
NVIDIA DGX Spark GB10 Grace Blackwell AI Desktop Computer
Best price now
Prices are checked against each store and labelled with when.

Prices are checked against each store and labelled with when.
The DGX Spark is a small desktop AI computer built around the GB10 Grace Blackwell superchip: a 20-core Arm CPU plus a Blackwell GPU rated at up to 1 petaFLOP of AI compute. Its 128GB of unified LPDDR5x memory is shared between CPU and GPU, so it can run and fine-tune models up to roughly 200 billion parameters locally instead of in the cloud. It ships with DGX OS, CUDA, PyTorch and Jupyter preinstalled, plus 4TB of self-encrypting NVMe storage and 10GbE networking, in a chassis about 15 cm deep. It suits developers, researchers and small teams who want local AI work without a server room; against cloud GPU rentals it trades per-hour flexibility for ownership, and against a workstation with a discrete RTX card it offers far more unified memory but less raw graphics throughput for gaming and some rendering.
In stockReturn policy
Hands-on coverage of the DGX Spark treats it as a genuine small-form-factor AI dev box: reviewers praise the 128GB unified memory and the ability to run large models on a desk, while noting the Arm CPU and modest memory bandwidth limit it against bigger GPUs.
The NVIDIA DGX Spark is a desktop AI supercomputer built on the NVIDIA GB10 Grace Blackwell Superchip. The chip pairs a 20-core Arm-based Grace CPU with a Blackwell GPU and delivers up to 1 petaFLOP of AI compute.
It has 128GB of unified LPDDR5x memory shared across the CPU and GPU, and can run and optimize models with up to 200 billion parameters locally. Storage is a 4TB self-encrypting NVMe SSD.
NVIDIA DGX OS comes preinstalled, with CUDA libraries and frameworks including PyTorch and Jupyter. The system has a 10GbE RJ-45 Ethernet port. The chassis is 15 cm deep and fits on a desk, and multiple units can be linked for greater scalability.
It checks today's price at every store, compares the alternatives and answers in about a minute.