Technology & AI
FineServe shows LLM serving traffic varies by model type
FineServe, a new dataset from a commercial LLM marketplace, reveals serving traffic varies fundamentally by model architecture and task type.
AI Chips
AI chips and hardware — GPUs, custom silicon and the data centres behind AI, and what they cost.
24 stories
Technology & AI
NVIDIA GB300 chips built in Fort Worth, 500 jobs live
NVIDIA partner Wistron opened a 324,000-square-foot Texas plant making GB300 chips, with 500 jobs and a $700M commitment.
Technology & AI
GigaPath-Flash cuts pathology AI compute 50x, keeps 97% accuracy
GigaPath-Flash and GigaTIME-Flash shrink billion-parameter pathology models to run on 50x less compute while predicting tumor profiles from routine slides.
Technology & AI
NVIDIA Spectrum-6 doubles AI factory network to 102.4 Tbps
NVIDIA's 102.4-terabit Spectrum-6 Ethernet switch is reaching gigascale AI factories first at CoreWeave, Microsoft and Nebius, doubling network capacity.
Technology & AI
NVIDIA Cosmos 3 Edge: 4B model runs robots at 15 Hz
NVIDIA's new 4-billion-parameter open model generates 32 robot actions per inference on Jetson Thor, bringing real-time reasoning to edge devices for the
Technology & AI
PagedWeight cuts MoE serving memory 72% with FP16 accuracy
A new preprint claims PagedWeight dynamically quantizes MoE model weights at runtime, saving 72% GPU memory and lifting throughput 1.94× for AI inference.
Technology & AI
TerraZero: zero-demo self-driving AI hits 1.3M steps/sec
A new simulator from Applied Intuition trains driving policies with zero human demonstrations, topping the InterPlan benchmark on a single GPU.
Technology & AI
WarpGuard: first joint CPU-GPU attestation, no new hardware
Researchers from NVIDIA and Oxford propose WarpGuard, first to jointly verify CPU and GPU code at runtime, targeting safety-critical embedded systems.
Technology & AI
NVIDIA NeMo Automodel trains Diffusers models with no code rewrites
NVIDIA and Hugging Face shipped an open-source tool that fine-tunes video and image diffusion models across GPU clusters with no checkpoint conversion
Technology & AI
Traffic center AI costs $34/month with mixed model portfolio
A new arXiv framework cuts transportation management AI costs 97% by mixing open-source and closed APIs, with a greedy heuristic that picks the cheapest
Technology & AI
NVIDIA Cosmos 3 Edge puts 4B-parameter AI on Japanese robots
NVIDIA's new on-device model lets Japanese robotics giants like FANUC and Yaskawa run vision and reasoning locally on edge hardware, adapting to new tasks
Technology & AI
Bulkhead uses LLMs to detect and patch container escape bugs
Bulkhead, a new arXiv preprint, uses multi-agent LLMs to automatically find and fix path traversal vulnerabilities in containers running AI workloads.