The Cyber-Industrial Revolution

Semiconductors won’t save the world, but they certainly will help. In fact, it’s arguable whether any significant progress will be made on such issues as global warming or future medical breakthroughs without the aid of ICs.

After decades of struggling just to get chips to work at each new process node, the semiconductor industry is moving into a new phase. Processing is now almost ubiquitous. Compute resources can be dialed up and down as a service, and for most compute jobs they can be done locally. As the edge is built out, it will add yet another layer of compute infrastructure.

That will be essential, because the missing piece is increased functionality at the edge. In the past, control of mechanical systems was highly centralized, either inside of companies or in the cloud. In the future, more of that will be will be moved closer to the source because these edge systems will generate enormous amounts of data, and the cost of moving that data is too high to make it useful. So for industrial and corporate applications, much of that will likely be processed at the edge, with some oversight functions at a higher, such as an enterprise server.

The cloud will continue to play an important role, but probably not the same one. Cloud data centers will become increasingly vital for massive calculations, such as finding patterns and anomalies in weather over the past century, or achieving drug/vaccine discovery in record-breaking time. These operations also are a way of adding burst capabilities to chip/system verification and debugging, and that will continue. The cloud has almost unlimited compute resources to run simulations, and as quantum computers are added into the mix, there will be no upper limit to what is possible.

Still, it’s the localization and customization of computing that will drive the biggest changes. Companies across a variety of markets recognize that more sensors, better data, and more analysis will be required to optimize operations. In agriculture, rather than crop-dusting from a plane or irrigating entire fields, semi-autonomous farm equipment will identify which plants need attention and precisely apply whatever is necessary. And in automotive, chips already are being used to extend the range of electric and gasoline vehicles, identifying problems with increasing precision and helping to improve reliability.

On the manufacturing side, AI systems that are built using highly specialized accelerators and algorithms will be able to predict problems earlier with pattern recognition. The higher the yield, the less waste, and the more likelihood that parts will perform as expected throughout their lifetimes. And as technology lifetimes are extended, that likewise will improve overall utilization and efficiency, and it will greatly improve the carbon footprint of all technology, including chips.

Viewed individually, these changes don’t look particularly impressive. Reducing the number of RMAs or lowering a water bill may not seem that significant. But looked at from a broader vantage point, they add up to something much bigger. Mechanical technologies changed the world, starting in the late 1800s. Chips, and the growing amount of data they can process, will make all of those systems more efficient and effective, making it possible for the first time to maximize profitability across a wide range of markets by leveraging more environmentally friendly practices.