Playing With Bubbles

The first is in the social media world, where there simply are too many sites for a fixed population of users. Much has been written about this trend, including in this blog, and the big question is whether there is enough new business in new markets such as the IoT to soften the landing.

The IoT has a long way to go, and a long way to grow. That growth will propel everything from the design and development of new servers at the edge of the network all the way to the cloud, which is where the real crossover point is for social media. From a purely business standpoint, if there is enough need for data storage and processing, then at least in theory a drop in the social media world will be absorbed by IoT demand.

But there’s another problem creeping in here that could have more far-reaching consequences. There is not enough capacity at older process nodes, not enough equipment being made for those nodes, and not enough of a business case to push design and manufacturing of sensors and mixed-signal devices down to 28nm or beyond where there is plenty of capacity.

This is showing up in increasing lead times from sign-off to first silicon. And it’s showing up in congestion in the mask shop at foundries. Miss a window on the shuttle run and you might have to wait weeks more for your chip to be manufactured. You might even be penalized for it as demand continues to grow and capacity stays flat.

The semiconductor industry witnessed a similar set of circumstances in 1999 and 2000, just before the dot-com bubble burst, as companies placed redundant orders as a way of assuring they could get their products to market on time. At that time the constraints involved communications servers, cell phones, and Internet access devices.

Many more devices have flooded onto the market since then. The ubiquitous smart phone can trace its roots to the personal digital assistant introduced in the mid-1990s, but it really gained solid footing in its current form starting in 2010. Since then, more and more companies have built devices that are compatible with phones based upon Apple’s iOS, Google’s Android, ARM’s processor cores, and to a much lesser extent Windows and Intel chips.

What’s worth keeping track of here is that even though the main players have been around for awhile, many of the players that have benefited from the smart phone evolution or revolution haven’t been through a bubble or a downturn. With the IoT, that percentage drops even further. They don’t necessarily understand the market dynamics of constrained manufacturing or the discipline of managing a business through these kinds of cycles. And they don’t have any of the history that says you don’t double-book or triple-book orders because that pushes the entire supply chain out of whack.

This is a hard enough discipline to master by companies that have survived previous cycles and have the burn scars to prove it. It’s particularly difficult for a startup that has no reserve capital, a limited window for market success, and few resources with which to plan for these kinds of events.