NVIDIA RTX PRO Blackwell in MXM 3.1 form factor · up to 24GB GDDR7 · up to 7,680 CUDA cores · for LLM & data-intensive edge AI
Flagship embedded MXM module powered by NVIDIA RTX PRO 5000 Blackwell for the most demanding edge AI and graphics workloads.
High-performance embedded MXM module powered by NVIDIA RTX PRO 4000 Blackwell — the sweet spot for edge AI inference and rendering.
Mid-range embedded MXM module powered by NVIDIA RTX PRO 2000 Blackwell for compact, power-efficient edge AI systems.
Entry embedded MXM module powered by NVIDIA RTX PRO 500 Blackwell for cost-optimized edge AI and multi-display deployments.
ADLINK Technology
NVIDIA Blackwell
MXM 3.1 Type A & Type B
PCIe Gen5 x16
(2× Gen4 bandwidth)
Up to 24GB GDDR7
7,680 (BW4000)
38.7 TFLOPS (BW4000)
240 Tensor · 60 RT (BW4000)
Ada · Ampere · Turing · Intel Arc
LLM & data-intensive edge AI
Available
Quote
ADLINK's EGX-MXM Blackwell family brings NVIDIA's RTX PRO Blackwell architecture to the compact MXM 3.1 embedded form factor, delivering up to 24GB of high-speed GDDR7 memory and PCIe Gen5 (2× the bandwidth of Gen4) for large language model (LLM) and data-intensive edge AI workloads. Spanning four models — BW5000 (RTX PRO 5000 Blackwell), BW4000 (RTX PRO 4000 Blackwell), BW2000 (RTX PRO 2000 Blackwell), and BW500 (RTX PRO 500 Blackwell) — the lineup scales from flagship 7,680-CUDA-core modules down to cost-optimized entry options in Type A and Type B form factors.
Announced March 2026, the Blackwell MXM modules are purpose-built for space-constrained embedded systems across healthcare, manufacturing, transportation, aerospace, and retail, where a standard PCIe graphics card won't fit. The modules complement ADLINK's existing MXM portfolio — Ada Lovelace (AD5000/AD3500/AD2000), Ampere (A4500/A2000/A1000/A500), and Turing generations, plus Intel Arc MXM-AXe — giving OEMs an industrial-grade, long-lifecycle GPU upgrade path. The BW4000 delivers 7,680 CUDA cores, 240 Tensor cores, 60 RT cores, and 38.7 TFLOPS peak FP32, enabling on-device AI inferencing, machine learning, and accelerated graphics directly at the edge.
Need ADLINK EGX-MXM Blackwell modules?
Contact QS Compute for availability, configuration, and volume pricing on embedded MXM GPU modules.
Request Quote