The scalability of edge AI

When developing a product using an embedded microcontroller or microprocessor, it’s always important to consider how the product may evolve and scale over time. Engineers don’t want to spend months developing a solution on one microprocessor and then have to start from scratch when they update their product to a higher-performance processor.

Semiconductor manufacturers that create these embedded devices need to develop portfolios with scalability in terms of features, performance and cost. This approach helps ensure that there is a seamless migration strategy between their various embedded processors for AI, in order to make it as simple as possible for developers to reuse their work across different devices.

Edge AI is no exception. For example, a designer making a home robot may want to produce both a high-end version with three cameras for surround vision and an entry-level version that only has a single front camera. A scalable portfolio of edge AI-accelerated devices enables the porting of software from the high-end model to the entry-level model, minimizing the amount of resources needed to produce both products. Scalability also allows developers to transfer their R&D investment from one platform to the next as their product evolves.