YOU ARE AT:AI Infrastructure

BROWSING: AI Infrastructure

LG explores Saudi deal on AI data center cooling

Talks between LG and Saudi authorities center on expanding cooperation in thermal management for what is expected to become the Middle East’s largest net-zero AI data center In sum – what to know: Talks on Neom cooling tech – LG and Saudi officials explored cooperation on...

Huawei to double output of Ascend AI chips

Huawei unveiled its roadmap for the next generation of Ascend AI chips during the recent Huawei Connect event in Shanghai, China In sum – what to know: Doubled chip output – Huawei's plan calls for 600,000 Ascend 910C units next year and up to 1.6 million...

Deloitte opens global AI infra center of excellence

Deloitte said the CoE is part of the firm’s Silicon2Service AI factory-as-a-service model In sum – what to know: Global AI Infrastructure CoE launched – Deloitte has established a hub to help clients design, deploy, and operate specialized AI-ready data centers worldwide. Performance and resilience – Capabilities...

Nokia, Nscale partner to accelerate AI infra build-out

Under the terms of the deal, Nokia has become Nscale’s preferred provider of networking technologies for large-scale AI clusters In sum – what to know: Nokia named Nscale’s networking partner – Agreement covers data center switching, IP routing, and optical networking for AI-ready clusters across Nscale’s...

US mulls 1:1 chip production rule: Report

The measure aims to reduce dependence on overseas chip supply and bolster economic security In sum – what to know: Proposed 1:1 production rule – Chipmakers would need to match imports with U.S. production or face tariffs, with exemptions for pledged domestic capacity under construction. Policy...

Telus launches Canada’s first sovereign AI factory

The new AI factory from Telus is supported by Nvidia GPUs and infrastructure designed by HPE In sum – what to know: Canada’s first sovereign AI factory – TELUS has launched a secure, Canadian-operated facility to provide domestic compute capacity for AI workloads while maintaining data...

SK Telecom launches independent AI unit

SK Telecom has committed $3.6B to AI over the next five years In sum – what to know: Independent AI CIC launched – SK Telecom creates a new in-house company unit to consolidate AI assets and streamline its transformation strategy. KRW 5T investment pledged – The company...

CoreWeave expands OpenAI partnership with up to $6.5B deal

CoreWeave noted that the deal raises the total contract value between the two companies to about $22.4B In sum – what to know: $6.5B new agreement – CoreWeave’s latest deal raises the total value of its contracts with OpenAI to $22.4 billion, reflecting growing demand for...

OpenAI, Oracle, and SoftBank expand Stargate

The five new locations were selected from more than 300 proposals across 30 states, according to OpenAI In sum – what to know: New Stargate sites – OpenAI, Oracle, and SoftBank confirmed U.S. data centers in Texas, New Mexico, Ohio, and the Midwest, adding significant AI...

Samsung secures Nvidia approval for HBM3E chips

Samsung recently passed Nvidia’s qualification tests for its 12-layer HBM3E chips In sum – what to know: Test passed – Samsung secured qualification for Nvidia's 12-layer HBM3E, overcoming past performance and thermal challenges. Validation restores credibility – While already supplying AMD, recognition from Nvidia is seen as...

Alibaba Cloud announces international expansion

Alibaba Cloud is announcing its first data centers in Brazil, France, and the Netherlands, with additional facilities planned In sum – what to know: Global expansion – New data centers in Brazil, France, the Netherlands, and additional planned sites extend its infrastructure beyond 90 existing availability...

Meta files application to sell wholesale power

Meta made the request to the Federal Energy Regulatory Commission through its newly formed subsidiary Atem Energy In sum – what to know: Meta enters wholesale power market – Atem Energy has applied to FERC to sell energy and capacity, aiming to secure supply and flexibility...

Ciena to acquire Nubis in $270 million deal

The deal will expand Ciena's data center portfolio and add more than 50 engineers to its R&D team In sum – what to know: Ciena acquires Nubis for $270 million – The deal adds advanced optical and copper interconnect technologies designed to handle AI-related traffic growth...

AI Pathfinder commits $24.9 billion to UK data centers

Newcomer AI Patfinder said it will allocate £3.4 billion to build a data center in Northamptonshire, in the U.K, with the first phase expected online by December 2025 In sum – what to know: $24.9 billion investment – AI Pathfinder will build two large AI data...

Malaysia pilots AI traffic intelligence system

The project is being developed with MyDigital Corporation, Malaysia Digital Economy Corporation, Digital Nasional Berhad, and the Petaling Jaya City Council In sum – what to know: Malaysia pilots AI traffic platform – CelcomDigi and government partners launch the country’s first large-scale AI-powered traffic intelligence system...

Alibaba, Baidu start using own developed chips: Report

Alibaba has been deploying its custom chips for smaller AI models since early 2025 In sum – what to know: Alibaba and Baidu deploy in-house chips – Both Chinese companies are training AI models with custom silicon, partly reducing dependence on Nvidia. U.S. export curbs accelerate shift...

SoftBank Vision Fund weighs 20% cuts amid AI pivot

The cuts reflect the diminishing role of the Vision Fund as SoftBank allocates resources into AI projects In sum – what to know: Vision Fund plans 20% cuts – More than 50 jobs may be eliminated as SoftBank streamlines operations under CEO Masayoshi Son’s AI-first strategy. AI...

Huawei outlines roadmap for Ascend AI chips

Huawei has said the new chips will strengthen the foundation of AI computing power, both in China and around the world In sum – what to know: Huawei lays out three-year chip roadmap – Ascend 950 (2026), Ascend 960 (2027), and Ascend 970 (2028), with each...

LG, SK partner on AI DC cooling solutions

LG Electronics will supply air-cooling products such as chillers and fan wall units, as well as liquid cooling technologies including coolant distribution units (CDUs) In sum – what to know: First energy-focused LG–SK collaboration – The MoU marks their entry into the AI data center cooling...

Neocloud segment sees five-year CAGR of 82%

In the study, JLL identifies around 190 “distinct operators” beyond established neoclouds such as CoreWeave, Nebius, and Crusoe In sum – what to know: 82% CAGR since 2021 – Neoclouds are rapidly expanding revenue as AI GPU demand accelerates, with nearly 200 operators entering the market. Flexible...

Huawei showcases new AI DC solutions

Michael Ma, Huawei VP and president of ICT product portfolio management and solutions, said that data centers are "at a turning point to become AI DCs" In sum – what to know: Huawei pushes AI-ready data centers – The Chinese vendor highlights compute clusters, data lakes,...

Here’s what you need to future-proof your AI-ready data center (Reader Forum)

Preparing data centers for AI means rethinking capacity, efficiency, and cost in a rapidly growing market By 2034, the global AI data center market will reach approximately $157.3 billion. That’s a significant increase from the $14.3 billion in 2024. While hyper-increased demand for computing power is needed...

Microsoft inks $6.2 billion AI compute deal in Norway

Microsoft noted that the location was chosen for its abundant hydropower and low local electricity demand, because of its small population and transmission capacity In sum – what to know: Microsoft signs $6.2B AI deal – Five-year contract from 2026 will provide AI compute capacity in...

Nscale outlines major UK AI infra commitment with partners

Nscale highlighted that the core of the planned investment is its AI Campus in Loughton, which will host the U.K.’s largest AI supercomputer In sum – what to know: Largest UK AI supercomputer planned – Nscale’s Loughton campus will house 23,040 Nvidia GPUs from 2027, scaling...