News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud

National AI hub, Japan

Kobe (Fugaku)

RIKEN's computing centre on Port Island, home of the K computer and Fugaku, the world's fastest machine from 2020 to 2022.

Kobe's Port Island holds the RIKEN Center for Computational Science, Japan's flagship supercomputing site. The K computer, the world's fastest in 2011, ran here until 2019. Its successor Fugaku, built by Fujitsu, debuted in 2020 and led the TOP500 list from June 2020 to June 2022. Japan used it to train Japanese language models, an unusual use for a CPU machine.

Fugaku

Fugaku uses no GPUs. It is built from about 160,000 Fujitsu A64FX chips, Arm processors with wide vector units designed in Japan, which also handle the matrix maths of AI. In 2024 a group of Japanese institutions released Fugaku-LLM, a language model trained on the machine. It is used for drug discovery, weather, earthquakes and, during the pandemic, modelling how droplets spread.

What comes next

RIKEN chose Fujitsu to design FugakuNEXT, the successor, and in 2025 said Nvidia GPUs would be added alongside Fujitsu's new MONAKA-X processors. It is planned for around 2030 at the Kobe site. Japan treats the line as a national asset, much as Europe treats EuroHPC machines, with access granted to researchers rather than sold.

Providers of Kobe (Fugaku)

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.