Hardware
Chips, devices and infrastructure · 3 stories
Nvidia and Microsoft open RTX Spark laptop preorders for local Windows agents
On October 7 Jensen Huang and Satya Nadella opened preorders for RTX Spark Windows laptops, pairing up to 128GB of unified memory and 1 petaflop of local AI with generally available Microsoft Execution Containers for sandboxed agents.
Nvidia Groq 3 LPX enters full production for low-latency agent inference
On August 24 Nvidia said Groq 3 LPX, a rack-scale SRAM inference accelerator for the Vera Rubin platform, is in full production, citing 3,400 output tokens per second on Gemma 4 31B at 100,000-token context in Artificial Analysis numbers.
Nvidia's 64GB DGX Spark brings a $4,999 desk-side box for 100-billion-parameter local AI
Nvidia announced a 64GB configuration of its DGX Spark desktop AI computer on October 2, starting at about $4,999 and shipping from six PC makers on October 23.