In recent years, geopolitics and Covid have brought the chip industry into the limelight. From the very apparent supply-chain-related shortages to the extreme dependence on technology platforms just to continue living our modern lives, governments and mainstream audiences have finally understood that “tech” doesn’t just mean social media and e-commerce platforms, and that semiconductors are “a thing.”
As a result, the chip industry has escaped the claws of regulation for the last 50 years or so. But that is rapidly changing, as a recent conversation with Silicon Labs CTO Daniel Cooley highlighted. We caught up with him at the embedded world North America conference and exhibition, just after he and Silicon Labs President and CEO Matt Johnson gave the opening keynote there in Austin, Texas.
Cooley highlighted both sides of the argument as far as regulation is concerned, with three specific areas of focus: privacy, security and geopolitics.
We also chatted about how the embedded industry has evolved over the last five years and how embedded systems are doing a lot more than five years ago. Because silicon takes two to three years to emerge in end products, it takes time to see innovations in silicon and software surface in the market.
And what about the challenges customers are facing? Cooley said that wireless is now an accepted feature in everyday applications. But customers are feeling the weight of code, so conversion from bare metal to an OS, such as Zephyr and others, is becoming more common and they need to speed up code. The other two areas that customers are now looking to address are security and privacy. On the latter, he said privacy and AI are linked: When you’re trying to do AI on products, you want to ensure you can do as much of it as possible locally to ensure privacy.
We touch on the very different aspects of two key markets: India and China.
Bringing ‘smart’ to IoT and the far edge
In their keynote address at embedded world North America, Johnson and Cooley talked about bringing more processing power to the far edge and making IoT devices smarter.
“AI is rapidly becoming a key growth catalyst that will enable the number of IoT devices to reach over 100 billion in the next decade,” Johnson said. To address this, Silicon Labs’ upcoming Series 3 platform would provide the connectivity, compute and security needed to “unlock new applications and new capabilities” across a vast range of industries. Johnson and Cooley listed manufacturing and retail, transportation, healthcare, energy distribution, fitness and agriculture as examples.
In addition, select Series 3 devices would feature Silicon Labs’ second-generation matrix-vector processor, which offloads complex machine-learning operations off of the main CPU onto a specialized accelerator designed to increase ML performance by up to 100× in wireless, battery-powered devices to drastically reduce power draw.
Data flows between edge devices to the cloud and back again are an important part of enabling this, Johnson and Cooley said in a summary of the keynote. “This two-directional flow makes IoT edge devices an ideal partner in the growing world of AI. Not only can the devices be used to make limited decisions at the edge, like smart thermostats that assess ambient temperatures and make adjustments to a home’s HVAC, but also with the advent of massive cloud-based AI applications, edge devices can also play a critical role as data collectors and apply their ML abilities to filter the valuable data from the chaff better. The ability to identify and transmit the data for the ‘corner cases’ that AI operators value to make their systems more intelligent.”
Silicon Labs added that its first Series 3 SoC is currently being sampled by customers and that more information will be revealed in the first half of 2025.
----Form EE Times