In business, experts generally recognize two platform types: innovation and transaction.¹

Innovation platforms, such as the developer platforms from Amazon Web Services, Facebook, and Salesforce, enable other companies to create complementary services or products using the resources of the platform. For example, independent software vendors can use the AWS cloud computing infrastructure to build their own applications.

Transaction platforms enable multiple parties to exchange goods, services, software, or data in exchange for money. Such marketplaces as Airbnb, Uber, and the iOS and Android smartphone app stores are classic examples.

To be successful, innovation platforms must provide a broad ecosystem of players with the tools they need to build complementary services or products. Transaction platforms must reduce the transaction costs between two parties. In many cases, companies create a hybrid model that combines both types of platform. Salesforce, for example, offers a combined app development platform and a marketplace for third parties to sell their enterprise apps.

The value of data in B2B is difficult to extract; companies need both the domain expertise to develop new data-driven solutions and the customer relationships required to monetize them. This complexity means that digital ecosystems, such as those built around IoT platforms, are insufficient for capturing the value of data on their own. New, built-for-purpose data ecosystems are required to organize the collective data assets, capabilities, and customer connections of a group of business partners to deliver new products and services—both within and across traditional industry verticals. While digital ecosystems provide the underlying platforms, data ecosystems enable B2B companies in asset-heavy industries to generate additional revenues and build enduring competitive advantage with their IoT data. (See the exhibit.)


Note


1. A distinction drawn from The Business of Platforms: Strategy in the Age of Digital Competition, Innovation, and Power by Michael Cusumano, Annabelle Gawer, and David Yoffie.

To better understand the likely development of platforms and ecosystems in B2B industrial sectors with different dynamics, we took an in-depth look at five sectors: precision agriculture, commercial buildings, automotive (with a focus on autonomous vehicles), trucking, and manufacturing automation. We explored the range of potential use cases in each and the degree to which they benefit from network effects and data aggregation. The frequency and value of use cases that benefit from both indicate the potential for platforms and ecosystems and whether there will be a consolidation trend. (See Exhibit A1.)

Precision Agriculture

Conditions favor IoT platform adoption in agriculture. Farmers have a clear need to improve productivity and yield, and adoption of precision sensors, drones, advanced analytics, and autonomous equipment connected to data aggregation platforms is already on the rise. Farmers own their data and can choose multiple parties for data sharing on the basis of the solutions and services these parties provide.

Among the 50 precision agriculture use cases in our analysis, few inherently exhibit network effects, but 17 (35%) benefit from data aggregation. For example, customers of Farmers Business Network (FBN), which is backed by more than $200 million in venture capital, share anonymized data (such as seed prices, historic yields, and chemical and fertilizer treatments) with FBN in return for operational benchmarks. Using these benchmarks, farmers can compare input prices and see the yields of similar farmers, helping them negotiate with input distributors and retailers and optimize their farming practices.

FBN has generated network effects throughout its business model. As more farmers share data, the validity and scope of the benchmarks improve. Moreover, FBN has built an e-commerce marketplace (a transaction platform) where farmers can purchase seeds, chemicals, and equipment from FBN or third parties. As the number of third-party sellers increases, farmers get a wider selection of inputs from which to choose, and sellers can reach a larger number of farmers. The fragmentation of the farming customer base enhances both the value of data aggregation and the value of a transactional marketplace. (About 90 farms, each with more than 2,000 acres, work 60% of the land in the US, while about 1,900 small and midsize farms work the rest, according to USDA.)

In agriculture, there are relatively mature efforts to drive data standardization. Ag Gateway, a nonprofit consortium with more than 200 member companies, including John Deere, AGCO, and Monsanto, has developed the ADAPT framework, which acts as a common data model, API standard, and set of open-source and proprietary data plug-ins that enable precision agriculture software to integrate with data generated by many different types of farming equipment. Over time, as these standards mature, we expect new entrants providing precision agriculture point solutions to enter the market and potentially join the ecosystems forming around major precision agriculture platforms. John Deere has a suite of APIs for third-party developers and has released a precision agriculture marketplace where farmers can contact companies selling farming software that leverages data from Deere’s farming equipment.

