Renhui Huang: Semiconductor Industry Scale May Need to Expand 10-fold in the Next 10 Years

BlockBeats News, July 26th - NVIDIA CEO Jensen Huang said in an interview with Bloomberg News that South Korea is in a "golden age," with its semiconductor and industrial capabilities able to help build global AI infrastructure. NVIDIA has reached multiple agreements with the SK Group, and the scale of their future business dealings in areas such as storage procurement, AI supercomputers, and data center construction will exceed $500 billion. NVIDIA will purchase a large number of storage chips from SK Hynix for several years and collaborate on the technical roadmap from HBM2, HBM3, HBM3E, HBM4, HBM4E to future products. Meanwhile, SK Telecom plans to expand its AI cloud infrastructure to a maximum of 2 gigawatts soon, and NVIDIA will sell supercomputers to them. The collaboration between the two parties includes not only NVIDIA's purchase of storage but also the SK Group's purchase of AI infrastructure.

Jensen Huang believes that as AI intelligent agents and robots begin to use computers on a large scale, chip demand will no longer be solely driven by human users. In the next roughly 10 years, the global semiconductor industry's scale may need to expand to ten times its current size. Current resources such as HBM storage, land, power, and data center construction are all in tight supply, and the industry may be capable of doubling its capacity annually. However, faster expansion will be very difficult. The entire industry may be able to double its supply scale annually, but land, power, and factory space cannot expand as rapidly as consumer electronics, so the construction of AI infrastructure may continue to advance at a constrained pace over the next 10 years.

Regarding the AI competition between China and the United States, Jensen Huang said that both countries have excellent AI researchers. Although the resources, conditions, and restrictions are different, both will continue to drive AI technological progress. The number of AI researchers trained in China each year may exceed the total for the rest of the world, while the United States still needs to continue learning, cooperating, and maintaining competitiveness.