News Nov 17, 2026 US moves to block China's remote access to advanced AI chips The Trump administration is tightening export rules so that foreign data centers can no longer freely rent restricted American AI chips to customers in China. The change follows a spying scandal involving Nvidia's chief executive and could reshape the global race for AI computing power. Giulio Prisco 2 min read
News Nov 6, 2026 A simpler way to train AI on light-based hardware Researchers in Japan show that a stripped-down training method and simple network design can run artificial intelligence on optical hardware while cutting computation. Giulio Prisco 2 min read
Community Sep 16, 2026 Silicon Clips AI’s Wings The future of artificial intelligence is no longer being decided only in laboratories of algorithmic design. It is also being decided in wafer fabs, high-voltage grids and server rooms. Eduardo Martínez de la Fe 3 min read
News Oct 13, 2026 Microsoft announces Maia 300 AI chip for 2027 deployment Microsoft has unveiled the Maia 300, its third-generation AI accelerator designed for multi-gigawatt datacenter deployments starting in 2027. The chip delivers up to 7 times the rack-scale throughput of its predecessor and targets long-context reasoning, conversational AI, and agentic workloads with significantly improved energy efficiency. Giulio Prisco 2 min read
News Oct 5, 2026 Google completes Project Suncatcher prototype launch for orbital AI compute Google's Project Suncatcher prototype satellite successfully reached orbit aboard a SpaceX mission to evaluate whether machine learning infrastructure can function reliably in space. The mission will test Tensor Processing Units under radiation, thermal extremes, and launch vibration conditions, with future milestones planned for 2027 including satellite laser interconnection tests. Giulio Prisco 2 min read
News Sep 25, 2026 Three tech giants demonstrate optical die-to-die interconnect for AI scale-up Lumentum, Qualcomm, and Corning will showcase a high-density optical die-to-die interconnect designed to extend AI system bandwidth beyond electrical limits. The demonstration at ECOC 2026 features VCSEL-based optical engines and aims to support roughly 1 Tb/s/mm of bandwidth density with a path to 4 Tb/s/mm. Giulio Prisco 2 min read
News Sep 22, 2026 When the hardware itself becomes the neural network Self-organizing networks of nanowires and nanoparticles compute in the material itself so devices can process sensor data locally with little power. Giulio Prisco 2 min read
News May 19, 2026 Georgia Tech develops radiation-resistant memory for space missions New ferroelectric NAND flash technology withstands extreme radiation to support artificial intelligence data processing on spacecraft far from Earth. Giulio Prisco 2 min read
News May 18, 2026 Researchers map progress and future of magnetic topological materials for computing Scientists from the University of Ottawa and MIT review more than 20 years of research and identify pathways to room-temperature devices that could make electronics faster and more energy efficient. Giulio Prisco 2 min read
News Oct 8, 2025 xAI raises funding to $20 billion with Nvidia investment Elon Musk's AI startup gets more money than planned, including from chipmaker Nvidia, to speed up its work on powerful data centers. Giulio Prisco 2 min read