Fujitsu, Nvidia set to build energy-efficient AI chip for Japan by 2030

The new chip will combine Nvidia's graphics processing units with Fujitsu's central processing units on electronic boards and servers

Japanese firm Fujitsu and chip giant Nvidia plan to jointly develop an energy-efficient artificial intelligence (AI) chip by 2030, expanding the partnership to support Japan’s growing data centre, robotics and technology sectors, Nikkei Asia reported. 

The new chip will combine Nvidia’s graphics processing units (GPUs) with Fujitsu’s central processing units (CPUs) on electronic boards and servers. It will use Nvidia’s high-speed technology to connect multiple chips so they function like a single processor. The companies have not disclosed how much they will invest, the news report said. 

“We are creating Japan's AI infrastructure, from silicon to systems to AI models and software, designed with Japan for Japan's industries and society,” Nvidia CEO Jensen Huang told reporters in Tokyo on Friday. He added that Fujitsu’s CPUs will be “fused with” Nvidia’s technology to deliver “a new class of energy-efficient and high-performance computing systems”. 

 

Central to this collaboration is Japan’s flagship Fugaku supercomputer, built by Fujitsu and the Riken research institute. In August, Riken, Fujitsu, and Nvidia revealed plans for FugakuNEXT, expected to launch around 2030, with five to ten times the computing power of its predecessor. 

Expanding AI chips beyond national projects

Until now, Fujitsu’s chips have mainly been used for national projects such as Fugaku and are based on British chipmaker Arm’s architecture. The tie-up with Nvidia aims to expand its use into commercial AI applications while leveraging energy-saving technology developed for the supercomputer. 

Nikkei Asia quoted Fujitsu CEO Takahito Tokita telling journalists “robust AI infrastructure with sufficient processing power and capabilities is essential” for the widespread adoption of AI across business and society. He said the new chip will be part of the company’s “full-stack” AI infrastructure, designed to support autonomous AI agents for businesses.

Japan’s bid in the global AI race

The partnership also tests whether Japan can establish a presence in the global AI race, which is currently led by US and Chinese companies. Fujitsu is promoting the idea of “sovereign AI”, where countries develop their own AI systems, from computing infrastructure to data governance. 

The collaboration also targets “physical AI” or AI for autonomous machines that perceive, interpret, and act in the real world. Fujitsu, Nvidia, and Japanese industrial robot maker Yaskawa Electric have started talks to use the AI chips and software in Yaskawa’s machines, the news report mentioned. 

Nvidia to invest $100 bn in OpenAI

Last month, Nvidia announced plans to invest up to $100 billion in OpenAI, adding at least 10 gigawatts of Nvidia AI data centres to boost computing power for ChatGPT’s developer, news agency Reuters reported. 

Under a signed letter of intent, the first gigawatt of Nvidia systems will be deployed in the second half of 2026. Both companies said they will finalise the details in the coming weeks.

  This partnership builds on the work OpenAI and Nvidia are already doing with collaborators like Microsoft, Oracle, SoftBank, and Stargate Partners, aiming to develop the world’s most advanced AI infrastructure.

  The announcement came just days after OpenAI said it reached a tentative agreement giving Microsoft a $100 billion equity stake in its for-profit corporation, though OpenAI remains technically controlled by its nonprofit entity.