DEEPX Builds Physical AI Deployment Environment With AWS IoT, Expands Global Reach
DEEPX linked its NPUs with AWS IoT Greengrass to build a physical AI deployment environment and listed the DEEPX Greengrass Solution on AWS Marketplace.
- 'DEEPX Greengrass Solution' listed on AWS Marketplace, giving global customers a deployment and operations environment
- Cloud-to-edge operations covering everything from AI model conversion to remote deployment of the NPU runtime
- CEO Lokwon Kim attends AWS Startup Partner Summit to expand cooperation with global partners

Sept. 1, 2026 — DEEPX, a developer of ultra-low-power physical AI semiconductors led by CEO Lokwon Kim, is expanding globally with a physical AI deployment environment that links its NPUs with the cloud and IoT infrastructure of Amazon Web Services (AWS).
On Aug. 27, AWS published a post on its official technical blog showing how to combine DEEPX's DX-M1 NPU with AWS IoT Core and AWS IoT Greengrass to prepare AI models and deploy and manage the NPU runtime environment across many edge devices.
With the DEEPX Greengrass Solution, now available on AWS Marketplace, users can convert ONNX models — a standard format that lets models built in different AI development environments be moved and reused — into DXNN, the format that runs on DEEPX NPUs. They can then use AWS IoT Greengrass to remotely deploy NPU drivers, firmware and the runtime, the software environment that lets AI models run on field equipment. This makes it possible to apply the same AI runtime environment to large fleets of equipment in factories, logistics centers, stores and robots, and to keep software and models updated without visiting each site.
As physical AI spreads to robots, cameras and industrial equipment, a cloud-to-edge architecture is becoming more important. In this setup, AI models are developed and managed in the cloud, while low-power NPUs — chips specialized for running AI models' real-time inference — handle inference on edge devices such as cameras, robots and industrial PCs that run AI close to where data is generated. DEEPX plans to develop this into a "center-edge federated AI" architecture in which the cloud and the edge divide roles to develop, deploy and operate AI.

Meanwhile, DEEPX CEO Lokwon Kim is attending the AWS Startup Partner Summit 2026, held Aug. 31 to Sept. 1 in San Francisco. His attendance is meant to go beyond the technical integration of DX-M1 and AWS IoT Greengrass already achieved, widening DEEPX's contact with AWS's global partner ecosystem and turning it into customer projects and business opportunities in the U.S. and other markets. It is unusual for a semiconductor company — the hardware foundation for running physical AI — to take part in AWS's startup partner ecosystem, which is centered on cloud and software companies, and the summit is expected to open new opportunities for collaboration linking the cloud with edge hardware.
DEEPX plans to extend the cloud-to-edge deployment architecture it built on DX-M1 to customer projects in U.S. robotics, industrial vision and smart factories, and is also considering applying it to DX-M2, its next-generation generative AI chip.
"For physical AI to take off, we need not only low-power AI chips but also infrastructure that makes it easy to deploy AI developed in the cloud to countless field devices and manage it continuously. By connecting AWS's cloud and IoT ecosystem with DEEPX's ultra-low-power AI chip technology, we will broaden the foundation for physical AI to spread into real industrial settings." — A DEEPX official



