Fujitsu and NVIDIA Jointly Develop AI Semiconductor
On October 3rd, Japan’s Fujitsu announced that it would jointly develop a semiconductor with US-based NVIDIA for artificial intelligence (AI) applications, connecting their chips. Fujitsu’s semiconductors were previously used primarily in the national supercomputer “Fugaku.”
The collaboration will connect NVIDIA’s semiconductors at high speed to improve computing efficiency and significantly enhance energy efficiency, aiming to tap into new demand markets such as data centers and robotics.
NVIDIA is a global leader in graphics processing units (GPUs), which excel at AI computing. Fujitsu develops central processing units (CPUs), the “command center” of computing. The two companies plan to connect their semiconductors on a single substrate by 2030. Utilizing NVIDIA’s technology, multiple chips, including GPUs and CPUs, will be interconnected at ultra-high speeds, acting as a single chip.
At a press conference that day, NVIDIA CEO Jensen Huang emphasized that connecting with Fujitsu’s CPUs “will enable a new level of energy conservation and efficiency.”
Power Efficiency Improved by 2 Times
Fujitsu President Takahito Tokita stated, “This marks a significant step towards an AI-driven society.” The semiconductors to be jointly developed by the two companies will not only be used in AI data centers, but will also be expanded into “physical AI” fields such as robotics and automotive.
To advance the field of AI, Fujitsu is advancing the development of a new CPU. Based on the architecture of British semiconductor design company Arm, Fujitsu is developing the “MONAKA” CPU, featuring circuit lines as thin as 2 nanometers (a nanometer is one billionth of a meter).
Fujitsu aims to achieve power efficiency that is 2 times that of other companies’ CPUs and plans to put it into practical use in 2027. Fujitsu excels at integrating systems that include technologies from other companies to achieve overall energy savings.
After the launch of “MONAKA,” Fujitsu plans to update its models every two years. By 2029, it aims to release a CPU with a line width of 1.4 nanometers, offering even greater energy efficiency and performance. By partnering with Nvidia, which holds approximately a 70% share of the AI semiconductor market, Fujitsu will further expand its CPU sales channels.
While expanding its collaborations with other CPU manufacturers, Nvidia also places high hopes on Fujitsu’s energy-saving expertise, cultivated through supercomputer development.
Overseas technology companies are generally optimistic about the energy-saving technologies of Japanese companies.
On October 2nd, Hitachi, Ltd. and OpenAI in the United States announced a collaboration on power-related technologies for AI data centers. OpenAI will utilize Hitachi’s energy-saving technologies in power transmission and distribution equipment, air conditioning, and other areas to advance data center construction.
NTT is collaborating with Intel, Microsoft, and Google in the United States on the development of the next-generation optical communications infrastructure “IOWN (Innovative Optical and Wireless Network). IOWN aims to achieve high-speed, low-latency data processing and communication while reducing power waste.
They are also collaborating on supercomputers.
From the perspective of economic security, “sovereign AI,” which uses a country’s own data and infrastructure to develop and operate AI, is attracting attention. Nvidia also hopes to leverage its partnership with Fujitsu, which has extensive experience building systems in Japan, to further expand its presence in the Japanese market.
Furthermore, in August, Fujitsu and Nvidia announced a collaboration on the development of a successor to the RIKEN “Fugaku” supercomputer, scheduled for operation in 2030. The two companies will combine CPUs and GPUs to achieve powerful computing power more efficiently.
Nvidia is also interested in Fujitsu’s extensive domestic customer base. Fujitsu provides core systems to various industries, including manufacturing and finance, as well as systems to government agencies, including those in the security field. Nvidia hopes to further accelerate its global business expansion by leveraging this customer base, which demands the most from its products and systems.