The future of M2M: Eternal Fragmentation or Winner-Takes-All?
[In The M2M Ecosystem Recipe report, which we published in June, we presented some ideas on how the full promise of the Machine-to-Machine market might become reality. Not all aspects of M2M ecosystem thinking were fully explored, however. In a response to the report, telecom specialist Tsahi Levent-Levi posed several to-the-point questions and counter-arguments. We always welcome debate, so we invited Tsahi to have a conversation about the issue of fragmentation in M2M. In this blog post, Tsahi and Stijn Schuermans, the author of the M2M report, try to advance the thinking on the topic of fragmentation in the emerging M2M market. Read on – and join the debate!]
Tsahi: I love what you guys are doing at VisionMobile. Most times I fully agree with the reasoning behind it. The M2M piece? I think there’s a real opportunity here – it is definitely a market that is bound to happen. My main difficulty is in envisioning how this will play out. The use of the analogy you did between smartphones and M2M it is the chart indicating the rapid growth of M2M. In a way, this chart suggest that a similar growth pattern can be achieved in M2M. To me, these businesses are apples and oranges. Where do you see the similarities between the two?
Stijn: Rather than focusing on the uniqueness of M2M, we focus on the similarities with past industries. M2M and smartphone businesses are indeed not the same thing. This said, we see many parallels from an economic perspective with other industries that have made the transition from a fragmented vertical structure to an ecosystem-led business model. Smartphones, but also PCs, game consoles, or online retail. The conditions in the M2M market today closely resemble those in other markets before they made the transition to ecosystems. For example:
- A highly fragmented market, but with a technology that’s becoming more and more modular and affordable.
- A complex technology and an inefficient market that is mostly only accessible to highly skilled organizations with deep pockets, like utilities or car manufacturers, while a much larger group of small businesses or consumers is not participating in the market (due to the high costs of transacting) and therefore are not innovating or creating value.
- A focus on a few high-profile, uniform use cases. A lack of diversity that fails to address unexpected customer needs. (Some analysts still strive to map the entire spectrum of M2M use cases. Try that with iPhone apps.)
We therefore think that the same principles that caused the flip in those historical examples also come into play with M2M.
Tsahi: But what about the differences between these two markets?
To me, M2M is about devices. Lots of devices. Varied ones. Each focused on solving a specific problem. Smartphones, on the other hand, are swiss army knives – they can do everything. That being the case, it is hard for me to see a clear single winner of the M2M ecosystem.
Stijn: Smartphones were not always the swiss army knives they are today. The predecessors of the modern all-in-one smartphone were described at the time as feature phones (heavily sub-segmented, with very distinct features for distinct consumer groups), email/telephone combos (BlackBerry), but also PDAs (Pocket PCs, as one manufacturer called them), industrial mobile data collection terminals and even digital cameras or music players. A device designed to solve the needs of a Western businessman had a very different design from that of device for Asian youngsters. Of course there were similarities and common features as well, but that’s true of most M2M devices today.
The core issue here is fragmentation. You’re absolutely right: it’s a major challenge in M2M and a big point of friction to the emergence of an ecosystem. (And admittedly under-addressed in the paper.)
Today’s M2M devices are conceived as purpose-specific, vertical devices, and therefore highly fragmented. In most if not all of the historical examples in the history of computing, however, this situation was a temporary phase, followed by an era of modularization and multi-purpose devices. From computers designed for science or accounting (but not both) to personal computers with scientific and accounting software. From feature phones, email phones and PDAs to converged smartphones with apps. The modular solution was difficult to imagine back then, but looks completely natural today.
A multi-purpose M2M device is not unthinkable; it’s actually likely to emerge if history is a guide. Arduino and the likes are early prototypes, if you will, of such devices. (Btw, there are literally hundreds of Arduino-compatible modules out there in all sorts of form factors, including tiny ones, cheap ones and with specific sensors already attached.) There is no fundamental reason why M2M devices must be developed for specific verticals only, just inertia.
If you don’t assume that M2M devices will remain extremely fragmented and organized along vertical lines forever, then the presence of multi-purpose M2M devices opens up the possibility for a fundamental rearrangement of the value chain in M2M. Suddenly service and device don’t need to be manufactured by the same vendor anymore, or even sold together as a package. This opens up the market for 3rd party developers to create services based on existing modules, or even on top of an installed base of modules (abstracted as data streams) already deployed. Once multi-purpose M2M devices and/or abstraction layers like cloud services emerge, the road is paved for large service ecosystems, quite possibly with winner-takes-all market dynamics.
