QoM1608: The coming IOT tsunami or not

Saw an interesting chart the other day in a post in TechPinions (Searching for What’s Next) showing the sales in millions over time of PCs, Tablets and Smart Phones.  From the chart, PC sales peaked 2010-2012 and that Tablet sales have at flat lined (2016). Not sure what’s projections vs. actuals but the story on SmartPhones have yet to run out and they had rapid sales growth between 2008 and 2014.

The other thing to take from this chart is that device adoption is speeding up. It took 20 years to reach peak PC sales but it only took ~10 years to reach peak Smartphones sales.

IOT device forecast

Intel predicts that by 2020 there will be 26 IoT devices for every human on earth. According to UN, the estimated human population by 2020 is expected to be 7.7B, so that means, we will have 200B IOT devices then, not sure if this is peak sales but it’s quite a step up from today.

Intel’s infographic says the majority of these devices will not be home or on your person but rather in the business/manufacturing (40%), healthcare (30%), retail (8%), security (8%), and transportation (4%) sectors.

As for today’s (2015) IOT device sales, there are estimates that range from 4.6B to 10.0B to 15B to 18B. Some differences in these estimates are no doubt definitional (what constitutes an IOT device), but to be different by a factor of almost 4X, is a lot.

Let’s just use 10B for a moment. What will it take to reach 200B in 2020?  To start at 10B in 2015 and to reach 200B IOT devices in 2020, IOT device sales will need a CAGR of 86.2%. Even with device adoption speeding up, 86% CAGR seems high

My IOT forecast is different

But we are talking about factories, hospitals, stores, and cars & trucks. From my perspective these places, services and things just don’t modernize that fast. Trucks on the road are mostly ancient, cars in use today at least 5 years old, hospitals and healthcare offices may be somewhat quicker to adopt technology (if it has proven value) but factories take forever to adopt new technology.

Smartphone sales went from 122.3M in 2007 to 1.4B in 2015 for a CAGR of 35.9% which seems huge.  But these are people adopting fashion technology, with a ~2 year lifecycle. If anything, the entities adopting IOT have a much longer life cycle than phones. The people adopting the technology are also supposed to be smarter about this sort of decision making than the average person.

I cannot conceive of IoT device sales growing ~1.6 to ~2X faster than smartphones.  Will we get to 200B IOT devices being sold/year, maybe some day, will it get there by 2020, no way.

So a significant part of the sales of IOT devices are tied to the lifetime of the equipment/facilities that it’s used in. For the categories above (business/manufacturing (40%), healthcare (30%), retail (8%), security (8%), and transportation (4%) sectors) business/manufacturing equipment has more like a 10 year cycle, healthcare is mixed but I’ll give it a 7 year cycle, and retail, security and transport probably fit in the more like 5 year life cycle.

Farming view of IOT device sales

For example, take farms for instance. In 2007 there were 408.1M acres (16.5K 100 sq km) of cropland in the USA in ~2.1M farms (in 2012). According to another source ~210.4K tractors were sold in 2015 across the USA. There has been only modest growth in tractor sales ~3.8%/year, so if that growth holds up we project 2020 tractor sales to be ~253.5K.

In 2007 there were ~271.2 tractors/100 sq km in USA and that ratio has been flat for years. So that means there were ~4.5M tractors in the USA in 2015 and with 210.4K tractor sales/year that says that the average tractor lasts about 21.4 years on US farms.

Lets assume ~50 IOT devices across the farm/tractor (SWAG) on average. So backing into the 10B IOT devices sold in 2015 (40% for business/manufacturing, says 4B IOT devices). Tractors would account for 10.5M IOT devices in 2015. Farms contribute about 1% of USA GDP and if that ratio holds for IOT devices, this would say that in 2015, ~1.05B IOT devices were sold in the USA for GDP (business/manufacturing) activities. USA accounts for ~24.4% of world GDP so that gets us to 4.3B IOT devices sold for business/manufacturing activities in the world for 2015 – close enough to 4B.

So the worldwide multiplier of IOT sales per tractor is (50 * 100 * 1/0.4 * 1/0.244 = 51,229.5). For an estimated 2020 tractor sales of 253.5K, we would estimate 13.0B IOT device would be sold worldwide.

What do we need to adjust

How could we be so short of the 200B Intel’s projecting.

• The amount of IOT devices used across the farm/tractor is not the best ratio to use for worldwide IOT device sales.
• The ratio of worldwide IOT device sales/tractor sales will not remain flat and should increase over time. More IOT devices will be used across the farm per tractor over time as IOT hardware and software are developed. Lets say this goes up by 5X in 4 years to 2020.
• The growth rate in tractor sales is not representative of other business/manufacturing sales.
  • Number of MRI units in operation in USA grew 2.1X from 1995 to 2007 (5.87% GAGR)
  • Cars sold in USA were 17.5M units, up 5.7% from prior year.
  • Truck sales in USA from now until 2024 are expected to grow 3% per year.
  • Fab equipment sales is expected to go up 13.2% between 2016 and 2017.

So lets use a 6% sales unit growth rate as more typical throughout industry.  This would say the tractor “equivalent” sales in 2020, which would be 281.6K for 14.4B IOT devices at current rat. But when we add the 5X multiplier, thats says 72.1B IOT device worldwide sales/year for 2020.

So my forecast for 2020 worldwide IOT device sales is 72.1B.

I would give this an 0.7 probability. The 6% growth rate seems ok. The 5x multiplier seems about right for 4 years. But I can’t help but feel I am missing something else.

Photo Credit(s): Searching for what’s next article and Jefferb at Pixabay