en.Wedoany.com Reported - On June 2, South Korean low-power AI semiconductor company DEEPX and Taiwan's industrial computing platform provider AAEON Technology signed a three-year mass production memorandum of understanding during COMPUTEX TAIPEI 2026. The two parties will integrate DEEPX's neural processing units into AAEON's product lines of industrial computers, single-board computers, and edge gateways, targeting commercial mass production for scenarios such as smart factories, robotics, smart cities, and intelligent transportation.

This collaboration moves edge AI chips from prototype demonstrations to a standardized delivery system for industrial hardware. Under the arrangement, DEEPX will provide AI semiconductor chip sets, dedicated compilers, and SDK support, while AAEON will handle hardware design, board form factor matching, and product line development. The two parties plan to establish a mass production system for AI acceleration modules covering standard form factors such as M.2, mPCIe, PCIe cards, and COM Express boards, and develop edge AI products for customer projects based on OEM/ODM requirements. For the industrial computing market, whether low-power NPUs can enter the mass production catalogs of mature hardware vendors is closer to real commercial validation than simply releasing chip specifications. AAEON has long served industrial PC, embedded platform, edge computing gateway, and industrial IoT customers, and its channels and product lines can allow DEEPX chips to directly reach equipment manufacturers, system integrators, and industry projects, rather than remaining at the development board or pilot verification stage.

The timing of the collaboration also aligns with DEEPX's previous commercialization progress. The platform combining AAEON hardware and DEEPX semiconductors received initial pre-orders after completing official certification in December 2025, and mass production verification by key customers has also driven related products from verification orders to commercial batch production.

Edge AI is extending from consumer-level smart terminals to factories, robotics, transportation, and urban infrastructure. Compared with cloud-dependent inference models, device-side AI computing emphasizes low latency, low power consumption, offline operation, data localization, and system stability. These requirements dictate that chips must be co-designed with industrial-grade hardware, thermal management structures, long-term supply cycles, and field deployment environments. The focus of this collaboration between DEEPX and AAEON is to embed NPUs into existing hardware categories such as industrial computers, SBCs, and edge gateways, enabling machine vision, AMR/AGV autonomous mobile robots, smart transportation perception, edge security, medical devices, retail terminals, and smart agricultural equipment to perform some AI inference tasks locally. For industrial customers, this approach can reduce cloud bandwidth and data backhaul pressure, as well as decrease reliance on high-power GPU equipment at project sites.

AAEON is backed by the ASUS Group's industrial hardware ecosystem, while DEEPX enters the Physical AI market with its low-power NPU architecture and DXNN SDK. If the two parties can package chips, boards, drivers, development tools, and industry software interfaces into a stable solution, the subsequent competitive focus will shift from "chips can run models" to "whether the complete system can be continuously delivered, whether the system can be maintained long-term, and whether customers can quickly adopt it." Such collaborations also indicate that edge AI chip companies are accelerating their ties with industrial hardware vendors, shortening the distance from chip design to industry project implementation through channels, certifications, and mass production systems.

Subsequent project variables focus on three aspects: first, the shipment pace of AAEON's product lines equipped with DEEPX NPUs; second, the stability of modules such as M.2, mPCIe, PCIe cards, and COM Express in different industrial scenarios; and third, whether the two parties can generate sustained orders beyond international exhibitions like COMPUTEX, CES, and Embedded World. If the mass production system progresses smoothly, edge AI chips will enter industrial equipment more quickly, becoming the underlying computing units for smart manufacturing, robotics, and urban perception systems.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com