scale Production

During the keynote at the Taipei Music Center, which doubled as Nvidia's GTC Taipei developer conference, Huang didn't just preview Vera Rubin; he declared it is ramping into full production, with high-volume shipments expected to begin in the second half of the year . He characterized the platform as "the most ambitious endeavor in the history of the company," an assertion backed by the scale of the operation .

"The supply chain we created for Vera Rubin is twice as large as Grace Blackwell," Huang stated, emphasizing the industrial feat underpinning the launch . Key components of the platform are moving in lockstep. The Vera CPU, custom-built for the platform, was formally introduced . Networking is also scaling up: Nvidia confirmed that its Spectrum-X Ethernet Photonics switches, built on co-packaged optics (CPO), are now in full production to support the massive data-center fabric Vera Rubin AI factories will require .

In a strategic move beyond the data center, Huang also unveiled RTX Spark, an Arm-based superchip co-developed with Microsoft designed for Windows PCs. The product marks Nvidia's most direct entry into the consumer PC chip market, with Huang promising it will "reinvent" the personal computer around agentic AI assistants .

The 'Please Make More' Moment: Nvidia's Unquenchable Demand for Memory

The most headline-grabbing event of Computex didn't happen on stage. On June 2, Huang made an unannounced walk across the show floor to the SK hynix booth, where he met with SK Group Chairman Chey Tae-won for the second day in a row . At the booth, Huang picked up a marker, walked to a display of a seventh-generation HBM4E wafer, and wrote four words: "Please Make More" .

The scene—the world's most valuable chip designer publicly cajoling its top supplier to accelerate production—perfectly captured the insatiable appetite for high-bandwidth memory (HBM) that is gating the entire AI industry's growth. Huang's visit was as much theater as it was a strategic signal. During his keynote the previous day, he had explicitly named SK hynix as a key HBM4 partner for the Vera Rubin platform .

What happened next turned theater into a major strategic commitment. Hours later, Chairman Chey, who in March at Nvidia's GTC conference had told reporters that adding capacity wasn't planned, announced a dramatic reversal. SK hynix would double its total wafer production capacity over the next five years, a direct answer to the bottleneck Nvidia is racing to solve . The episode highlighted a stark reality: Samsung was also at Computex, unveiling the world's first physical mock-up of its more advanced HBM5 chip, but Nvidia's public commitment—and Huang's signature—remained firmly with SK hynix .

A Pre-Computex Summit with TSMC and an Unspecified Commitment

The tone for Computex was set a week before the show even began. Huang arrived in Taiwan early, on May 23, telling reporters at Songshan Airport only that he had "a lot to do" . Reports confirmed the trip's central purpose was a direct meeting with TSMC Chairman C.C. Wei to secure production commitments for the Vera Rubin ramp—a system Huang called "the largest product launch, probably in the history of Taiwan" .

Despite this intense focus on the Taiwanese supply chain, reports from the event do not cite a specific dollar figure or formal financial commitment from Nvidia to Taiwanese suppliers. The commitment is operational and strategic. The meeting with TSMC, the doubling of capacity from key supplier SK hynix, and the declaration of full-scale production collectively demonstrate a massive, ongoing investment, but a singular financial number was not part of any formal announcement at the show.