On-Prem

Systems

AI chip startup Tenstorrent to train Japan's engineers in $50M government deal

Government wants to make semiconductors big on home turf again


Japan has asked AI chip startup Tenstorrent to train up to 200 chip designers from the country at its US offices over the course of five years to help inject new blood into Japan’s semiconductor industry.

The contract, announced on Tuesday, is valued at $50 million, invested between both Tenstorrent and Japan's Semiconductor Technology Center (LSTC).

Tenstorrent said it will “train and elevate Japanese engineering talent while simultaneously cultivating a strong CPU team for the company in Japan.”

Program participants are to be nominated from large tech companies and universities in Japan, with the first cohort expected to start in April 2025. They will be trained on tech including Tenstorrent's RISC-V Ascalon design, Tensix IP, and AI and HPC software stacks for either one or two years before repatriating to their home country.

Both the engineers and Tenstorrent will get financial support from the Japanese government for the training program and the startup will get to keep any of the RISC-V chip blueprints made during the engineers' stay.

The goal is that the engineers will subsequently work for silicon design companies in Japan, and in return potentially license Tenstorrent’s technology.

“This is a groundbreaking program between Tenstorrent and Japan, and sending Japan’s top engineers to become experts in Tenstorrent’s technology will be pivotal in growing their ecosystem and propelling forward our efforts here as we build our edge 2nm AI accelerator in partnership with them,” extolled LSTC charman Tetsuro Higashi.

Tenstorrent chief customer officer David Bennett said the deal was in addition to an already existing agreement with Japanese advanced logic semiconductors company Rapidus and would help train “Japan’s best and brightest in multiple disciplines related to high performance compute, AI hardware design and advanced software.”

Tenstorrent has linked-up with Rapidus to co-develop semiconductor IP for AI edge devices, which will be manufactured in Rapidus's planned, government-subsidized factory, and confirmed it will open a design center in Tokyo.

Japan’s semiconductor manufacturing thrived in the 80s when it controlled over half the market worldwide, but in the 90s, it began facing increased competition from the likes of South Korea and Taiwan. By 2019, Japan held only 10 percent of the global semiconductor industry.

Now the nation would like to revert back to its heyday, and is offering substantial subsidies and programs, like the ones granted to Rapidus and Tenstorrent, to get there. ®

Send us news
5 Comments

Biden said to weigh global limits on AI exports in 11th-hour trade war blitz

China faces outright ban while others vie for Uncle Sam's favor

US adds Chinese RISC-V player that TSMC suspected of helping build Huawei GPUs to risky company register

Sophgo scores a place on Entity List, Indian nuclear boffins taken off

UK unveils plans to mainline AI into the veins of the nation

Government adopts all 50 venture capitalist recommendations but leaves datacenter energy puzzle unsolved

Where does Microsoft's NPU obsession leave Nvidia's AI PC ambitions?

While Microsoft pushes AI PC experiences, Nvidia is busy wooing developers

Microsoft eggheads say AI can never be made secure – after testing Redmond's own products

If you want a picture of the future, imagine your infosec team stamping on software forever

Just as your LLM once again goes off the rails, Cisco, Nvidia are at the door smiling

Some of you have apparently already botched chatbots or allowed ‘shadow AI’ to creep in

RISC-V is making moves, but it has work to do if it wants to hit the mainstream

Can it topple x86 and Arm, or is the gap too wide to close?

Microsoft, PC makers cut prices of Copilot+ gear in Europe, analyst stats confirm

Double-digit reduction only served to 'stimulate some interest'

Additional Microprocessors Decoded: Quick guide to what AMD is flinging out next for AI PCs, gamers, business

Plus: A peek at Nvidia's latest hype

In AI agent push, Microsoft re-orgs to create 'CoreAI – Platform and Tools' team

Nad lad says 30 years of change happening in 3 years ... we're certainly feeling the compression of time

AI datacenters putting zero emissions promises out of reach

Plus: Bit barns' demand for water, land, and power could breed 'growing opposition' from residents

Enterprises in for a shock when they realize power and cooling demands of AI

Energy consumption set to become a key performance indicator by 2027