AI News
NVIDIA DGX Spark 64GB: A New Local AI Configuration with Cluster Scaling
NVIDIA announces DGX Spark 64GB unified memory configuration from six partners, with cluster support via NVIDIA Sync Cluster Assistant for scaling local AI workloads.
What this piece is grounded in
01
NVIDIA DGX Spark will be available with 64GB of unified memory from Acer, ASUS, Dell, Gigabyte, HP and MSI.
02
Two clustered 64GB systems delivered up to 1.7x performance compared with a single system.
03
DGX Spark 64GB is available on Friday, Oct. 23, starting at $4,999.
04
Two units can connect directly with a QSFP cable, pooling their memory to 128GB.
01 / FIELD NOTE
What Changed: A New 64GB DGX Spark Configuration
NVIDIA announced that DGX Spark will be available with 64GB of unified memory from manufacturer partners Acer, ASUS, Dell, Gigabyte, HP and MSI. The new SKU runs capable local agents on device privately, without cloud dependency, according to the NVIDIA blog post. The 64GB configuration retains the GB10 Grace Blackwell Superchip, DGX OS and full NVIDIA AI software stack, same as the 128GB model. The system supports up to 100-billion-parameter models and agentic applications built on them, fully on device. Availability is scheduled for Friday, Oct. 23, starting at $4,999.
02 / FIELD NOTE
Why It Matters for Practical AI Work
Editorial interpretation: The 64GB configuration gives developers a lower-memory entry point for running local AI agents and models without cloud dependency, based on the source's description of the new SKU. The source states two 64GB units can cluster together via NVIDIA Sync Cluster Assistant without additional setup, pooling memory to 128GB and expanding model support to up to 200 billion parameters. In NVIDIA's Qwen 3.8 27B test, two clustered 64GB systems delivered up to 1.7x performance compared with a single system. This suggests a scaling path from single-system work to multi-node clusters for larger workloads, as described in the source.
03 / FIELD NOTE
What a Reader Can Verify
A reader can verify the announced configuration details from the NVIDIA blog post: the 64GB unified memory option, the six named manufacturer partners, the $4,999 starting price, and the Oct. 23 availability date. The source also lists supported software including NVIDIA Agent Toolkit, CUDA-X AI libraries, Nemotron open models, and runtimes like Ollama, vLLM, and PyTorch with CUDA. The cluster capability via NVIDIA Sync Cluster Assistant and the QSFP cable connection for two units are described in the source. The NVIDIA Sync Model Launcher is stated as coming at the end of the month.
04 / FIELD NOTE
What Remains Unclear
The supplied feed entry is a concise summary; consult the canonical article for full context. The source does not provide details on how the 64GB configuration performs on specific workloads beyond the Qwen 3.8 27B test result. The source mentions Blender support with a prebuilt installer coming soon, but does not specify the exact release timing. The source lists three workflow examples without detailing their content. The source also references playbooks coming soon to 64GB devices without naming them. The relationship between the 64GB and 128GB models beyond the shared components is not fully elaborated in the source text.
Questions readers ask
What is the NVIDIA DGX Spark 64GB configuration?
NVIDIA announced DGX Spark will be available with 64GB of unified memory from manufacturer partners Acer, ASUS, Dell, Gigabyte, HP and MSI. The new SKU runs capable local agents on device privately, without cloud dependency, and supports up to 100-billion-parameter models fully on device. It retains the GB10 Grace Blackwell Superchip, DGX OS and full NVIDIA AI software stack, same as the 128GB model.
How does clustering work with two DGX Spark 64GB units?
Two 64GB units can connect directly with a QSFP cable, pooling their memory to 128GB and expanding model support to up to 200 billion parameters while delivering twice the memory bandwidth and up to 1.7x the performance, according to NVIDIA's Qwen 3.8 27B test. The NVIDIA Sync app configures this multi-node cluster seamlessly, detecting connected units and configuring the ConnectX-7 network.