NetApp and Iterate.ai detailed a private AI appliance that runs on corporate storage, while CoreWeave said GPU networking and power can swing AI latency by orders of magnitude.
Nvidia AI chips are still reaching China despite US export controls, Bloomberg reports. Officials are asking why the company missed red flags and say the curbs have created a sprawling shadow trade.
Products & Applications · Companies · AI Chips & Compute
TechRadar reported that at Intel Connection 2026 in China, Gunnir showed a compact dual-GPU workstation with two Intel Arc Pro B70 cards and 64GB of ECC memory, while Maxsun demonstrated an MCIO-based Mini-ITX system that relocates the graphics card from the motherboard. Neither company announced pricing or launch dates.
AI Chips & Compute · Markets · Products & Applications
NetApp CMO Gabie Boko says AI is drawing chief data officers and other new buyers to data infrastructure, as the company promotes its Intelligent Data Infrastructure strategy and its newly announced NetApp Novus architecture for AI factories.
Micron says RAM and storage shortages will worsen through 2028, with most 2027 capacity already sold, while forecasting record revenue and an 86.25% gross margin.
AI Chips & Compute · Companies · Large Language Models
Cognition and CoreWeave executives said always-on AI agents such as Devin require infrastructure that continuously links inference, feedback and training, as CoreWeave announced Forge to support that loop.
TechRadar reported that AMD's 256-core EPYC 'Venice' 9996 is its most expensive processor ever launched, with 600W, 1GB cache, PCIe 6.0 and MRDIMM support, likely enterprise-only until after 2030. Separately, TechRadar warned that cheap AMD Athlon Silver 10 laptops from Asus, Lenovo and HP are not worth buying.
Google launched a prototype orbital compute satellite aboard a SpaceX rocket from California on Thursday, carrying a Tensor Processing Unit into space for the first time. The mission is designed to test whether Google's AI chips can operate in orbit as the company pursues Project Suncatcher.
Large Language Models · AI Chips & Compute · Companies
Google this week unveiled Gemini 4 Argon, a frontier model it says advances coding, cybersecurity and complex tasks, and launched a prototype satellite for Project Suncatcher. Investors and consumers are pressing Google for a breakout personal agent, where Meta and OpenAI are moving fast.
NetApp and Nvidia are co-engineering Novus, a storage architecture for AI factories that separates data and metadata to scale independently. Executives say agentic AI and enterprise inference are raising concurrency and throughput demands.
Congress is weighing rival bills to shield ratepayers from AI data centers' power costs, as experts say residents will pay a share but the exact amount is unclear.
Texas Governor Greg Abbott ordered an audit of data centers seeking grid connections and paused permits, but NAACP groups and an investigation say thousands of diesel generators and on-site power plants may escape the crackdown.
Hugging Face released Olmo-core 3 on Oct. 1, 2026, an open training framework for large mixture-of-experts models. The system is designed to scale MoE training to more than a trillion parameters while reducing communication and memory costs, the company said.
AI Chips & Compute · Companies · Policy & Regulation
A months-long Verge investigation says Kevin O’Leary’s Wonder Valley, a planned 40,000-acre Utah AI data center, was halted by a political firestorm over secrecy and local opposition.
Markets · AI Chips & Compute · Large Language Models
The 10-year Treasury yield touched a 24-year high before easing, while U.S. stock futures rose. Micron reported strong results but fell premarket, Google launched Gemini 4 Argon, and Broadcom agreed to lend Anthropic up to $42 billion.
Kevin O’Leary and Fox News Media were sued for defamation after O’Leary claimed on Fox Business that critics of his planned Utah data center were linked to China, according to The Verge. Fox apologized, O’Leary said he had no evidence, and both moved to dismiss the suit.
Products & Applications · Companies · AI Chips & Compute
Huawei launched the Mate 90 series on October 1, 2026, with four flagships starting at 5,999 yuan and same-day sales. Top models add the Kirin 9050 Pro, a 12,000-nit display and a 5,999-yuan modular camera.
South Korea's September exports rose 83.5 percent to a record $120.9 billion as chip shipments passed $60 billion, while President Trump announced Korean investment projects in Texas, Alaska and eight nuclear reactors, and Seoul was ordered to pay Elliott $48.49 million.
CScale Inc. launched with $188 million in funding to develop optical interconnects that link thousands of accelerators across racks, aiming to improve reliability and performance in large AI scale-up clusters. Nvidia and Intel Capital are among the backers.