In the last few months, we have published a couple of reports looking at the ongoing deployments of Low Earth Orbit satellite constellations to deliver IoT. The LEO arena is a hot one at the moment. Many of the headlines have been grabbed by the likes of OneWeb and SpaceX which are focused on deploying thousands of satellites for delivering broadband internet. It is the IoT-oriented constellations that are of most interest to us.
The IoT focused LEO roll-outs typically consist of 20-150 satellites, or are planned to do so, operating between 160km and 2,000km altitude. They have non-geostationary orbits, meaning that they do not address a fixed point on the earth’s surface, but are constantly moving.
The result is that a small number of satellites can cover the whole world, although without continuous reception. This works fine for many IoT applications. At the end of January we published the report ‘A technology overview of Satellite IoT’ which examined all the various technology aspects of delivering IoT via satellite, including questions of frequency bands, protocol choices, and architecture.
Subsequently we published the ‘Low Earth Orbit (LEO) satellite IoT connectivity vendor landscape’ report in February. That report looked at the capabilities of 44 companies involved in the LEO space, the majority of which are operating (or planning to operate) constellations of LEO satellites. Even narrowing down to those where IoT will be a significant focus, the number is still around 20 players who seem to be serious about deploying.
We think that the market is not big enough to sustain 20, or probably even 10, IoT-focused LEO satellite companies. In recent months a few would-be operators, such as Hiber, decided not to launch their own constellation and instead rely on those of other companies. We expect many more to follow suit. Doubtless there is some unmet demand for lower cost satellite connectivity, but it’s unlikely to be at sufficient scale. Cellular networks, and particularly LPWA networks, have very good coverage of populated areas, which account for the lion’s share of where IoT devices will be located. There are use cases for rural areas, including agriculture and environmental monitoring, but the volume of this opportunity is currently still very much unproven.
And the networks have yet to really be deployed. Today only two-thirds of LEO satellite operators have deployed any satellites at all. The average is 11. Only Swarm Technologies has really launched a significant number, 120. Overall, the average LEO operator has deployed just over 20% of what they’re planning to launch.
There is also a particular challenge with economics. Satellite constellations have high fixed costs and low variable costs. In a highly competitive market, there will always be a tendency for price erosion down to the marginal cost, meaning that the operators will struggle to repay the capital investment. This will make for very unhappy investors who may pull the plug.
This may be a self-sustaining argument. The costs inherent for any end user that needs to migrate their devices to an alternative connectivity provider in the case that their existing provider goes under will be significant. Accordingly, end-users are likely to favour those vendors with deeper pockets and more skin in the game.
Who will win? It’s largely the satellite operators that can weather the storm of fierce competition. That will tend to be those with deep pockets, including backing by existing satellite players.
There is also a technology aspect to this as well. Some LEO deployments are using established LPWA technologies, most prominently LoRaWAN and NB-IoT, while others are using proprietary technologies. While many of the proprietary technologies are optimised for satellite connectivity and provide potentially superior capabilities, they have a commercial disadvantage. The market scale will sit with devices that are closest to terrestrial devices that ship in large volumes. Making LoRaWAN and NB-IoT devices satellite capable will require a premium of just a few dollars over a regular terrestrial connected device, e.g. for a better antenna. The scale of production of these types of devices should mean that they will have a much lower per-unit cost than any proprietary technology. Also, the ecosystem is better developed with a wider range of vendors.
Furthermore, the requirements of most customers will not be satellite-only. Most will want some combination of satellite and terrestrial connectivity, to support the device regardless of where it is located. In such a circumstance, it makes more sense to have a single mode NB-IoT or LoRaWAN device than multi-mode with a proprietary technology, both in terms of cost and complexity. Homogeneity of technology will help with simplicity of deployment.
At the moment LEO satellite for IoT looks like it has over-heated before it has even really begun, and only a few will thrive. It might be best if the majority of would-be players cut their losses and decided not to launch. Deep pockets and standards-based offers will carry the day.