Specifications

Architecture

NVIDIA GB10 Grace Blackwell Superchip (Dell Pro Max platform)

AI Performance

1 PFLOP FP4 - 1,000 AI TOPS

CPU

20-core Arm (10x Cortex-X925 + 10x Cortex-A725)

GPU

NVIDIA Blackwell GPU with 5th-gen Tensor Cores and 4th-gen RT Cores

CPU-GPU Link

NVLink-C2C unified CPU-GPU memory, up to 6x PCIe Gen5 bandwidth

Memory

128 GB LPDDR5X unified, 256-bit, 273 GB/s

Networking

Dual ConnectX-7 SmartNIC; 10/25/40/50/100/200/400GbE; up to 400 Gbps

Wireless

Wi-Fi 7, Bluetooth 5.4

Storage

Up to 4 TB M.2 NVMe with built-in self-encryption

Power Input

18-36 V DC-IN (optional MIL-STD-461 EMI filter)

OS / Stack

NVIDIA DGX OS 7.3.1, GPU driver 580.95.05, CUDA Toolkit 13.0.2

Clustering

Up to 3+ nodes connected via D38999 MPO 200Gbps cables for distributed training

Ruggedization

MIL-STD-810 and MIL-STD-461 compliant; DTL-38999 connectors

Ingress Protection

IP65 sealed compact chassis

Dimensions

250 x 250 x 138 mm; approx. 5 kg

Cooling

Conduction cooling with smart fan; corrosion-resistant aluminium chassis

Models

DGX10-R (industrial rugged), DGX10-M (military)

Overview

The 7STARLAKE DGX10 is a ruggedized AI computer built on the NVIDIA GB10 Grace Blackwell Superchip and Dell Pro Max architecture. It delivers 1 PFLOP of FP4 AI performance from a 20-core Arm CPU and Blackwell GPU with 5th-generation Tensor Cores, backed by 128 GB of coherent unified LPDDR5X memory at 273 GB/s.

Designed for defense, aerospace and industrial edge deployments, the DGX10 uses NVLink-C2C for up to 6x the bandwidth of PCIe Gen5, dual ConnectX-7 SmartNICs for up to 400 Gbps networking, and up to 4 TB of self-encrypting NVMe. It meets MIL-STD-810 and MIL-STD-461 with D38999 connectors and an IP65 chassis, and multiple units can be clustered over MPO 200Gbps cables for distributed training.

Key Benefits

1 PFLOP FP4 in a rugged IP65 chassis; 128 GB unified memory and NVLink-C2C for large-model inference at the tactical edge; MIL-STD-810 / 461 compliance and D38999 connectors for defense deployments, plus multi-node clustering.

Applications

Forward-deployed AI inference, ISR data processing, autonomous systems control, tactical decision support, electronic-warfare analytics and small-scale distributed model training.

Request a Quote — 7STARLAKE DGX10 — RUGGED GB10 GRACE BLACKWELL AI COMPUTER

QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.

Get Your Quote →

Related Products

7STARLAKE AV600-MH-A50 Rugged Edge AI Server NVIDIA DGX B300 ASUS Ascent GX10 Compact AI Supercomputer