NVIDIA and Microsoft released RTX Spark laptops and compact desktops on October 16, allowing users to run AI models locally on Windows PCs. It is the company's first major AI hardware update since the launch of the RTX 40 series.
NVIDIA reported 1 petaflop of FP4 AI performance, measured on the RTX Spark superchip, which combines a Blackwell RTX GPU with a Grace CPU. That compares with the previous RTX 40 series, which offered up to 1.5 petaflops of FP16 performance.
RTX Spark is built on the NVIDIA CUDA platform and targets developers, creators, and gamers. Availability begins October 16, initially for developers and creators.
"With up to 128 gigs of unified memory and up to a petaflop of AI compute, you can run models on this laptop that simply don’t fit on a traditional machine," said Pavan Davuluri, Microsoft’s EVP of Windows and Devices. The laptop is designed for running increasingly capable AI models locally without cloud dependency.
The announcement follows a Microsoft event in San Francisco, where NVIDIA and Microsoft outlined their co-engineered AI agent strategy for Windows. The collaboration aims to bring AI capabilities to the desktop, aligning with the growing demand for local AI solutions.
NVIDIA did not say how much RTX Spark will cost, and it remains unclear how the AI models will be optimized for different use cases. The company plans to expand the RTX Spark ecosystem with more hardware partners and software tools.
Source: nvidia
