The Next Big Thing

The “next big thing” is always a collection of things—technologies that come together at the right moment to produce a wildly popular new product at a time when the market can consume it, build on it and truly recognize and leverage its value.

What’s different about the Internet of Things is that, despite efforts to take control of it, there is no single owner, no company or even groups of companies with which it is most closely associated, and no one technology that has suddenly made it possible. It mixes older process technology with new technology at the latest process nodes, combines analog and digital in very predictable ways, and can be reshaped at any moment when there is a complication, either with software, different technology or even using different vendors’ parts and IP—meaning no vendor lock-in or apparent advantage.

The Internet itself falls into this camp, of course, except that it was created by a single entity—DARPA. Because DARPA didn’t want to own it, it has fostered a slew of commercial operations. In some ways the Internet of Things is an extension of that. Rather than just connecting people, it has reached the point of connecting things.

But there’s a difference in this case. The Internet is a bridge between mass communication and instantaneous communication that can be controlled by the user, in contrast to television, radio or even newspapers, which are controlled by big broadcasting or media companies. Communication between things on a mass scale has never existed. It’s new, uncharted territory. Moreover, unlike many new technologies, many of these things rely on older technologies as well as new technologies.

This has some interesting ramifications for the semiconductor supply chain. The key issues for most things in the IoT will be time to market and flexibility. Things are created horizontally but positioned vertically, and they actually serve to fragment markets rather than consolidate them. That means the emphasis will be on integration, programmability, and connectivity using very standard interfaces, standard IP, and programmable embedded software rather than design-from-scratch ASICs. Tools that focus on integration and verification of that integration will become very popular, even at older nodes.

Second, as a corollary to this, things in fragmented markets may have more price elasticity than chips controlled by a few very large companies in horizontal markets. They will be made in smaller batches, and profits will be split among more companies that are connected to those markets. Those profits, moreover, will likely extend well beyond just a chip or a thing, cascading into multiple new areas that have never existed before, either.

Third, while Moore’s Law will continue to be relevant for chips created specifically for mobile phones or powerful computers, the Internet of Things will provide a strong impetus for more generalized platforms and subsystems—and eventually for stacked die. How this shakes out on the manufacturing side, and where the big foundries decide to place their future investments as the IoT begins gaining steam will be interesting to watch. The battle between top-tier and second-tier foundries should get very interesting, and investment in capital equipment at older nodes could see a big resurgence.

As with the beginning of the Internet era, we are too early in this process to understand what will work and what will not, and where the biggest opportunities will be. But there are some interesting questions that need to be asked. What do things need most and what things will serve them most successfully and how? What makes one machine more successful than another, and what tools are needed to create that machine? This may sound ridiculous, but it’s probably going to decide who—or what—wins biggest in this new market.

—Ed Sperling