From Airbnb to Amazon and Uber, digital platforms have disrupted many industries, including travel, retail, music and others. From an AIoT point of view, we are interested in digital platforms that connect to physical assets and products. An AIoT platform operator does not necessarily have to manufacture or operate the physical assets. This is why platforms are covered independently of the Digital OEM and Digital Equipment Operator roles introduced earlier, even though the approaches can of course also be combined. This section provides an overview of the different concepts and looks again at the why, what, and how of platforms.
Successful digital platform businesses create network effects that scale globally, creating huge revenues and profits without having to invest in a physical infrastructure. Since digital platforms rely on an ecosystem of external producers and consumers, the only limit to scale is the size and value of the ecosystem. Uber relies on independent drivers, as well as people in need of rides. Airbnb does not own hotels or apartments, but has grown a multibillion business as a neutral broker, taking a cut off every transaction. Generally, it seems that there are two distinct motives for becoming involved in a digital platform business:
- Winner takes all: Many platform businesses are fairly dominant in their own domain, making them either extremely profitable or an extremely interesting growth investment
- Underdogs team up: Often, companies that cannot achieve a dominant position in a platform market on their own seek to align themselves with other players to try to catch up to the leader
Many industrial players have been trying to imitate the success of B2C platforms in the last couple of years, motivated by the scale these B2C businesses have achieved. However, it seems that a key reason for success of the B2C platforms was a combination of simplicity and focus. Many B2B platform scenarios suffer from the fact that industrial solutions often tend to be much more complex, and require a broader approach. For example, creating a platform that brokers holiday apartments cannot be compared to a platform that brokers automotive sensor data. The latter has to deal with more complex stakeholders, as well as a much higher diversity of data sources (there can be dozens of different sensors on a car, and they will differ from model to model). Since many AIoT use cases are of an industrial nature, there also seems to be an opportunity here to address this.
The basic concept of a multi-sided business platform has been well described by G. Parker et al. in Platform Revolution: How Networked Markets Are Transforming the Economy. The platform provides the infrastructure (e.g., an appstore) and brings together producers and consumers. The producers are creating the platform content (e.g., apps in the appstore), while the consumers are buying or using it (e.g., by downloading apps to their smartphones).
Three paradigm shifts are described as key for moving toward a platform business model. First, the move from resource control to resource orchestration. Traditional companies have tangible assets on their own balance sheets, e.g., real estate, factories or mines. Platform businesses are based on a less tangible asset, the ecosystem of providers and consumers. The value of Airbnb is the large community of holiday home owners and seekers. Second, from internal optimization to external interaction". Again, traditional companies are focusing on internal activities to create and sell products or services. Platform businesses focus on value creation by building external ecosystems. Third, "from a focus on customer value to a focus on ecosystem value". Traditional companies focus on the lifetime value of their customers. Platform companies focus on creating network effects between their customers.
AIoT-enabled platforms include physical products or assets as a key part of the ecosystem that creates the platform network effect. For example, this can mean that physical products provide data, which is then consumed by platform customers - either human users, or other physical products. The IoT enables connectivity, either in a producer or a consumer role or both. AI can support the producer in creating a meaningful offering. Equally, it can support the consumer in making use of the platform, e.g., by processing data from a platform in a customer-specific way. Alternatively, AI can be applied by the platform operator to create swarm intelligence that benefits from multiple data producers.
The authors of the platform revolution provide three recommendations for building a platform: magnetism (producers and consumers must attract each other), user-generated content, and implicit creation of value by the platform owner. Applying this to an AIoT-enabled platform, this means:
- Magnetism: The AIoT-enabled platform must find a match between consumers and providers that creates this magnetism. This will heavily depend on the type of physical assets or products involved, and the supported use case.
- User-generated content: For example, sensor data from assets in the field
- Implicit creation of value: For example, a swarm intelligence that combines the data from multiple sensors, e.g., to create a real-time map or road conditions
Since this is difficult to generalize without being too generic, let us take a look at a concrete example in the following.
Example: Parking Spot Detection (Multi-Sided Business Platform)
This is an example of a multi-sided business platform enabled by AIoT: cars equipped with ultrasound sensors can detect available parking spots as they are passing by them. These data are collected in a centralized platform and monetized, e.g., via a find-a-free-parking-spot app. Multiple OEMs might provide parking data to the platform operator, which integrates, consolidates and markets the data.
As mentioned earlier, building successful platform business models in industrial environments - or environments involving highly complex products such as cars - is not easy. Often, this is because in these environments we are facing a mixture of technical complexity, stakeholder complexity, and legacy (physical assets, products and equipment in the field are often suffering particularly badly from high levels of heterogeneity, because of their often long lifetimes).
A number of data marketplaces have emerged in the last couple of years that focus on bringing together OEMs and after-market customers, e.g., in automotive. The challenge here is manifold. First, OEMs do not always have an interest in making their data available, not even for payment. Second, the question is how to integrate - through basic APIs accessible over the internet, or through custom hardware deployed on the vehicles (which allows better integration, but increases costs). There are also some startups that are providing completely generic marketplaces for sensor data. The problem here is that they are often too generic, making it difficult for users to truly find a relevant offering (missing "magnetism").
Another potentially interesting area for AIoT-enabled platforms is industrial AppStores, or AppStores for complex consumer products such as cars or kitchen equipment. The example from before would not be possible without such an AppStore. However, there are at least two challenges here. First, the number of relevant consumers of the apps is most likely much smaller than in a smartphone AppStore, thus making it more difficult to build profitable apps. Second, the OEM would have to provide the app developer access to APIs, to get access to sensors such as the ultrasound sensors in the previous example. While many smartphone vendors are making increasingly more sensor APIs available to app developers, this might still take a while in industries such as automotive, simply because any security problems with the OEM's app sandboxes could have potentially catastrophic consequences. An interesting step along this journey could be app stores which are only accessible to trusted partners, as we have described in the co-creation section.
Therefore, while AIoT-enabled platform businesses are certainly not straightforward, it will be interesting to observe how the industry will approach this, and who will be the first players to succeed in their areas. It seems fair to say that smartphone app store players have already shown how to do this (using their own form of AIoT), and others will eventually follow in their domains.
- ↑ Platform Revolution: How Networked Markets Are Transforming the Economy - And How to Make Them Work for You, Geoffrey G. Parker, Marshall Van Alstyne, and Sangeet Paul Choudary, 2016, Norton & Company