Latest
A Pew Research Center survey found that 55% of Americans aged 18-29 said their concerns about AI outweigh their excitement. 73% predicted that AI will reduce the number of jobs in the U.S. over the next 20 years, with high usage rates and skeptical views coexisting.
Windows 11's Task Manager can now display per-app NPU utilization, execution engine, and memory usage. Standard tools make it easier to pinpoint where AI processing runs and what's driving the load.
The US critical and emerging technologies list has been reorganized from 18 domains to 14. While explicit references to data centers and battery storage have disappeared, federal R&D funding allocation is now tied to basic research and national security applications.
A ground-up explainer on HBM (High Bandwidth Memory), the technology shaping AI semiconductor performance. Covers the mechanics of stacking DRAM and connecting it via TSVs, the differences from GDDR, the three-way competition and base die logic shift in the HBM4 generation, the market's rapid expansion toward roughly $60 billion, the resulting DRAM price surge, and the business opportunities for Japanese companies.
Reuters reports that new SF4 orders from Samsung rose 10-15% in the US and China, and 5-10% in Taiwan. While demand from China exceeds supply capacity, prioritization of US customers and capacity reserved for in-house use are constraining the production allocation.
Anthropic's Q2 2026 revenue reached $11.6 billion, surpassing OpenAI's $6.7 billion for the first time. We break down the growth in enterprise sales and the differences in accounting scope.
We examine reports of a 30% cut in ABF supply to China against Ajinomoto's official IR disclosures and mass-production qualification processes, and outline the allocation factors and substitute-material conditions that determine the impact on advanced packaging.
Mesa's Intel Vulkan driver ANV now presents compressed memory types to general buffers on Xe2 and later. DXVK DX11 traces showed up to a 5.26% improvement, but reproducibility in actual games and frame time impact remain unverified.
A paper titled "Lost in Compaction" from Pennsylvania State University researchers demonstrates via the COMPINT benchmark that AI context compaction causes an average of 83% of user instructions to be lost. The study also points out that Anthropic's official documentation does not guarantee automatic instruction retention.
Seasonic's 1600W power supply has become the first 115V non-redundant ATX PSU to earn 80 PLUS Ruby certification. It recorded a measured 95.27% at 50% load and 88.88% even at a low 5% load, pushing the efficiency limit of the Titanium era up by one full point. A new milestone in power supply design for the age of AI workstations.
