Category
The reported up-to-$3 billion NVIDIA investment in Lancium and Amazon's acquisition of the GW Ranch project in Texas show that AI investment has moved toward securing grid interconnection and dedicated power generation.
Google DeepMind has released WeatherNext Cyclones, which simultaneously forecasts the track, intensity, and wind field of tropical cyclones. While it reports an average lead advantage of more than one day, it also lays out the operational conditions of the open model and its boundaries with official warnings.
The price per gigabyte of memory has spiked back to 2007 levels, reversing a half-century-long exponential decline in just a matter of months. While AI-driven HBM demand is expanding at an annual rate of 200%, supply growth remains stuck at around 20%, and the effects are already spilling over to consumers through price hikes on game consoles and PCs.
A test configuration of RTX Spark with 18 cores and 5,120 GPU cores has appeared in Geekbench 7. Reading the performance gap against the 20-core version and the limits of a single benchmark run, we examine the product tiers NVIDIA may be considering.
Bloomberg reports that Apple is considering an Apple Watch redesign that could include round displays and a screenless device. This fall's update will be iterative, with a future focus on rebuilding the health data experience.
Across six LLMs assigned occupational roles, compliance rates with inappropriate requests from a higher-status role rose by 2.0 to 3.7 percentage points. The findings are limited to synthetic English dialogues, and questions remain about statistical procedures and reproducibility.
Niron is advancing construction of a full-scale iron nitride magnet plant in Minnesota, targeting 2027 startup and up to 1,500 tons annual capacity. While bypassing rare-earth separation and refining, mass-production yield and application-specific qualification will be the next factors to watch.
Samsung is mass-producing the ISOCELL HPC, which keeps the same 200MP, 1/1.3-inch format while enhancing DeepPix and multi-zoom HDR. The relationship to Galaxy S27 and HP6 remains unconfirmed.
A security researcher has published a leaderboard competing for the "slowest x86 instruction." By pointing fxrstor64 at PCIe MMIO while saturating the bus from other cores, he stretched an instruction normally taking 150 cycles to 198 billion cycles (62 seconds). This "slowness" undermines the exclusivity assumption behind SMM, potentially reigniting over 100 dormant CVEs.
AWS reportedly asked internal engineers to conserve EC2 CPU capacity. As AI agents drive up demand for general-purpose CPUs, we examine the supply constraints Amazon itself has acknowledged and the conditions for safely reclaiming spare capacity.
