eSIM technology is being over-hyped for IoT

The last couple of weeks - especially with MWC - have seen lots of noise around embedded SIMs (eSIMs). In particular, the GSMA announced its remote provisioning standard (link). 

While interesting and a step in the right direction, I think the industry is over-hyping the potential of eSIMs.

eSIMs are still SIMs, but they are built-into devices as fixed hardware components (basically an extra chip soldered-in), rather than as traditional removable cards. They can be remotely-programmed to support different operators' profiles, or switch between them.

This development gets around some of the more awkward practicalities of physical SIM cards in non-phone devices:

• Physical space & design constraints needed for SIM slot & removable tray
• Vulnerability to vibration and dust by having a tray/slot
• Need to get SIM cards into the devices' normal distribution channels & retail stores
• Potential need for user to source a SIM separately in a different purchase
• Difficulty for user to swap operators (especially if device is locked to a particular network)

These were some of the problems which stopped widespread adoption of cellular radios and SIM cards in laptops, most tablets and other devices. (I wrote about this a lot in 2006-2008, eg here & here)

In that sense, eSIM is definitely a step forward. More use-cases become practical for cellular connectivity, just at the time when M2M/IoT is finally taking off. However, it would be wrong to assume this means that 4G-connected consumer devices will become the norm. While some categories (eg cars) are widely adopting cellular radios, others (eg wearables, home electrical appliances) are not.

The problem? Cost.

A 4G radio module and a SIM/eSIM remains a significant extra component on the per-unit BoM (bill of materials) cost for a manufacturer, plus the costs of extra design, engineering and testing in creating a cellular version of a product, amortised over the volume sold.

Today, normal 4G modules for devices cost perhaps $20-30, with SIM, battery & other components added to that. 

New versions of LTE are being designed to reduce costs, by cutting down some of the functions of "full" LTE. The target price for a new Cat-1 LTE module, optimised for M2M, is about $15. It's reasonable to imagine that Cat-1 (or Cat-M, its successor) will get to the $10 range over the next couple of years.

In parallel to this, when the new 3GPP NB-IoT low-power standard starts to ship (maybe mid-late 2017, being optimistic) the price should be more like $5-10, with an intention (I'm guessing 2018) to get that below $5. Add on an extra amount for the eSIM licence and design/test costs - probably a few dollars more. Then add on whatever is needed in terms of extra battery, software and so forth.

In other words even in two years' time, adding cellular to a consumer device will still cost the manufacturer at least $10 and perhaps $20 depending on the power/transmit speed needed, number of frequency bands, fallback to 3G, voice support and so on. While that's better than today, it's still significant for a manufacturer to wear.

Now $10 does not seem like much - or even $30 - until you consider the underlying costs of the devices they're supposed to be built into.

• A $20,000 car might have a 13% gross margin, or $2600, amortising the R&D and sales & marketing costs
• A high-end laptop might have a $100/unit gross margin, and a low-end one maybe $30
• FitBit (the largest wearables company) makes 46% margin on $87 average selling price, so about $40/unit
• A $300 washing machine might have 17% margin, so $50/unit
• A $30 toaster probably has a $5 profit margin

So in other words, adding a cellular module now, and also in the mid-term future, is a large % of gross margin for most consumer devices, irrespective of whether it uses SIM or eSIM.

Nobody is going to add a $10 extra cost to a toaster which has only a $5 margin, unless they can charge an extra $10 (or preferably $20) for it. And if only perhaps 10% of people actually (a) care enough to want a connected toaster, and (b) are willing to pay the extra cash upfront, then the product will become uncompetitive. Instead, the manufacturer could make two versions - normal & connected - sold at different prices. But that adds complexity in manufacturing, adds inventory costs, and there's no guarantee that retailers will stock both anyway. 

There's also no realistic way for cellular operators to subsidise the new mToasters down to the normal price, unless they sell them in their own stores, or find a way to reward the manufacturers with a sign-up bounty or rev-share once they get activated.

Result - the cellular Connected Toaster market is a non-starter, unless someone works out a way to print adverts on toast in shades of brown, and creates a new business model. And even then, you could probably do it more cheaply and easily with a WiFi Toaster.

Now obviously that's an extreme example - but it is designed to make the point that if (radio module+SIM) is a big % of the underlying device gross margin, and take-up rate is likely to be low, then the concept is not viable. In particular, if a device doesn't already come in (successful and well-used) WiFi-connected versions, it is unlikely to succeed in cellular variants, unless it has wheels or legs, plus high margin and a possible new revenue stream.

A $2000 specialist mountain bike might get a cellular radio built-in. A $20 bike sold in a developing country will not. 

In other words, while eSIM is a helpful advance for some types of connected IoT device, it's not a huge game-changer which will mean every home appliance, and every wearable product, will adopt 4G. Where it is realistic is in categories such as:

• Expensive items such as cars, which can wear the extra BoM cost of the radio module and SIM/eSIM easily, and which may be re-couped by extra revenue streams to the manufacturer such as warranty sales
• New categories of devices which always-on wide area connectivity to function - eg "lost and found" tags or wearables for wayward pets and children, or realtime heart-rate monitors with emergency alert capability for cardiac patients (notwithstanding insurance liability costs).
• Special "connected" premium-priced versions of products that normally just rely on WiFi or other short-range wireless (eg most wearables)
• Separately-sold accessories, eg aftermarket security "trackers" for bicycles
• Existing SIM-connected devices where the physical SIM creates extra complexities (eg some tablets, some industrial machiner - and maybe, finally, mainstream laptops etc)

This also means that the vast bulk of upcoming IoT devices (the quasi-mythical 10bn, 20bn, 50bn figures) will not support cellular, certainly by 2020, and perhaps even by 2025, unless 5G module prices get below $1, which seems unlikely. The majority will either use WiFi, cheaper (& non-SIM) LPWAN technologies, or maybe aggregated locally via a cellular gateway.

Cellular is definitely a player in IoT, but it will certainly not be ubiquitous, eSIM or not. There needs to be a specific reason and use-case for its inclusion - it is too expensive to be added in by a manufacturer just as an extra feature, except on very expensive/profitable products.

details here