PFN and Tenstorrent win NEDO project to develop optically integrated xPU through 2031
DavidBennett__ · x · 2026-10-20
Preferred Networks (PFN) and Tenstorrent, led by Jim Keller, have won a NEDO-funded project on "xPU design integrating optical-electronic fusion technology" under Japan's post-5G advanced semiconductor program, formally launching a joint R&D effort.
The project aims to build AI processor systems that run both training and inference with high performance and efficiency for generative AI, running through fiscal 2031 with commercialization for AI agents and beyond planned afterward. Division of work: PFN and Tenstorrent will develop processor systemization using optical-electronic fusion and advanced packaging; PFN will build ultra-high-memory-bandwidth AI chiplets via 3D memory integration; Tenstorrent will develop open-architecture, confidential-computing-capable CPU chiplets on advanced process nodes. The University of Tokyo, Kobe University, JAMSTEC and others will develop domain applications for AI for Science and Physical AI.
More from Infra
- Morgan Stanley: $700B may reach Nvidia by 2027, but US data centers face 32GW power gap — rohanpaul_ai · 2026-10-20
- Stop button often doesn't stop the LLM: tokens keep generating unless the app hangs up — No-Peanut-6988 · 2026-10-20
- ComfyUI hits 140k stars: text-to-image generation now runs on your own GPU — alex_verem · 2026-10-20
- Immich at 116k stars: face search across your photo library without Google — alex_verem · 2026-10-20
- whisper.cpp at 54k stars: hours of offline transcription on a normal laptop — alex_verem · 2026-10-20
- Five AI tasks that needed a data center in 2020 now run at home in 2026 — alex_verem · 2026-10-20