Nvidia Boosts AI Ecosystem with NVLink Fusion and Reveals Future Chip Roadmap at Computex
TAIPEI—Nvidia announced on Monday, May 19th, its strategic plan to sell a new technology designed to dramatically accelerate the communication speed between chips, a critical factor in the development and deployment of sophisticated artificial intelligence tools. The unveiling of NVLink Fusion took center stage at an event preceding the Computex AI exhibition in Taipei, where CEO Jensen Huang also disclosed plans for a new Taiwan headquarters and offered insights into the company’s evolving AI chip strategy.
NVLink Fusion: Powering Collaborative AI Systems
The newly launched version of Nvidia’s proprietary interconnect technology, dubbed NVLink Fusion, will be made available to other chip designers. This move aims to facilitate the creation of powerful, custom AI systems by enabling multiple chips to work together seamlessly and exchange vast amounts of data with unprecedented speed. Nvidia itself utilizes NVLink to link various chips in its own formidable hardware, such as the GB200, which integrates two Blackwell graphics processing units (GPUs) with a Grace central processing unit (CPU).
Several prominent technology firms are already poised to adopt NVLink Fusion for their custom chip endeavors. Nvidia confirmed that Marvell Technology and MediaTek are among the early adopters. The growing list of partners also includes Alchip, Fujitsu, and Qualcomm, signaling broad industry interest in leveraging this advanced interconnect solution to push the boundaries of AI system performance.
Expanding Roots: Taiwan Headquarters and Evolving Vision
During his keynote speech at the Taipei Music Center, the venue for the Computex AI exhibition running from May 20th to 23rd, Jensen Huang announced Nvidia’s intention to construct a new company headquarters in the northern suburbs of Taipei. This decision underscores Taiwan’s critical role in the global semiconductor ecosystem and Nvidia’s commitment to the region.
Huang’s address also traced Nvidia’s remarkable journey from its origins as a maker of graphics chips for video games to its current position as the dominant force in producing the high-performance chips that have fueled the artificial intelligence frenzy since the advent of tools like ChatGPT in 2022. He acknowledged that his past presentations often centered on the company’s graphics chips, but now the narrative has expanded significantly. This evolution includes previously reported efforts by Nvidia to design CPUs compatible with Microsoft’s Windows operating system, utilizing technology from Arm Holdings.
Roadmap to the Future: Next-Generation AI Chips Unveiled
Looking ahead, Nvidia is not resting on its laurels. Huang reiterated the company’s strategic shift, initially outlined during Nvidia’s annual developer conference in March, from primarily focusing on building large AI models to enabling the widespread deployment of applications based on them.
As part of this forward-looking strategy, several new generations of AI chips were highlighted. The Blackwell Ultra chip is slated for release later this year, promising even greater performance. Following this, Nvidia plans to introduce its Rubin series of chips, which will eventually be succeeded by the Feynman processors, currently anticipated to arrive in 2028.
Nvidia also launched a desktop version of its AI chips, named DGX Spark, specifically targeting AI researchers. On Monday, Huang confirmed that this powerful desktop computer is now in full production and would be ready for market in just “a few weeks,” offering researchers direct access to Nvidia’s AI processing capabilities.
Computex Buzz and the “Jensanity” Phenomenon
The Computex event itself, expected to host around 1,400 exhibitors, is a significant gathering for computer and chip executives in Asia. It also marks the first major industry congregation since the U.S. President Donald Trump raised the prospect of tariffs aimed at encouraging companies to increase production within the United States.
Last year, Huang’s presence at Computex sparked a phenomenon dubbed “Jensanity” in Taiwan, as he was enthusiastically followed by the public and media and mobbed by attendees at the trade show. His return and the latest series of announcements are set to generate similar excitement, reflecting Nvidia’s influential role in shaping the future of technology. Through these strategic initiatives in hardware innovation and ecosystem collaboration, Nvidia continues to solidify its leadership in the rapidly advancing field of artificial intelligence.
