NVIDIA is making its DGX Spark personal AI supercomputer a bit more accessible, launching a new 64GB unified memory configuration available from partner manufacturers Acer, ASUS, Dell, Gigabyte, HP, and MSI. It ships ready to go with DGX OS and NVIDIA’s full AI software stack pre-installed, letting developers, researchers, and AI enthusiasts run genuinely capable local agents without depending on the cloud at all.
Except for the changes in RAM, the rest of the kit is more or less the same – GB10 chip, ConnectX-7 networking, CUDA-accelerated AI software stack. 2 units of the new SKU can also be clustered together to deliver up to 1.7x the performance of a single unit, with room to scale further as workloads demand.
The NVIDIA Sync app is the main tech behind the cluster connection, automatically detecting connected units, validating their configuration, and setting up networking without manual setup. Every node runs the identical software stack, so scaling from one unit to two doesn’t require reconfiguring anything. Coming later this month, NVIDIA Sync Model Launcher will make running local models even simpler, letting developers download and launch something like Qwen3.8 27B with just a few clicks, whether on a single unit or a cluster, and making it accessible right from a connected laptop.
Out of the box, DGX Spark comes ready for real agent development work, supporting the NVIDIA Agent Toolkit, CUDA-X AI libraries, Nemotron open models, and popular runtimes like Ollama, vLLM, and PyTorch with CUDA already built in. Blender is also among the first major creative apps lining up to support the platform, with a prebuilt installer coming soon.
In terms of practical use, NVIDIA points to a few solid examples: running a coding or research agent continuously in the background, offloading AI model inference to DGX Spark while your actual laptop or desktop stays free for other tasks, or scaling up to a two-unit cluster when a single system runs out of room for a bigger model or longer context window.
The DGX Spark 64GB configuration becomes available on October 23 from Acer, ASUS, Dell, Gigabyte, HP, and MSI, starting at US$4,999.