New types of data are also becoming available—from unexpected sources. Airbus, the aircraft manufacturer, offers a set of API services that provide crop analytics based on satellite images. These services, delivered via a precision farming software portal, enable Airbus to act as an analytics supplier to precision agriculture manufacturers and software vendors. In the future, the fusion of satellite and farmer-owned data, aggregated on a platform such as FBN’s, will further enhance analytics for crop optimization. While FBN could emerge as a dominant platform and ecosystem, it faces competition from the platforms of incumbent players such as John Deere and Monsanto’s Climate Corp.

Going forward, we expect to see a few, complementary precision agriculture platforms remain, with some, such as FBN, focused on specific parts of the value chain (crop procurement and sales, for example), others focused on managing farm equipment, and yet others on crop analytics. These platforms can coexist, with farmers sharing data with multiple platforms for different uses. But farmers will share sensitive crop data only with platforms that do not violate their trust by reselling data to third parties that could use the data in ways that are not in the farmers’ best interest. For example, a recent case concerned the potential use of farm and soil productivity data to increase land lease prices. Platform providers face the strategic questions of how to engage farmers to share data, which use cases to address, and how to monetize the data without violating farmers’ trust. The answers will shape the evolution of the competitive landscape.

Commercial Buildings

In the commercial building industry, there is potential for connected-building solutions to deliver compelling value propositions to developers, landlords, builders, tenants, and others. Connected HVAC and lighting systems can reduce energy costs, which can account for 40% of a building’s total operating costs. However, there are fewer use cases that benefit from network effects and data aggregation than in other verticals. Of the 65 use cases in our analysis, only seven—among them smart parking, flexible workspace allocation, identity management, and predictive signage—exhibit network effects.

The data standards in commercial buildings are less mature than those in the precision agriculture sector. Every connected building is filled with equipment and systems from multiple OEMs; one supplies the lighting, another the HVAC system, and others the elevator, access control, and security systems. Each system, if connected, transmits data in different formats. Existing building management software cannot easily integrate data across different equipment types without a significant system integration effort. Innovation platforms with open APIs are not yet available in commercial buildings because there is no common data model across the sector’s domains and equipment suppliers.

The fragmented commercial building customer base—in New York City, for example, the top five building owners own only about 2% of the commercial square footage—and the lack of standards and APIs make it difficult to implement industrywide solutions.

This may be changing as equipment and building management software OEMs identify opportunities to promote data standardization. For example, Siemens is supporting Project Haystack, a nonprofit consortium that has built a common framework for tagging building data. Johnson Controls, a provider of HVAC and fire safety equipment, is supporting the Bricks Schema, another open-source effort to build a common data model for the building industry. Haystack and Bricks are both working with established industry bodies to integrate their data models into new building specifications. As standardization efforts progress, we expect to see major building controls OEMs offer platforms with similar suites of building applications, essentially forming a competitive oligopoly with little room for new entrants.

Automotive

With the advent of new mobility models enabled by rideshare platforms and vehicle autonomy, change in the automotive industry is accelerating dramatically. The rapid rise of connected vehicles and the resulting flood of data across OEMs point to a ripe opportunity for data aggregation across highly fragmented sets of consumers and car manufacturers. Of the 49 use cases we analyzed, 31 demonstrate either network or data aggregation effects. We focus specifically on autonomous vehicle (AV) development as an underlying technology trend with significant implications.

Multiple platform scenarios are possible in the AV technology stack, depending on which layer is under consideration: the AV software stack running in the car, the high-definition mapping layer, or the fleet management level. (See Exhibit A2.)

Autonomous Driving Software. The AV software companies with first-mover advantage can collect large driving data sets to develop more reliable software by covering more “edge cases.” They can then deploy autonomous vehicles at scale and pursue platform-based business models, such as licensing their software to multiple OEMs or to rideshare companies for application in robotaxis or last-mile delivery use cases. For example, if Waymo (Alphabet’s autonomous driving venture) can develop such an advantage in software performance and reliability, it has the potential to become the digital operating system for the autonomous automotive industry. (GM’s Cruise, another autonomous driving enterprise, could well have something to say about that.)

High-Definition Mapping. HD mapping may become a winner-take-most data aggregation platform because there are clear economies of scale in map creation and maintenance. In this scenario, multiple makers of autonomous vehicles would license HD maps from a sole provider, analogous to mapping software provider HERE, which is owned by a consortium of Mercedes, Audi, and BMW and competes with Google Maps. Numerous other specialized mapping providers and consortia have formed. For example, Toyota, Aisin, and Denso have put together a group that OEMs can join. They share their raw camera data and, in return, can access the HD maps created from the consortium.

