The AI hardware landscape is shifting fast. A new GPU‑kernel optimization platform from doubleAI, announced just hours ago, promises to squeeze more performance out of existing NVIDIA B300 GPUs. The startup raised $220 million and immediately invested in 1,000…
Continue reading DoubleAI Unveils GPU‑Kernel Optimizer as Nvidia Reconsiders Chiplet CPUsKey developments this week show a mixed but accelerating AI‑hardware landscape. South Korean startup DeepX moved its DX‑M1 neural‑processing unit into mass production on Samsung’s 5‑nanometer process, securing 77 orders from more than ten countries and expanding its…
Continue reading AI Hardware Surge: New Edge Chips, Memory Chip Rout, and Major Data‑Center ShiftsKey developments: – OpenAI confirmed it is building a humanoid robot and plans a public demonstration in 2027, emphasizing that the AI “brain” is the crucial component and expanding its robotics team to 27 specialized roles. This marks…
Continue reading OpenAI Announces Humanoid Robot Demo for 2027 Amid Chip Shortages and New AI Chip ProductionKey developments: – Li Auto has separated its high‑performance M100 chip unit into a standalone subsidiary valued at $2.5 billion, underscoring the accelerating demand for custom AI silicon in automotive applications. – Amazon unveiled two consumer‑focused AI wearables—the $75…
Continue reading AI‑Hardware Surge: Li Auto’s $2.5 B Chip Spin‑off, Amazon’s New Bee Wearables, and Taiwan’s Crackdown on Chip SmugglingNvidia announced the launch of its Nemotron 3.5 Super VL, a 120‑billion‑parameter mixture‑of‑experts model with a 1‑million‑token context window and up to five times the throughput of comparable open models on a single GPU. The release includes smaller…
Continue reading Nvidia Unveils 120B Nemotron 3.5 Model as Samsung Rolls Out 400‑Layer V‑NAND and Taiwan Plans $4B US AI‑Hardware ParkMeta and Apple teamed up to launch Muse Spark Open, an 816 billion‑parameter mixture‑of‑experts model that runs on paired Apple M5 Ultra Mac Studios. The model’s massive size (42 billion active parameters and a 1 million‑token context) is made feasible by…
Continue reading New AI hardware breakthroughs accelerate open‑weight models and on‑device AIKey developments: – China’s operational AI compute capacity has reached 24GW and another 50GW is under construction, narrowing the gap with the United States’ 56GW total, according to a Financial Times analysis of SemiAnalysis data. – Jeff Holden’s…
Continue reading Global AI‑Hardware Race Accelerates with China’s 24GW Data‑Center Rollout, Microsoft’s RTX Spark Dev Box, and New 3‑D Micro‑Machine Relays
