Supermicro Super AI Station GB300
Use this to tell your different configurations apart in your cart — handy if you order the same product with different options.
Full Specifications
Compute & Memory
Superchip
NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip
CPU
1× NVIDIA Grace CPU Superchip, 72 Arm Neoverse V2 cores
GPU
1× NVIDIA Blackwell Ultra GPU
Coherent Memory
748 GB unified: 496 GB LPDDR5X + 252 GB HBM3e
AI Performance
20 PetaFLOPS (FP4)
Storage & Physical
Storage (Slots)
4× M.2 2280 PCIe 5.0 x4 NVMe
Form Factor
Convertible deskside tower or 5U rack-mount, shared chassis design
Dimensions
21.5 x 48 x 49 cm
Networking & Cluster Scaling
High-Speed Networking
2× 400G QSFP112 ports via NVIDIA ConnectX-8 SuperNIC
Cluster Scaling
Up to 8 units in a single coherent cluster over 400GbE, expandable further into HGX B300 (8 GPU) or GB300 NVL72 (72 GPU) infrastructure
PCIe Expansion
PCIe 5.0, 1× double-width x16 slot + additional x8 slots
Power & Thermal
Power Supply
1600 W power supply, Titanium efficiency, standard office outlet
Cooling
Closed-loop liquid cooling, serviceable, near-silent sustained operation
Management & Software
Remote Management
IPMI 2.0 and Redfish API, data-center-style monitoring and control
Software Stack
Preloaded Linux environment with NVIDIA AI Enterprise tooling, ready for NVIDIA NemoClaw agent deployment
Supermicro introduced the Super AI Station GB300 on October 28, 2025, as a deskside system based on the NVIDIA GB300 Grace Blackwell Superchip platform — a system built to run trillion-parameter models locally, while offering a documented path to cost efficiently and gradually grow to data center-scale AI capacity.
Hardware Specifications
The Super AI Station is built around a single NVIDIA Grace CPU Superchip — 72 Arm Neoverse V2 cores and 496 GB LPDDR5X memory — paired with an NVIDIA Blackwell Ultra GPU with 252 GB HBM3e GPU memory, unified through NVLink-C2C into a coherent memory pool of 748 GB. Up to 7 isolated MIG partitions let a single station serve several teams or workloads at once without contention. The system provides 4× M.2 2280 PCIe 5.0 x4 NVMe storage and a dual-port NVIDIA ConnectX-8 400G SuperNIC — the same networking fabric used to connect additional stations as a deployment grows.
Local AI Server for Business Use
The Super AI Station is designed to function as a shared local AI resource. Multiple employees can submit document processing tasks — contract review, invoice extraction, report drafting, email triage — throughout the working day, with jobs handled centrally on-site instead of routed through external cloud services.
Through NVIDIA NemoClaw, autonomous agents can run continuously on the system, handling recurring tasks such as inbox monitoring or scheduled report generation without requiring a person to initiate each request. This allows the Super AI Station to operate as ongoing AI infrastructure for the business, supporting steady document workloads.
Built to Scale: From One Station to an AI Factory
A single Super AI Station is a complete, self-contained system, but it is also the base unit of a larger deployment. Up to 7 additional stations can be added over time — up to 8 units — communicating over the same dual-port 400G fabric, without re-architecting existing infrastructure or workflows. Each added unit contributes its own 748 GB of coherent memory and compute to the pool, giving a growing organization a straightforward, incremental way to expand capacity as adoption increases.
When demand eventually outgrows an 8-unit cluster, the same 400G networking foundation extends into Supermicro's NVIDIA HGX B300 and NVIDIA GB300 NVL72 platforms. Stations purchased earlier keep running in production rather than being retired or replaced — the initial investment remains part of the deployment at every stage of growth.
Every station added along this path keeps working — nothing purchased earlier is replaced.
| Stage | Coherent Memory | Indicative Cost | What It Means |
|---|---|---|---|
| Single Station | 748 GB | $100K | Starting point; runs fully standalone. |
| 2-Unit Cluster | 1.5 TB | $190K | Adds 1 station over the same 400GbE fabric. |
| 4-Unit Cluster | 3 TB | $370K | Adds 3 stations; no change to existing setup required. |
| 8-Unit Cluster | 6 TB | $700K | Maximum single-cluster capacity on Super AI Station hardware. |
| HGX B300 Expansion | 9+ TB combined | ~₹17L (additional) | Existing stations keep running; dense multi-node training capacity is added alongside them. |
| GB300 NVL72 Expansion | 20+ TB combined | ~₹37L (additional) | Unified rack-scale tier for exascale inference; prior stations remain in production. |
HGX B300 and GB300 NVL72 figures are indicative starting prices for additional infrastructure connected to an existing Super AI Station deployment, not a replacement for it.
Cost Comparison: Scalable On-Premises vs. Cloud AI APIs
For SMBs processing high volumes of documents on a daily basis, the difference between a fixed on-premises cluster and usage-based cloud AI billing compounds significantly over time.
| Cost Category | Cloud AI APIs | Supermicro (On-Premises) |
|---|---|---|
| Initial hardware investment | None | $100K (one-time) |
| Year 1 operating cost | ~€100k – ~€200k | ~€4k (power & maintenance) |
| Year 2 operating cost | ~€100k – ~€200k | ~€4k (power & maintenance) |
| Year 3 operating cost | ~€100k – ~€200k | ~€4k (power & maintenance) |
| Cumulative 3-year cost | ~€300k – ~€600k | ~₹1.09Cr |
Because processing takes place entirely on-site, documents and business data are not transmitted to third-party servers. This is a relevant consideration for organizations in legal, financial, or other data-sensitive sectors.
Summary
The Super AI Station GB300 offers access to local, trillion-parameter-capable AI at limited cost, able to grow station by station, and eventually into HGX B300 or GB300 NVL72 infrastructure. For SMEs with growing AI workloads, it can provide a lower cost alternative to cloud AI expenditure.