Creating HD maps is complicated because there are no common data standards, and each car model produces its own mapping data based on its specific sensor configuration (which involves camera, LIDAR, or radar and different sensor locations). Fusing data from different vehicles’ sensor stacks to create an aggregate map is technically challenging, and numerous autonomous vehicle and mapping companies are redundantly mapping the same territories. Governments, such as the UK, are encouraging the formation of common mapping standards and data sets to mitigate these inefficiencies. A startup from Silicon Valley, DeepMap, has developed HD mapping software that can process data from a variety of sensor configurations. As the underlying sensor and hardware stack for self-driving vehicles continues to evolve, being able to generate HD maps from, and for, various types of autonomous vehicles will be a competitive advantage for DeepMap if the company has truly solved the data fusion challenge. If mapping data can be fused, then network effects also come into play as multiple OEMs can contribute their data, making the data sets more accurate at the same time. If technical challenges can be overcome, HD mapping could be a winner-take-all platform.

AV Fleet Management. The fleet management layer of the AV technology stack may become a transaction platform, operating much like the consumer ride hailing platforms Uber and Lyft, to connect drivers and riders. Companies such as RideOS have already developed sets of APIs that are specific to ride hailing to enable third parties, including autonomous car manufacturers and cities, to implement their own ride hailing platforms. Because drivers and riders can use and easily switch between multiple ride hailing platforms, they exhibit weaker localized network effects, and multiple autonomous vehicle fleet management platforms can coexist.

Trucking

Of the 41 telematics use cases we analyzed from the trucking industry, roughly 40% exhibit network effects or benefit from data aggregation. The combination of industry data standards, such as J1939, and the fragmentation of trucking fleets has spurred rapid innovation in the industry.

Three types of platforms address different use cases in commercial trucking: load boards, digital freight brokers, and fleet management application marketplaces. Load boards provide real-time postings of available trucks and cargo shipments, enabling carriers to find loads to transport (an example of indirect network effects). Using machine learning algorithms and aggregated data, digital freight brokers (which have raised more than $600 million in venture capital funding over the past five years) seek to disintermediate the traditional broker model through automation. To do so, they will need to overcome the fragmentation and heterogeneity of carrier truck fleets and shipper needs. Carriers, for example, often operate on specific lanes in a given geography, and shippers frequently have specific requirements, such as safety ratings or equipment types (flatbed or refrigerated trucks, for example).

Fleet management applications have existed for years, but by aggregating data from trucks across fleets, management companies such as Geotab have created platforms and app marketplaces that enable innovation in the industry. For example, Geotab aggregates fuel economy data across truck types and enables carriers to compare their fuel economy performance with that of similar vehicles and industry benchmarks.

The advent of autonomous trucks combined with digital freight platforms may accelerate consolidation of platforms by altering how carriers and shippers transact. In addition to replacing the driver, autonomous trucks reduce the need for human dispatchers. Autonomous trucking companies can develop APIs that integrate with digital freight brokers. Shippers will be able to automatically book the capacity of a truck via APIs instead of human brokers. The necessary scale and investment in both autonomous trucks and digital freight platforms could drive consolidation in the carrier industry and convergence or interoperability among digital freight brokers, fleet management platforms, and autonomous trucks. As autonomous trucks lower total cost of ownership and generate more revenue (they can operate 24/7), the largest carriers will gain substantial economies of scale over smaller competitors that still rely on human drivers and could rapidly gain market share. We expect that, over time, significant consolidation of trucking platforms or interoperability across different platforms will maximize network effects.

Manufacturing Automation

Manufacturers have embraced IoT solutions to enhance their operations. Most companies with an industrial software and controls heritage, such as Siemens, Emerson Electric, Schneider, ABB, Honeywell, Bosch, and Rockwell Automation, have launched or taken equity stakes in IoT platforms that improve their customers’ operations and processes.

In our analysis of 51 use cases in manufacturing, however, less than 5% exhibit network effects, while 20% exhibit potential for data aggregation. For example, automated spare parts procurement and inventory management show network effects. Data aggregation use cases include operational benchmarking; customers can compare their average mean time to repair (MTTR) of equipment with peers based on anonymized customer data. Other use cases in manufacturing suggest strong potential for cross-company data sharing.

