On-device inference, computer vision, and ML acceleration. Run models at the edge without cloud dependencies — from Coral TPUs to Jetson GPUs to Kendryte NPUs.
The boards we'd recommend first if someone walked up and said “I want to build a ai & edge computingproject.”
Edge-AI boards ship with one of a half-dozen accelerator families — each with its own quantization story, framework support, and thermal envelope. UBDS catalogs accelerator name, TOPS rating, supported runtimes, and the exact model formats that will actually run.