[News] Fujitsu and NVIDIA to Co-Develop AI Chip by 2030, Powering Supercomputers and Robotics

According to Nikkei, Fujitsu and NVIDIA plan to co-develop an energy-efficient AI chip by 2030, expanding their collaboration to serve Japan’s data center, robotics, and other industries. The companies did not disclose the scale of the investment, the report notes.

As the report further explains, the chip will integrate NVIDIA’s GPUs with Fujitsu’s CPUs on electronic boards and servers, using NVIDIA’s high-speed interconnect technology that enables multiple chips to function as a single processor.

The report highlights that the partnership also centers on Japan’s flagship Fugaku supercomputer, developed by Fujitsu and the Riken research institute. In August, Riken, Fujitsu, and NVIDIA announced FugakuNEXT, slated for around 2030, with performance expected to be five to ten times greater than its predecessor.

In addition, the report highlights that the collaboration will extend to “physical AI,” or AI for autonomous machines. Fujitsu, NVIDIA, and Japanese industrial robot maker Yaskawa Electric have begun talks on applying the two companies’ AI chips and software to power Yaskawa’s robots, the report adds.

So far, Fujitsu’s chips, built on Arm architecture, have mainly been used in national projects like Fugaku. As the report indicates, the tie-up with NVIDIA is aimed at expanding their use in commercial AI markets.

Meanwhile, the report indicates that the collaboration is seen as a test of Japan’s ability to carve out a role in the global AI race, currently led by U.S. and Chinese firms. It adds that Fujitsu is advancing the concept of “sovereign AI,” where nations build their own AI systems from infrastructure to data governance.

Read more

• [News] NVIDIA’s OpenAI Mega-Deal: AI Boom Proof or Customer Bailout Amid Antitrust and Conflict Concerns?
• [News] Mega NVIDIA-Intel Deal: No Chipmaking Orders, Co-Developed Products Might Launch 2027–28

(Photo credit: Fujitsu’s LinkedIn)