The Future of Embedded Chip Design Navigating the Chip Creation Space

Custom Silicon at Lower Cost, Reduced Development Time

- Michael Wishart

Empowering a new generation of developers to create more specialized, efficient and innovative chips opens paths for smaller enterprises to take on the challenges of machine learning, IoT, artificial intelligence and automation.

The proliferation of embedded systems has reshaped industries by providing intelligence to products, thereby improving ease of use, functionality and life cycle cost. Historically, embedded systems providers could meet most customers' requirements using standard, off-the-shelf components, providing differentiation through software. In certain cases, the embedded systems had particularly demanding performance, cost, or power requirements (or some combination of all three) and required tailored hardware in the form of customizable silicon solutions, most notably FPGAs or custom silicon such as ASICS.

We are now moving to the next generation of embedded systems, one powered by machine learning (ML). These new ML-enabled systems are expanding the applications served and the value that microelectronics provide in traditional embedded markets.

Importantly, they also bring the power of embedded intelligence to a wide array of existing and new consumer markets. In this new world, the benefit of tailored hardware becomes more broadly relevant, often making the difference between products that serve a market's needs and those that do not.

As such, enabling simple, fast and affordable customizing of chips has emerged as a critical differentiator among OEM's, software and service providers. Once again, FPGAs are one choice. They are relatively easy to implement by non-chip experts but are not energy efficient and do not integrate the extensive analog functionality necessary for cost-sensitive or sensor driven applications.

Alternatively, ASICS give engineers a wide