Specifications
Model
DX-AIPlayer N97 (ordering code DX-AIBOX-M1-N97)
SoC
Intel Processor N97 — quad-core up to 3.6 GHz, 6 MB cache, 24 EU Intel UHD @ 1.20 GHz, 12 W TDP
AI Accelerator
DEEPX DX-M1 M.2 2280 M-Key module — 25 TOPS INT8, 4 GB dedicated LPDDR5, 1 Gbit QSPI NAND, 5 W max TDP
NPU Power
1 to 5 W (accelerator only)
System Memory
8 GB LPDDR5 (up to 16 GB supported)
Storage
64 GB onboard eMMC (up to 128 GB supported)
Networking
2x Gigabit Ethernet RJ45
USB
3x USB 3.2 Gen 2 Type-A, 2x USB 2.0 via internal 10-pin header
Expansion 1
M.2 2280 M-Key (PCIe Gen 3 x2) — occupied by the DX-M1 accelerator
Expansion 2
M.2 2230 E-Key (PCIe x1, USB 2.0) — Wi-Fi / Bluetooth
Power
12 V DC-in 5 A via threaded locking barrel jack
Weight
450 g net / 600 g gross
Operating Temperature
0 °C to 60 °C (0.5 m/s airflow)
Storage Temperature
-20 °C to 70 °C
Certification
CE / FCC Class A, RoHS, REACH
OS Support
Windows 10/11, Ubuntu 20.04 / 22.04 / 24.04 LTS, Yocto Project 5.1
SDK
DXNN (DEEPX Neural Network) SDK plus DX-AllSuite compile/optimise/run toolchain, local or Docker
Frameworks
PyTorch and other common frameworks via the DX-AllSuite toolchain
Target Workloads
Real-time vision AI — robotics, smart cities, factory automation
Reference Price
USD 995.01 for the 8 GB version on DigiKey (out of stock at time of writing; est. ship 17 July 2026)
Overview
The DEEPX DX-AIPlayer N97 is a compact edge AI mini PC that pairs an Intel Processor N97 Alder Lake-N SoC with the company's own DX-M1 M.2 2280 NPU module mounted in a PCIe Gen 3 x4 M-Key slot. The NPU delivers up to 25 TOPS of INT8 inference while drawing only 1 to 5 W, and carries 4 GB of dedicated LPDDR5 plus 1 Gbit QSPI NAND so model weights do not consume system memory.
That separation is the point of the design: the x86 host handles the operating system, camera capture and application logic, while the DX-M1 runs the vision model at a fraction of the power a GPU would need. The system ships with 8 GB LPDDR5 and 64 GB eMMC (both upgradeable), dual Gigabit Ethernet and a threaded locking 12 V barrel jack — hardware-clearly intended for unattended industrial deployment.
Software support covers Windows 10/11, Ubuntu LTS releases and the Yocto Project, with the DXNN SDK and DX-AllSuite package providing model compilation, optimisation and local or containerised execution. DEEPX states the platform is designed for real-time vision AI in robotics, smart cities and factory automation.
Key Benefits
25 TOPS at 1-5 W — a dedicated NPU with its own 4 GB LPDDR5, so inference does not compete with the host for memory bandwidth. x86 host plus NPU accelerator keeps standard Linux/Windows tooling while offloading the model. Threaded locking DC jack and dual GbE for unattended industrial mounting. DXNN SDK with Docker support shortens model-deployment time.
Applications
Real-time video analytics at the industrial edge, robotics perception, smart-city camera nodes, factory-automation inspection, retail and traffic vision systems.
Request a Quote — DEEPX DX-AIPLAYER N97 — INTEL N97 MINI PC WITH 25 TOPS DX-M1 NPU
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →