Share
Linkedin
Twitter
Facebook
Whatsapp
Copy Link
Bain Capital's Bridge Data Centers terminated its contract with Megaspeed International at its Malaysian site following a US government investigation into alleged smuggling of restricted Nvidia AI accelerators to China. The data center operator reallocated 68.4 MW of power capacity to Zenlayer, distancing itself from a tenant suspected of spending roughly $2 billion on AI processors for illicit distribution.
Alibaba and China Telecom unveiled a data center in southern China powered by 10,000 Zhenwu AI chips developed domestically. The facility in Guangdong province supports AI models with hundreds of billions of parameters and marks a shift toward large-scale industrial implementation of homegrown AI infrastructure as U.S. semiconductor restrictions push Chinese firms to develop domestic chip alternatives.
Intel has signed on to help Elon Musk's SpaceX, Tesla, and xAI build the ambitious Terafab semiconductor facility in Texas. The chipmaker will leverage its expertise in designing, fabricating, and packaging high-performance chips to support the project's goal of producing 1 terawatt of computing power annually. But questions remain about the partnership's structure, as no formal SEC filings have been submitted yet.
Uber is expanding its AWS contract to run real-time ride-matching infrastructure on Amazon's Graviton4 processor and piloting AI model training on Trainium3. The move adds Uber to a growing roster of major tech companies choosing Amazon's custom silicon over Nvidia, while simultaneously challenging Google Cloud and Oracle in the cloud computing battle.
Nvidia's December acquisition of SchedMD has raised concerns among AI specialists and supercomputer experts about fair access to Slurm, open-source software that powers 60% of supercomputers worldwide. Engineers fear the chip giant may favor its own hardware over rivals like AMD, though Nvidia insists it will maintain vendor-neutral development.
Anthropic expanded its compute agreement with Google and Broadcom, securing 3.5 gigawatts of tensor processing unit capacity starting in 2027. The deal reflects explosive demand for Claude AI models, with the company's annualized revenue jumping from $9 billion to $30 billion in just three months. More than 1,000 business customers now spend over $1 million annually on Anthropic's services.
Samsung Electronics reported a preliminary operating profit of $37.9 billion for Q1 2026, an eightfold increase driven by explosive demand for memory chips used in AI computing and data centers. The result nearly triples the company's previous quarterly record and exceeds its entire 2025 profit, signaling an unprecedented supercycle in the semiconductor industry despite Middle East conflict concerns.
Broadcom has signed a multi-year agreement with Google to develop custom AI chips and components for next-generation AI racks through 2031. The chipmaker also expanded its deal with Anthropic, providing 3.5 gigawatts of AI computing capacity starting in 2027. The agreements cement Broadcom's position in AI infrastructure as demand for alternatives to Nvidia accelerates.
Australian AI data center company Firmus has raised $505 million led by Coatue at a $5.5 billion valuation, bringing total capital raised to $1.35 billion in six months. The Nvidia-backed startup is now targeting a $2 billion IPO on the Australian Securities Exchange in June or July, supported by a $10 billion Blackstone debt facility to build AI infrastructure across Australia.
TSMC controls over 90% of advanced semiconductor manufacturing, but Taiwan's reliance on liquefied natural gas imports through the Strait of Hormuz creates a critical vulnerability. With LNG stockpiles lasting only days, the Middle East crisis threatens to halt chip production that powers Nvidia, AMD, and the entire AI industry.
Intel is in talks with Google and Amazon to provide advanced chip packaging services worth billions annually, positioning itself as an alternative to capacity-constrained TSMC. The chipmaker's CFO projects packaging revenue exceeding $1 billion, up from hundreds of millions, as AI demand creates a critical bottleneck in chip production. Intel's EMIB technology competes directly with TSMC's CoWoS as tech giants scramble for custom AI chip capacity.
Intel unveiled Texture Set Neural Compression (TSNC), an AI-powered compression technology that can reduce game texture sizes by up to 18 times while maintaining image quality. Nvidia demonstrated similar capabilities with its Neural Texture Compression, cutting VRAM usage from 6.5GB to just 970MB in demo scenes. Both solutions address mounting storage and memory challenges as modern games demand increasingly detailed assets.
Apple has officially approved TinyGPU drivers from Tiny Corp, enabling AMD and Nvidia eGPU support on Apple Silicon Macs for the first time since the M1 launch. The drivers are designed exclusively for AI workloads, allowing users to run larger local AI models without bypassing system protections. Mac Mini owners running OpenClaw agents can now tap into external GPU power via Thunderbolt/USB4 connections.
DeepSeek's anticipated V4 model will run on Huawei chips instead of Nvidia hardware, marking a strategic shift in China's AI development. Chinese tech giants including Alibaba, ByteDance, and Tencent have ordered hundreds of thousands of Huawei chips ahead of the launch. The move signals China's push for AI self-sufficiency amid US export restrictions.
Elon Musk's SpaceX has confidentially filed for what could become the largest IPO in history, targeting a valuation exceeding $2 trillion. The company, now merged with xAI, plans to raise up to $75 billion to fund ambitious projects including orbital data centers, Starship development, and a lunar base.
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.