Hardware·
NVIDIA DGX Spark 64GB Expands Local AI Development Options
The new 64GB configuration of NVIDIA’s DGX Spark workstation brings unified memory and the full AI software stack to developers, enabling private local agent execution and easy two‑node scaling via Sync Cluster Assistant.
The DGX Spark platform now ships with a 64 GB unified memory option, retaining the GB10 Grace Blackwell Superchip, DGX OS and the complete NVIDIA AI software stack.
Local AI without cloud reliance
Developers can run agents, inference and fine‑tuning workloads entirely on the device, keeping data private and eliminating per‑task cloud costs.
When a project outgrows a single node, two units link directly through a QSFP cable and the built‑in Sync Cluster Assistant automatically configures the ConnectX‑7 NIC, pools memory to 128 GB and delivers up to 1.7× performance in benchmark tests.
The system supports models up to 100 billion parameters on one node and up to 200 billion when clustered, with out‑of‑the‑box compatibility for Ollama, vLLM, PyTorch, the NVIDIA Agent Toolkit and Blender’s upcoming installer.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- dgx spark
- local ai
- gpu cluster
By AiGpu Editorial · Editorial rewrite based on public reporting (NVIDIA Blog)
← All articles