Local AI Clustering with Dell's Pro Max GB10: Pairing Two Nvidia Grace Blackwell Systems
Tom's Hardware tested a pair of Dell Pro Max with GB10 systems (based on Nvidia's DGX Spark platform) as a local AI cluster, achieving 256GB of unified memory for running large language models that exceed single-system capacity. Each system costs $6,332 (with 4TB SSD) and features Nvidia's ConnectX 7 200Gbps NICs for RDMA clustering via RoCE. The setup offers a cheaper, quieter, and more desk-friendly alternative to traditional GPU servers with multiple discrete GPUs, which can cost over $20K and require high-power circuits and loud cooling. The Dell variant adds a power LED and 280W USB-C adapter, improving on the reference DGX Spark design. The article concludes this is the most cost-effective way to run large local AI models with high-performance networking.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page itself is projected from evidence records.
- Current automated evidence projection