The internet of things is quickly achieving the same levels of froth and excitement as big data in the venture and entrepreneurial community. And like “big data,” the prevalence of smartphones, cheap computing and connectivity all are combining into a substantial and real opportunity under all the hype. So instead of yet another smart light bulb or connected hub (yes, I love those too) let’s dig a little deeper where the internet of things is already changing the fortunes of several large companies.
To build the internet of things we’re going to need a lot of chips — orders of magnitude more than we have in use today. Generally those chips will fall into three categories, and each of those categories is poised to become a booming business with a lot of volume and room to grow. Let’s break it down:
Connectivity: This one is a no-brainer. If we want things to connect to the internet, we’re going to have to put radios in them. It may be Wi-Fi, Bluetooth, Z-wave, ZigBee or even a 3G or 4G cellular standard (or all of the above) but there has been and will continue to be a land grab for radios among the big chip companies. The rise of connected devices is the reason Qualcomm bought Atheros back in 2011 and the reason little known microcontroller company Atmel purchased Ozmo Devices in December.
We’ll also see new products aimed at integrating radios together, not just from Broadcom — the king of radio integration — but also smaller companies such as Redpine Signals, Altair and others. And these radios will be going into more devices. Just a quick scan of Kickstarter or Indiegogo shows a plethora of home gateways, Wi-Fi enabled devices and sensors that have radios integrated from a variety of vendors. A report from the OECD on the internet of things estimates that a family of four will go from having an average of 10 devices connected to the internet now to 25 in 2017 and 50 by 2022. Every single on of those will have a radio — or multiple radios.
Control: These chips are the brains of the operation. But unlike in the personal computer or server market, where Intel and AMD fought for dominance (more truthfully, AMD tried to at least achieve profitability), or the smartphone market where Qualcomm has taken out competitors ranging from Texas Instruments and Freescale on the application processor side (leaving Apple, Samsung, Broadcom and Mediatek standing), this market has a much wider variety of players known for their embedded processors and microcontroller. The one name that spans all of these industries is ARM (s ARMH).
At the low end, microcontrollers can range from 8-bit processors that manage setting on your microwave to higher-end chips inside a set-top box. Companies like Freescale, Texas Instruments, Atmel, Intel and STMicroelectronics all are pushing their microcontrollers (MCUs) inside the internet of things. The variety of use cases and devices inside connected devices mean some gadgets will need more power savings than performance or merely just a cheap 32-bit chips designed for a more industrial application. Many of these companies have an advantage for the internet of things because they are used to supporting a wide variety of end products with their firmware and sales teams.
They have designed their chips to be modular. If the bigger players want to play here they will have to build out multiple lines of chips with differing performance specs that can be supported across a wide range of end devices. That’s very different from building out a line of chips with slightly different specs all designed for servers. I bet a few of the big vendors, especially on the connectivity side, might try to acquire this knowledge.
And since many of these sensors will be integrated onto small packages with radios and maybe even MCUs there will be a lot of value for a company that can pop all of the above onto a system on a chip — it’s cheaper, smaller and more power efficient. So consolidation will happen within these categories as well as across them as more devices get online and we ask them to share more information about their environment.
So be they MEMs, microcontrollers or radios, there’s a lot of silicon (or maybe gallium arsenide) inside the internet of things. And the types of chips required will stretch the silicon industry — that has been primarily focused on keeping up with the performance requirements of Moore’s Law — into new directions. Power savings, integration and size will matter when it comes to connected devices more so than the all out race for performance that has dominated the chip industry for decades.