SoftBank's $676M AI Data Center Move, Google Eyes $3B Oklahoma Investment - Silicon Brief #9

Hi there,

In this edition of Silicon Brief, we're diving into the latest developments in AI chips and data centers. Nvidia is expanding its AI efforts beyond chips, while Meta is testing its in-house AI chip to reduce dependence on Nvidia. Meanwhile, Compal Electronics is set to unveil its AI-HPC servers at GTC 2025, and AMD has launched its new Instella-VL-1B Visual Language Model. These advancements highlight the growing focus on AI hardware innovation.

On the data center front, Kao Data has opened a new liquid-cooled AI data center in Harlow, and CoreWeave is collaborating with Bulk Infrastructure for a major NVIDIA AI deployment in Europe. Additionally, SoftBank has acquired a Sharp plant for AI collaboration with OpenAI, and Google is planning a $3 billion AI data center in Oklahoma. These initiatives underscore the increasing demand for robust infrastructure to support AI technologies.

FLUIX has secured $2.1 million in funding to advance its AI platform, A.I.M.I, which aims to optimize energy use in data centers by integrating HVAC, water, and IT systems. LS Eco Energy, via its subsidiary LS-VINA, has provided power cables for the STT Fairview Campus, the largest data center in the Philippines, under construction near Quezon City. Microsoft Corporation is focusing on Finland for its AI data centers due to its cold climate and carbon-neutral energy, aiming to become carbon-negative by 2030. Energy executives and regulators at the CERAWeek conference raised concerns about the potential strain on the U.S. electrical grid due to the rapid growth of AI data centers.

Vertiv has introduced the CoolLoop Trim Cooler, a new thermal management solution aimed at improving energy efficiency and space utilization in AI and high-performance computing environments. Huawei announced upgrades to its Xinghe Intelligent Network solutions at MWC Barcelona 2025, focusing on AI integration to boost network performance and security. Juniper Networks has launched a new networking solution for GPUaaS and AIaaS providers, enhancing AI delivery and reducing costs with advanced security features.

The European Union is taking steps to enhance its GPU reserves and strengthen its role in the chip industry following US export restrictions. EU leaders are committing funds to AI development and reducing regulations to promote technological growth.

Kao Data has launched KLON-03, a new 17.6MW liquid-cooled AI data center at its Harlow campus. The facility is designed to support AI workloads with 100% renewable energy and hybrid-cooling technology. CoreWeave and Bulk Infrastructure are collaborating to create a significant NVIDIA AI deployment at the N01 Datacenter Campus in Norway, set to be operational by Summer 2025. SoftBank has acquired the Sharp Sakai Plant in Osaka for $676 million to transform it into an AI data center, supporting its collaboration with OpenAI to advance AI model commercialization in Japan.

Nvidia CEO Jensen Huang is focusing on expanding the company's AI initiatives beyond chip manufacturing, addressing challenges like competition and geopolitical issues. Qualcomm and Palantir are collaborating to integrate Palantir's Ontology and AI capabilities with Qualcomm's edge computing platforms, aiming to enhance real-time insights and decision-making in industrial sectors. AMD has introduced Instella-VL-1B, a visual language model with 1.5 billion parameters, trained on AMD's MI300X GPUs, noted for its superior performance in benchmarks compared to similar open-source models.