Fragmentation must be managed, but not completely eliminated, for the benefits of modularity to emerge. Not all smartphones have the same capabilities, features, processing power, operating system version, screen size, quality and price points. Fragmentation is still an issue in smartphones: between OS’s (about five of them are seriously in the running at the moment) and notably within the Android ecosystem. However, we’re now at a much better situation than the “three million platforms with a hundred users each” from 2007-era mobile computing. At the moment there are thousands of incompatible M2M modules. Even reducing this to a few dozens would be a major breakthrough; one that could enable the transformation of the value-chain.
Tsahi: I beg to differ. Smartphones are all alike. They come to solve the same job. They are personal. You carry them in your pocket. You do things with them. They might have some different types of accelerometers or different resolutions, but can you really distinguish between them?
M2M? How does a toaster similar to a UPS package? How can a refrigerator behave like a humidity sensor on a farmland? Can you really say that these are only fragmentation scenarios of the same device? Sure – put an Arduino in all of them, but then you will see that the service being developed must come from the company that designs the hardware, as the use cases for each is extremely different.
A refrigerator vendor might publish an API (or extend a baseline API), but that API has nothing to do with the people counter at the entrance to a shopping mall.
Stijn: Smartphones are not just personal, though – they’re also personalized. While the devices might all be similar (nowadays, not in the past), the ways they are used are not. Users install apps to tune the device to their own needs, and apps cater to just about every need imaginable. Smartphones have commonalities (personal, pocketable, connected) but fundamentally cater to thousands of use cases. That’s what differentiates them from feature phones in the first place.
M2M devices and services cater to many different use cases, but they also have commonalities. They all collect data, turn that data into actionable insights or automated response, and through actuators allow you to drive that response. The technologies needed to do that will be very similar across a very wide range of specific use cases. Farming, home automation, logistics, fleet management, and foot traffic measurement are the equivalent of apps that can be built on top of a common platform. Attributes like ruggedness, range or battery life are the counterparts of smartphone screen sizes or build quality.
The co-design of hardware and service for every distinct use case, historically, is unlikely to last. It’s like saying that computers are mostly use-case specific devices, and therefore the motherboard, processor, memory, peripherals and software must all be co-designed by the same manufacturer. That was actually true at some point, but then minicomputers and PCs came along. Once the performance of a device becomes good enough for most use cases (in M2M that might mean lifespan, battery life or processing power, for example), modularization ensues and general purpose devices emerge. This is because once performance becomes good enough, the flexibility and “mix and match” possibilities of a modular design becomes more economical and more valuable than the expensive full co-design.
You might argue that jet engine monitoring (the industrial internet), for example, will never use the same platform as say home automation. That might be true. The way it typically unfolds is that “easy” applications will start using the new technology (in casu the M2M platform) first, and then will suck in more and more advanced use cases. The industrial, high performance ones will be the last to succumb (if at all – people are still buying mainframe computers). However, are home automation, in-car connectivity, smart metering, etc so disconnected in people’s lives that they warrant different technologies?
Tsahi: You gave the example of the app store and apps as a way to drive an ecosystem in smartphones, and the explosive growth we see there today. What do you think will happen in the M2M business? Can apps be the prevalent driving force as well or will there be a different ecosystem?
Stijn: Digital ecosystems as we understand them emerge when separate groups, e.g. developers, consumers and device makers, are empowered to interact and transact with each other efficiently. In smartphones, the catalyst to this interaction are apps and app stores. Apps are where the value is created in smartphone ecosystems (but not necessarily where it is captured).
Where will value be created in a possible M2M ecosystem? It could be devices, but that’s unlikely as devices by themselves don’t solve user needs. It’s rather the services which will be the focal point (fed with data by said devices). Services might be closely tied to devices, or not. If you’re monitoring a jet engine, then they likely are. If you’ve monitoring pollution or traffic conditions with thousands of sensors owned by a community of people rather than a single organization, then the link is more tenuous. Services can include centralized software components in the cloud, APIs, device management systems or managed services (i.e. involving actual humans).
If you think about services as a focal point for the ecosystem, then the diversity in devices might become less important, as well. The ecosystem winner(s) could emerge on another layer, like cloud services.