Samsung zHBM: Memory Stacked on AI Chips — 8x HBM5 Speed
Samsung unveiled zHBM at FMS 2026 — a memory architecture that stacks HBM directly above AI accelerators for 8x HBM5 bandwidth. Here is ...
AMD Buys Taalas: AI Weights Etched Into Silicon Today
AMD acquired Taalas today: AI model weights etched into silicon, 17,000 tokens/sec. 74x faster than ...
Anthropic Custom Chip Team: What It Means for Claude Devs
Anthropic confirmed an in-house silicon team to co-design custom chips for Claude inference. Here is ...
NetBSD 11.0: 10ms MicroVM Boot and Stable RISC-V Port
NetBSD 11.0 ships a MICROVM kernel that boots in 10ms, the first stable RISC-V BSD ...
Galaxy Unpacked 2026: The Developer Action List
Samsung announced Galaxy Z Fold 8, Galaxy Glasses, and Watch 9 today. Here is what ...
NVIDIA Cosmos 3 Edge: On-Device Robot AI for Developers
NVIDIA Cosmos 3 Edge is a 4-billion-parameter World Action Model that runs on a GeForce RTX 3070. Released July 20, 2026, it brings ...
TSMC Q2 2026: Agentic AI Is Now Your Chip Problem
TSMC's record $40.2B quarter reveals agentic AI is driving a new CPU demand wave on ...
AMD Helios Hits Azure: 72 GPUs, 31 TB HBM4, Rival Nvidia
AMD Helios launches with 72 MI455X GPUs and 31 TB HBM4. Microsoft Azure is deploying ...
NVIDIA Cosmos 3 Edge: Run a World Model on Jetson Hardware Now
NVIDIA's Cosmos 3 Edge is a 4B-parameter on-device world model for robots. Here's what the ...
OpenAI’s Screenless AI Speaker: What Developers Need to Know
OpenAI revealed its first hardware product — a screenless AI speaker powered by GPT-Live. Here’s ...