A Brookings study by Stijn van Nieuwerburgh estimates the AI buildout could average 3.63% of U.S. GDP per year and double residential electricity use, while complex financing and unproven revenue create downside risk.
AI Chips & Compute · Products & Applications · Companies
Asus has released the ExpertCenter Pro ET900N G3, a 27 kg desktop workstation powered by Nvidia’s GB300 Grace Blackwell Ultra Superchip, with 748GB coherent memory and estimated pricing of $100,000 to $120,000.
U.S. stocks head into a data-heavy week with jobs and inflation reports, a Federal Reserve rate-path debate, Micron earnings as a test for AI shares, and an Eli Lilly obesity-drug update.
Zhongke Tongliang's SmarCo series appeared at the fifth Global Digital Trade Expo in Hangzhou, featuring a 200-TOPS AI workstation, local 70B large-model support, AI digital employees, and RISC-V-based video and storage products.
AI-driven demand for video memory has pushed new graphics card prices up sharply in 2026, and Engadget reports that used GPUs with large VRAM buffers now resell for 40 to 60 percent of their original price — sometimes above it.
Australian politicians and industry leaders argue that public opposition to AI datacentres is largely imported from the United States, while local protests and inquiries focus on energy, water and planning concerns.
SiliconANGLE customer research says GPU scarcity opens the door for CoreWeave, but performance, cost and operating experience determine whether customers stay. The study of 13 interviews found inference and training workloads both growing, hyperscalers still embedded and not every proof of concept becoming a signed customer.
AI Chips & Compute · Products & Applications · Companies
The 10-year Treasury yield hit 5.23% Friday, its highest since 2007, as sticky inflation, Fed rate-hike expectations and heavy government and AI-related bond issuance pushed yields higher.
The AI buildout is creating blue-collar jobs in data centers and related infrastructure, with welders and pipefitters seeing sharp demand gains, even as most Americans oppose local data centers, CNBC reported.
TechRadar advises buying a mainstream GPU now rather than waiting for Black Friday 2026, pointing to rising U.S. prices, a memory shortage and expected further hikes.
Android Authority reports that Qualcomm’s regular Snapdragon 8 Elite Gen 6 omits High Performance Memory and 8K60 recording found on last year’s Snapdragon 8 Elite Gen 5, and also lacks the APV codec, while the Extreme variant keeps them.
AI Chips & Compute · Large Language Models · Companies
Inferact says 16 Google TPU v7 chips ran Kimi K3 at 709 tokens/s, 57% faster than Nvidia GB200, using DeepSeek's DSpark and a megakernel. Separately, open-source Colibri runs 744B GLM-5.2 on a 25GB laptop by offloading MoE experts to SSD.
Anthropic will spend $11.6 billion over seven years on Akamai's cloud infrastructure, the largest contract in Akamai's history and more than six times a previously reported $1.8 billion agreement between the two companies.
Alibaba wants to expand data center capacity beyond 20 GW by 2032 and says its new V900 AI chip can scale to 500,000-card clusters, according to TechRadar. The report calls the V900 China's most powerful AI chip and asks whether it can displace Nvidia GPUs.