Latest news, developments and reporting on AI Chips & Compute.
Chinese TPU chip maker Zhonghao Xinying showcased its new 'Xuyu' chip at the Global Unicorn Summit in Hong Kong, claiming inference costs as low as 35-50% of overseas GPUs.
Cursor open-sourced MoK to cut MoE communication latency on Nvidia GPUs, while Meta released Muse Glimmer, a 30B local agent model with 128K context. Both tackle software-level bottlenecks in AI execution.
Global pure-play foundry market rose 29% year-on-year in Q2, driven by AI chip demand, with TSMC leading at 73% share.
SK hynix broke ground on a $4bn US HBM plant; South Korea also announced AI projects, military drills, and JTBC rehabilitation.
WRC 2026 embodied AI shifts from demos to ROI; domestic chips power commercial video generation and GLM-5.3-Flash.
Nvidia has agreed to acquire AI model repository Hugging Face for $12.9 billion, according to The Information, a deal that would place the largest open-source AI platform under Nvidia's control.
Zhipu open-sources a 320B-parameter multimodal model, while Google, NVIDIA and Perplexity launch new AI products.
Jim Cramer says strong Nvidia and Salesforce earnings shattered bear narratives, sending shares up 8% and 22%.
Nvidia's strong results reignited AI stock rally; market awaits Fed's Warsh at Jackson Hole.
Nvidia reported $96 billion in quarterly revenue, up over 100% year over year, but guided to only 70% growth in fiscal 2028 due to supply limits, while demand may double. The gap is reshaping AI infrastructure procurement.
Reports say a new Apple TV 4K could debut at the September 9 event, featuring Siri AI, a faster A-class chip, an N1 wireless chip, and an updated Siri Remote.
SK Hynix held groundbreaking for its first U.S. HBM packaging plant in Indiana, aiming to make the state a key memory base by 2030.
Nvidia has reportedly agreed to acquire open-source AI platform Hugging Face for $12.9 billion, expanding its reach into the AI software ecosystem.
Anthropic has agreed to rent AI compute capacity from UK-based Nscale in a six-year, $45bn deal, gaining access to 460MW of power powered by Nvidia's Vera Rubin chips.
Zhipu launched open-source GLM-5.3-Flash, running on SenseTime's domestic AI infrastructure, scoring 57 in A2I.
The largest planned AI datacenter in New Jersey is accused of running dozens of unpermitted gas-powered generators, according to a report by The Guardian. The DataOne facility in Vineland is also facing noise complaints and a lawsuit.
Nvidia has agreed to acquire AI platform Hugging Face for $12.9bn, according to reports, in one of the biggest deals of the generative AI era.
Apple and Xiaomi have pushed AI chip memory bandwidth toward terabyte-per-second levels, exposing the 'memory wall' as the key constraint in on-device AI. The trend is reshuffling the memory industry and driving diverse architectural fixes.
Chinese AI firm Z.ai says its new model runs only on domestic chips; shares jumped over 8% on the news.
German minister: AI demand strains data centers; Europe must act, and approvals will be sped up.
Nvidia agrees to buy open-source AI platform Hugging Face for $12.9 billion, report says.
Nvidia posted $96.2B quarterly revenue, but analysts see its next growth frontier in distributed AI compute at the edge, using high-speed fabrics to overcome power density limits.
Nvidia reported Q2 revenue of $96.2 billion, up 106%, and announced Amazon will add 2 million more GPUs to AWS, expanding their partnership.
Nvidia posted record $96.2 billion quarterly revenue and forecast $108 billion next quarter; economist Mark Zandi says AI drives at least a quarter of growth.