Artificial intelligence news.
Straight from the original announcements.

PFN and Tenstorrent Win NEDO Backing for Photonics-Based AI Processor R&D

October 21, 2026, 14:00By the AI News Bites Editorial TeamSource: ACN Newswire: Artificial Intelligence

Preferred Networks (PFN) and Tenstorrent USA have been selected by Japan's New Energy and Industrial Technology Development Organization (NEDO) for a joint project on xPU design using photonics-electronics integration technology. The companies will run a five-year research and development program and then work to commercialize the results.

Key points
  • PFN and Tenstorrent were selected by Japan's NEDO for an xPU project using photonics-electronics integration technology.
  • The R&D program will run for five years, followed by commercialization efforts for applications including AI agents.
  • PFN will develop AI chiplets with 3D memory stacking, while Tenstorrent will develop open-architecture CPU chiplets supporting confidential computing.
  • Five Japanese universities and research institutions join as subcontractors for scientific AI and physical AI applications.

What NEDO selected

According to the companies, their joint proposal was chosen for the theme "xPU Design and Development Incorporating Photonics-Electronics Integration Technology". The theme sits under a NEDO-sponsored program called the "Research and Development Project of the Enhanced Infrastructures for Post-5G Information and Communication Systems / Development of manufacturing technologies for advanced semiconductors". The selection notice was issued on Oct 21, 2026.

PFN and Tenstorrent say the aim is a system-level integration of AI processors that delivers high performance and high efficiency for both training and inference. They point to the growing size and computational complexity of AI models and workloads, which has increased the compute needed and raised demand for more scalable and efficient AI infrastructure. The project targets the next wave of AI deployment in industry and academia, including agentic AI, physical AI and scientific AI.

Division of work

The research program will run for five years. After that, the companies say they will work to commercialize the outcomes for applications including AI agents. Responsibilities are split as follows:

  • PFN and Tenstorrent together: development of system integration technologies for AI processors that use photonics-electronics integration and advanced packaging technologies.
  • PFN: development of AI chiplets with ultra-high memory bandwidth, enabled by 3D memory stacking based on leading-edge semiconductor process technologies.
  • Tenstorrent: development of CPU chiplets built on an open architecture that support confidential computing.

Subcontractors will also take part. The University of Tokyo, Kobe University, Toyo University, the Japan Agency for Marine-Earth Science and Technology and the Japan Atomic Energy Agency will research and develop domain-specific applications for scientific AI and physical AI. Tenstorrent Japan will design the CPU chiplets and develop software enabling confidential computing.

The companies involved

PFN was founded in 2014 in Tokyo and describes a vertically integrated approach covering AI chips, computing infrastructure, generative AI foundation models and industry solutions. It currently develops the MN-Core series of AI processors, the PFCP cloud platform for AI computing and the generative AI foundation model PLaMo.

Tenstorrent is an AI compute company led by CEO Jim Keller. It builds RISC-V-based AI processors and systems and licenses its TT-Ascalon RISC-V CPU and Tensix AI cores to chip designers including Samsung and LG. The company says it has raised over $1B+ and operates from Santa Clara, Austin, Toronto, Belgrade, Tokyo and Bangalore.

What it means

The selection gives the two companies public research backing for a multi-year effort on chip-level and system-level integration of AI processors. The announcement does not give budget figures, performance targets or product timelines, so it is unclear when any commercial products might follow. What the companies do state is the direction: combining photonics-electronics integration with advanced packaging, high-bandwidth stacked memory and open-architecture CPU chiplets, aimed at agentic, physical and scientific AI.

Related news

Quantinuum and RIKEN expand Japan quantum AI supercomputing partnership
Chips and infrastructureNovember 11, 2026, 16:55

Quantinuum and RIKEN expand Japan quantum AI supercomputing partnership

Quantinuum and RIKEN are expanding their partnership to advance quantum AI hybrid supercomputing in Japan after completing an upgrade of the Reimei quantum computer to…

Quantinuum and RIKEN expand Japan’s quantum AI supercomputing push
Chips and infrastructureNovember 11, 2026, 12:48

Quantinuum and RIKEN expand Japan’s quantum AI supercomputing push

Quantinuum and Japan’s RIKEN have expanded their partnership to advance hybrid quantum AI supercomputing, following the upgrade of Quantinuum’s Reimei system to System…

Quantinuum and RIKEN expand Japan’s quantum AI supercomputing effort
Chips and infrastructureNovember 10, 2026, 20:48

Quantinuum and RIKEN expand Japan’s quantum AI supercomputing effort

Quantinuum and RIKEN are expanding their collaboration to accelerate hybrid quantum AI supercomputing in Japan. The partnership includes an upgrade of RIKEN’s Reimei…