AI Briefing#blackwell
The latest AI and developer news about #blackwell, with the original source and a short summary.
The latest AI and developer news about #blackwell, with the original source and a short summary.
At Dell Technologies World, Dell and NVIDIA unveiled new Vera Rubin-based AI factory products and an on-premises agent strategy.
The article argues that portability in AI kernels is a myth, and top performance comes from hardware-specific DSLs.
The llama.cpp b9095 update enables tensor parallel processing on Blackwell PCIe GPUs without NCCL.
Atlas, a high-performance inference engine built on Rust and CUDA, has been released as open source.
The specifications for NVIDIA's personal AI supercomputer, the DGX Spark, equipped with NVIDIA's Blackwell superchip, have been revealed.
On dual RTX 5060 Ti 16GB GPUs, Qwen3.6 27B ran at about 60 tok/s.
llama.cpp b8967 adds native NVFP4 support for Blackwell.
NVIDIA B200 GPU spot prices surged 114% in six weeks.
NVIDIA unveiled Blackwell Day-0 support for DeepSeek V4 with performance of 3,500TPS per second.
Google Cloud and NVIDIA are expanding AI Hypercomputer to grow agentic and physical AI.
DeepSeek has updated DeepGEMM with Blackwell optimizations and Mega MoE support.
Nvidia's moat lies not in the chip itself but in its power to control the supply chain and ecosystem together.
Cursor's multi-agent system made 235 CUDA kernels 38% faster in just 3 weeks.
Surging Blackwell GPU prices and longer contract lock-ins signal that AI compute shortages have become a real problem.
Verified Qwen3.5-122B running at 198 tok/s on 2x RTX PRO 6000 Blackwell.
Roche is expanding its AI infrastructure globally by deploying more than 3,500 NVIDIA Blackwell GPUs to boost drug discovery and manufacturing efficiency.