Nvidia Launches 64GB DGX Spark Starting at $4,999
Nvidia is expanding its hardware lineup to accommodate modern local AI models and a wider silicon supply environment.
Radwan AbdulhadiEditor and technology writer
A New 64GB Configuration for Local AI
Nvidia is tailoring its popular DGX Spark platform and the GB10 SoC to better fit the realities of today's artificial intelligence models and the broader silicon supply crunch. The company is introducing a new 64GB version of the Spark designed to be more affordable for local AI trailblazers who do not require a full 128GB of memory. Further details are available from Tom's Hardware in the original source material.
Earlier expectations in the industry assumed that massive amounts of memory would be mandatory to execute the most intelligent models locally. Consequently, unified memory systems utilizing architectures like AMD's Strix Halo, Apple's M-series chips, and Nvidia's GB10 were originally configured with 128GB of RAM or more.
Adapting to Evolving Model Sizes and Market Realities
However, the artificial intelligence landscape moves rapidly. Highly intelligent dense models such as Qwen 3.8 27B can now run comfortably inside 32GB of RAM, albeit with a limited context window. Because of these optimizations, the original Spark's 128GB capacity is no longer strictly essential for basic local inference workloads.
Furthermore, skyrocketing memory prices mean that hardware permanently paired with large amounts of costly LPDDR5X can act as a barrier to entry. A streamlined Spark with a lower memory footprint provides a practical alternative for users who want the supporting software stack and capabilities of the GB10 Superchip without paying for unused capacity.
Pricing, Availability, and Hardware Specifications
Affordability in the current market remains relative. The 64GB GB10 systems built by hardware partners including Acer, Asus, Dell, Gigabyte, HP, and MSI are scheduled to start at $4,999 when they launch on October 23. Market volatility, however, means prices may fluctuate due to shifting memory and storage costs.
By comparison, finding an in-stock 128GB GB10 system currently runs between roughly $7,000 and $9,000, which sits far above Nvidia's initial adjusted MSRP figures. Despite the reduction in memory capacity, the 64GB variants retain the identical ConnectX 7 RDMA NIC found on their 128GB counterparts, allowing users to cluster multiple units together if they eventually outgrow the 64GB limitation.
The internal architecture also carries over the same 20-core Arm CPU complex alongside 273 GB/s of shared memory bandwidth designated for both the CPU and GPU subsystems.
New Software Tools for Clustering and Model Deployment
To help users leverage the included ConnectX 7 NIC, Nvidia is introducing a new software utility called the Nvidia Sync Cluster Assistant. This tool automates the process of connecting multiple Spark systems together to scale up performance. Additionally, the Sync software serves as a remote connectivity application enabling users to harness their workstation's processing power from separate Mac and PC systems.
Nvidia is also adding a Model Launcher feature to the Sync application. This addition automatically downloads and executes models like Qwen 3.8 27B across one or more connected systems, linking them directly to the OpenCode browser-based coding agent for immediate agentic workflows.
Coexistence within the Existing Product Ecosystem
The original 128GB DGX Spark configuration will remain available for users who require extra space to support larger models, extensive cluster scaling, or memory-intensive tasks such as model fine-tuning. Nvidia's introduction of the 64GB variant expands consumer choice, giving local artificial intelligence enthusiasts an option that better aligns with their specific operational budgets and technical workflows.
Sources
- Tom's HardwareNvidia introduces 64GB DGX Spark to throw local AI fans a lifeline amid the RAMpocalypse
Continue chronologically
Related entity coverage
Related stories
NVIDIA DGX Spark 64GB Launches Local AI Development
NVIDIA announces the DGX Spark 64GB personal AI supercomputer, featuring unified memory and partner support starting at $4,999.
Google Launches Googlebook Laptops for Android Users
Google has officially unveiled Googlebooks, a new line of laptops designed to provide seamless connectivity and synergy with your Android smartphone.
Google Launches Android-Powered Googlebook Laptop Lineup
Google is introducing its new line of Android-based Googlebook laptops on October 4th, featuring deep mobile integration and Gemini AI capabilities.
MSI Stealth A18 AI+ RTX 5090 Laptop Gets $520 Discount
Discover a major $520 discount on the MSI Stealth A18 AI+ gaming laptop featuring an RTX 5090 GPU, 64GB DDR5 RAM, and a 2TB SSD.
