Hardware design, driver development, firmware, FPGA and SDK work all happen under one roof. The team that lays out a board also writes the driver that runs it — which is what integrated support actually means when a deployment goes sideways.

About YUAN
YUAN High-Tech has spent more than three decades moving video into computers — and, lately, teaching computers what to do with it. Founded in Taipei in 1990 and listed on the Taipei Exchange as 5474, the company has grown from a maker of capture cards into a platform partner for industrial vision computing.

YUAN was founded in Taipei in 1990, in the era when getting a video signal into a computer was itself the hard problem.
The company built its name on image capture — the boards that turn SDI and HDMI feeds into data a machine can use — and became a globally leading provider of image capture modules along the way. Thirty-plus years on, the four verbs of the business have not changed: acquisition, conversion, streaming and analysis.
The heritage shows in the catalogue. YUAN today ships 372 products across 45 series and five families — PCIe and USB capture cards from HD to 8K60, 12G-SDI and HDMI 2.0 converters, and AV-over-IP codecs speaking SDVoE, NDI, Dante AV and SRT. Behind them sits a full-stack R&D team that designs the hardware and writes the drivers, firmware, FPGA logic and SDKs itself — which is why the same boards turn up in broadcast galleries, operating rooms and roadside cabinets.
The pivot came when capture met inference. As a strategic partner of NVIDIA, YUAN rebuilt its platforms around Jetson: VPP systems that put up to 275 TOPS of AI compute (VPP6N0-MB) beside the capture pipeline, the Pandora developer kits, and the NexVDO SDK that carries a frame from ingest to analysis with microseconds of overhead. Video stopped being something to record and became something to understand — across nine industries, from smart cities to precision agriculture.
That partnership now spans NVIDIA's full edge line — Jetson Orin and Jetson Thor, IGX, and the Holoscan Sensor Bridge — and reaches into the field through collaborations with Foxconn on smart healthcare and with National Taiwan University Hospital on AI-assisted ultrasound screening. The ambition is stated plainly: to be the next-generation leader in industrial vision computing, and the shortest path from NVIDIA silicon to a deployed product.
Milestones
2025
The EYE-I AI camera is honoured in the Best Smart Manufacturing category at the 2025 EE Awards.
2026
YUAN launches its Holoscan Sensor Bridge line, feeding synchronised high-bandwidth sensor streams directly into NVIDIA edge AI pipelines for ultra-low-latency inference.
2026
With National Taiwan University Hospital and the Good Liver Foundation, YUAN deploys 50 iGood Liver AI ultrasound systems to rural and underserved clinics — Jetson-powered screening as a second pair of eyes for frontline physicians.
2026
With production platforms already shipping on Jetson AGX Thor T5000, YUAN announces development on the new Jetson T3000 and T2000 modules — one Blackwell-architecture software stack, every deployment scale.
2 / 2
Values
Hardware design, driver development, firmware, FPGA and SDK work all happen under one roof. The team that lays out a board also writes the driver that runs it — which is what integrated support actually means when a deployment goes sideways.

Awards

From capture silicon to Jetson inference — if the work described here sounds like yours, talk to the YUAN USA product team.
Contact YUAN USA