Lost: 19 Billion Connected Widgets

Just five years ago the pundits at Ericsson were predicting that there would be 20 billion industrial devices with built-in cellular connectivity by 2020.   Other pundits jumped on board with staggering numbers of their own — also in the tens of billions.  Granted, I do love reading these grandiose predictions.  They inspire me to imagine vast potential use cases, value propositions, and technical possibilities.

But then I filter it all against what I think I know about technology, the rate
of market adoption, and product development cycles.  I stop to consider whether the various markets really have the speed, the willingness or the budget to transition to alternate technologies.  I also ponder costs, and whether any of the altMike's Napkin Filterernate technologies is really one-size-fits-all.  When I applied my set of “Mike’s filters” to the predicted cellular adoption rates I had serious doubts.  So when Ericsson updated their estimates a few months ago, and the prediction dropped from 20 billion to 1 billion, I couldn’t help but feel a little vindicated.


As useful as cellular connectivity is, it’s not automatically the answer to every industrial networking question.  Cellular connections involve investments in data plans, routers and other expenses.  When wired connectivity is an option it can often be the better choice.  Furthermore, not every device needs its own connection.  That’s what network gateways are for.

Consider your car.  Unless it’s older than I am it can be expected to have somewhere between 30 and 80 sensors on board.  At various times you might like to know what they all have to say. But you don’t need 80 cellular data plans. You can do it through a single gateway port: wired or cellular.  Twenty billion connected sensors and devices won’t need 20 billion new cellular connections, even if their data is ultimately transmitted across the cellular networks.

Here at the B+B we know a few things about cellular. We’ve been connecting machines over cellular networks for years, and through three generations of cellular technology.  We also know a thing or two about connecting devices in other ways, even sensors over IP networks.  So when we look at the cellular option we don’t just think about obvious costs like data plans and cellular routers, we also consider things like the costs involved in provisioning and maintaining every sensor or connected device as a unique IP connection.

I suspect that the predictions about the IoT network of the future simply neglected to consider how often cellular connectivity isn’t the only solution, and how often our new sensors and devices will be networked.  Data will travel over wireless mesh networks, Wi-Fi bridges, fiber optic cable and copper Ethernet cable.  Most of it will aggregate through central gateways, or even networks of gateways. Some of the gateways will be cellular, but many will have wired Ethernet connections.  It’s not that we aren’t going to connect tens of billions of sensors and devices to our networks by 2020.  Of course we are.  But we won’t have to buy every single one of them its own cellular data plan.

I think Ericcson’s new estimate of 1 billion new cellular-enabled devices is a bit closer to the mark.  Is anyone placing bets? Talk back; I’d love to hear your comments.