ROHM & NVIDIA's 800V Breakthrough, Intel & Lenovo's $10M Supercomputer, EDGNEX's $2.3B Jakarta Data Center - Silicon Brief #23

Hi there,

In this edition of Silicon Brief, we're diving into the latest developments in AI chips and data centers. ROHM and NVIDIA are collaborating on an 800V HVDC architecture, which is expected to enhance power efficiency in AI applications. Meanwhile, Intel and Lenovo are teaming up to build a $10 million supercomputer at Imperial College, aiming to boost research capabilities with cutting-edge technology.

On the data center front, EDGNEX is making headlines with its $2.3 billion investment to build an AI data center in Jakarta, a move that underscores the growing demand for AI infrastructure in Southeast Asia. Additionally, SK Group and AWS are set to construct South Korea's largest AI data center, highlighting the region's commitment to advancing AI technology. Stay tuned as we explore these exciting developments and their implications for the tech industry.

Equinix Inc. is planning a data center in Minooka, Illinois, which is raising concerns over its potential impact on local water resources, as it is expected to use a significant portion of the municipality's drinking water supply. Meanwhile, Bloom Energy predicts that by 2030, 27% of data center facilities will rely entirely on on-site power generation, driven by increasing AI power demands and utility delays.

EDGNEX Data Centers by DAMAC is investing $2.3 billion to build an AI-powered data center in Jakarta, while Intel and Lenovo are collaborating on a $10 million supercomputer at Imperial College London. Gcore has launched an AI-native startup program, and Fujitsu Limited is designing the next-gen supercomputer for RIKEN. Hitachi Vantara has been recognized in the GigaOm Radar for AI storage, and TAC Data Centers plans a 600MW campus in Montana. Nvidia and Foxconn are planning humanoid robots in a Houston plant, and SK Group with AWS is set to build South Korea's largest AI data center.

ROHM Semiconductor is collaborating with NVIDIA on an 800V High Voltage Direct Current architecture to enhance data center efficiency. Groq has become an official inference provider on Hugging Face, supporting Alibaba's Qwen3 32B language model. Analysts from Oppenheimer highlight potential gains for Nvidia from sovereign AI demand. China introduces a hybrid AI chip to improve computing efficiency, while Muon Space raises $146 million for satellite infrastructure. Synopsys and Cadence both expand collaborations with Samsung Foundry for AI-driven design solutions. MiniMax unveils the MiniMax-M1 AI model with a 1 million token context, and Latent AI launches the Latent Agent platform for edge AI management.