NVIDIA Will Launch 64 GB DGX Spark Model Soon
The new AI supercomputer entry will debut in late October with a $4,999 price tag.
Updated on Oct. 2, 2026 in Artificial Intelligence
Live Poll
Is now a good time for individual developers to purchase expensive specialized AI hardware?
NVIDIA is set to expand its DGX Spark AI supercomputer lineup with a 64 GB model hitting the market on October 23, 2026. Priced at $4,999, the new system includes a ConnectX-7 NIC and utilizes the NVIDIA AI software stack.
Why it matters
The release aims to offer a more accessible entry point for AI development as rising costs for high-end memory and components have forced the price of the 128 GB model above $6,000. This shift addresses the growing demand for scalable computing hardware while managing ongoing manufacturing price pressures.
The new 64 GB system features a ConnectX-7 NIC and runs the DGX OS. Units can be connected via QSFP cables, allowing users to pool unified memory through the NVIDIA Sync Cluster assistant.
The players
NVIDIA
NVIDIA is a global technology company known for designing graphics processing units and AI-focused hardware and software stacks.
AMD
AMD is a semiconductor designer that produces competitive AI-focused processors and hardware platforms.
The details
Produced by manufacturing partners including Acer, Asus, Dell, Gigabyte, HP, and MSI, the systems are designed for modular AI workloads. Clustering two units together is projected to provide a 1.7x performance increase, positioning the hardware to compete against AMD Ryzen AI Halo systems that currently range from $3,500 to $4,500.
Timeline
September 2026: Testing of the DGX Spark configuration occurred.
October 23, 2026: The 64 GB model will become available through manufacturing partners.
The Tech Race
This release follows the integration patterns of the NVIDIA AI software stack as manufacturers look to standardize high-performance computing hardware. It marks a significant shift in the competitive landscape as companies battle to lower barriers to entry against rival platforms like AMD Ryzen AI.
Developers and AI researchers can expect a new price-conscious option for deploying local AI supercomputing workloads. The standardized use of the NVIDIA software stack may also simplify integration for those already familiar with the existing 128 GB systems.
The takeaway
The move demonstrates a strategic pivot toward tiered performance options as hardware costs rise across the sector. Users should evaluate whether the 64 GB capacity meets their specific model training needs or if the higher-memory 128 GB unit remains the necessary long-term investment.
Further reading
Explore more about evolving hardware standards in the Artificial Intelligence section.
Live Poll
Is now a good time for individual developers to purchase expensive specialized AI hardware?
