Nvidia is set to sell its NVLink Fusion technology to other chip designers, aiming to accelerate chip-to-chip communication for building and deploying AI tools. The announcement was made by…
Nvidia is set to sell its NVLink Fusion technology to other chip designers, aiming to accelerate chip-to-chip communication for building and deploying AI tools. The announcement was made by CEO Jensen Huang at the Computex AI exhibition in Taipei. This new version of NVLink is designed to facilitate the creation of powerful, custom AI systems by linking multiple chips together.
Key partners, including Marvell Technology, MediaTek, Alchip, Fujitsu, and Qualcomm, plan to adopt NVLink Fusion for their custom chip endeavors. The move comes as Nvidia navigates challenges in the Chinese market, where U.S. restrictions on chip shipments have resulted in significant lost sales.
Huang revealed that Nvidia had to forego $15 billion in sales in China due to these limitations. Despite these setbacks, Nvidia continues to innovate, with the launch of NVLink Fusion and plans for a new Taiwan headquarters demonstrating its commitment to the AI sector. During his keynote, Huang highlighted Nvidia's evolution from a graphics chip maker to a dominant force in AI, emphasizing its role in powering the AI boom since the launch of ChatGPT.
He also discussed the company's strategy to address the shift in computing needs, from building large AI models to running applications based on them. The company unveiled new generations of AI chips, including the Blackwell Ultra and the upcoming Feynman processors. In addition to the NVLink Fusion announcement, Nvidia also launched a desktop version of its AI chips, the DGX Spark, targeted at AI researchers, which will be ready in a few weeks.
The Computex exhibition, with 1,400 exhibitors, marks a significant gathering of computer and chip executives in Asia, particularly given the backdrop of potential trade tensions.