I found five edge AI servers built on NVIDIA IGX Orin, from three NVIDIA partners: YUAN High-Tech, Advantech and MBX Systems (AHEAD).
IGX Orin edge AI servers yuan.com.tw +4yuan.com.tw ↗“NVIDIA IGX Orin and RTX 6000 Ada, up to 1705 TOPS, 2x 100Gb QSFP28 + 2x GbE”advdownload.advantech.com ↗“Industrial short-depth 4U rackmount edge AI system based on NVIDIA IGX Orin (64 GB)”advantech.com ↗“Medical-grade small-tower edge AI system based on NVIDIA IGX Orin”ahead.com ↗“2U edge AI server platform from MBX Systems (AHEAD) powered by NVIDIA IGX Orin”ahead.com ↗“4U edge AI server platform from MBX Systems (AHEAD) powered by NVIDIA IGX Orin”
| Product | Manufacturer | Form factor | Max AI performance (INT8, sparse) | Highlights |
|---|---|---|---|---|
| VPP6N0 IGX 700 | YUAN High-Tech | Edge AI workstation/server | Up to 1705 TOPS (with RTX 6000 Ada) | 2x 100Gb QSFP28 + 2x GbE, PCIe Gen5, 850 W |
| MIC-735I-IO | Advantech | Industrial short-depth 4U rackmount | 248 TOPS | Up to 25GbE, 2x PCIe slots (Gen5) |
| MIC-735M-IO | Advantech | Medical-grade small tower | 248 TOPS | PCIe x16 slot for an optional discrete GPU card |
| Optio L100 | MBX Systems (AHEAD) | 2U | 248 TOPS | 64GB LPDDR5, 2x QSFP ConnectX-7 100GbE, 4x 2.5" drive bays |
| Optio L150 | MBX Systems (AHEAD) | 4U | 248 TOPS | 64GB LPDDR5, 2x QSFP ConnectX-7 100GbE, room for up to two double-width GPGPUs |
Things to know:
- Advantech marks the MIC-735I-IO datasheet as preliminary. advdownloa...advdownload.advantech.com ↗“Advantech datasheet is marked Preliminary”
- The VPP6N0 IGX 700's 1705 TOPS is for the configuration that includes the RTX 6000 Ada GPU. The other four list the 248 TOPS baseline.
- A related system: the Aetina AIE-MIX640 is also built on IGX Orin, but it's listed as an IGX Edge AI System rather than a server. Aetina gives a 248 TOPS baseline and an optional RTX 6000 Ada configuration of up to 1,705 TOPS, aimed at industrial and medical use. aetina.comaetina.com ↗“Aetina AIE-MIX640 industrial edge AI inference system built on NVIDIA IGX Orin”