Supermicro’s Super AI Station puts NVIDIA’s Grace Blackwell GB300 platform into a system that can sit beside a desk or mount in a rack. The useful detail is its capacity: ServeTheHome lists 252GB of GPU HBM3E and 496GB of system memory, making this a substantial local-AI machine rather than a graphics card with ambitions.
NVIDIA Watch analysis
What happened
ServeTheHome’s review describes the Super AI Station, model ARS-511GD-NB-LCC, as part of NVIDIA’s DGX Station initiative. The system pairs a 72-core Grace CPU with a Blackwell Ultra GPU, runs Ubuntu 24.04 LTS with NVIDIA AI Developer Tools, and has a 1,600-watt power supply.
Its convertible chassis can be used as a desktop workstation or installed as a 5U rack server. ServeTheHome also lists two 400Gbps network ports, two M.2 PCIe Gen 6 slots and an optional NVIDIA RTX PRO Blackwell graphics card. The review positions the system for local AI work; those specifications describe the hardware, not a particular model’s speed or performance.
Why it matters
A machine with 252GB of GPU memory gives AI teams a sizeable local compute platform, while the rack option offers a route to deploy it outside a conventional workstation setup. The design also brings server-scale power and cooling needs into a system that can, technically, live under a desk. Your power bill may have questions.
Our read
This is an interesting bridge between the workstation and server markets, with local AI as the reason to care. The memory capacity and flexible form factor are concrete; without benchmarks or pricing in the supplied review extract, they do not tell us what models it can run at what speed or cost. Buyers will want those answers before confusing “fits in the chassis” with “fits the budget”.
What to watch
- ServeTheHome’s performance results and the workloads used to produce them.
- Pricing, availability and the practical power and cooling requirements.
- How the rack-mounted configuration compares with a conventional server for local AI work.
Discussion spark: For local AI work, would you choose a desk-side system that can also go in a rack, or keep workstation and server hardware separate?
Sources and evidence
- Source update (5 November 2026, 21:15 UTC)
Independent WittyWires tracker for public updates about NVIDIA. Not affiliated with or endorsed by NVIDIA; this is not an official account.