To make it easier to implement use cases and access third-party solutions, providers such as Schneider have built both transaction and innovation platforms. On Schneider’s Electric Exchange, for example, users can share information via Q&A forums, developers can access software development kits, and customers can access potential suppliers, including third-party independent software vendors. However, while these efforts support the development of new platform-accessible solutions, they do not address the fundamental challenge of scaling IoT solutions within a customer.

Multiple equipment providers offer IoT platforms. But the last thing a manufacturer wants is multiple industrial IoT platforms that all offer native connectivity and data modeling capabilities for the provider’s equipment but only limited integration with third-party equipment. Multiple costly platforms weaken the business case for IoT use. (A 2019 Microsoft-sponsored survey indicated that roughly one-fifth of IoT proofs of concept fail because there are “too many platforms to test.”)

Customers lack the in-house talent to build and integrate end-to-end IoT solutions from multiple providers. To overcome this issue, platform providers and equipment OEMs are partnering to lower integration costs and aggregate use cases into broader enterprise applications that address a larger value proposition.

Customers can follow a few paths in implementing IoT solutions. (See Exhibit A3.) One is to adopt a single IoT platform and bear the integration and complexity costs of connecting their equipment and plants and modeling their operations. In this approach, customers can leverage only their own data for analytics, limiting the potential value of the platform.

A second option is to adopt multiple point solutions, each with native connectivity and data modeling capabilities covering particular types of equipment but limited integration across platforms and equipment types. If the point solutions are provided by the equipment OEM and the OEM is aggregating data across its customers’ installed base, then the OEM should be able to access a much larger data set and provide better analytics-based solutions. In this situation, the customer benefits from OEM knowledge and data, but the solutions are not integrated to optimize end-to-end processes.

To address this tradeoff, industrial IoT platform providers can strike partnerships with equipment OEMs, which enables solutions that access all equipment data from each OEM across the customer’s manufacturing plant network and integrates data for the customer at the cloud level. Use cases requiring machine data from multiple OEMs—such as production process optimization that improves throughput, reduces scrap, and enhances quality—can be unlocked more easily. A customer can use one IoT platform to access multiple OEM equipment data streams and solutions and build end-to-end process models. This should significantly reduce the cost and complexity of implementing use cases.

In addition to reducing the integration cost involved in accessing shopfloor equipment data, platform providers can aggregate use cases into new enterprise applications. Just as we have seen enterprise software solutions evolve from customized company-specific code to enterprise applications such as ERP, MRP, SCM, and CRM, we now see six new categories of IoT applications emerging: asset optimization, process optimization, industrial worker productivity, energy management, augmented reality, and industrial cybersecurity. (See Exhibit A4.) The benefit of these new application classes is that they package individual point solutions into broader value categories that generate a more attractive return on investment, provide more mature functionality, and reduce the cost of implementation.

But the industrial incumbents launching IoT platforms do not have a monopoly on the market; they both partner and potentially compete with “hyperscalers” such as AWS and Microsoft Azure. Almost every industrial IoT platform leverages infrastructure-as-a-service (IaaS) in the public cloud. Honeywell, Schneider, ABB, and Emerson all position themselves as IoT solution providers, based on their decades of domain expertise in their core verticals, but they all embed Microsoft Azure’s platform-as-a-service (PaaS) capabilities within their industry-specific platforms. Siemens and PTC emphasize their own IoT platforms and are recruiting developers to build IoT applications for their proprietary platforms. For example, PTC offers IoT solution “building blocks” (such as domain-specific business logic and common user-interface elements) as well as a suite of connectors with enterprise systems to accelerate and simplify IoT application development for its customers and partners. Bosch has taken a different path, leveraging an open-source IoT platform, Eclipse, to build its offering—trying to follow the model of companies such as Red Hat in the technology industry by enhancing open-source software with support and maintenance services.

With substantial R&D budgets and a history of rapid innovation (Microsoft launched more than 100 new IoT services on Azure in 2019), platforms continue to evolve. The hyperscalers are investing in new IoT capabilities. For example, AWS’s IoT SiteWise service enables communication with industrial controls via the manufacturing-specific OPC-UA protocol and comes prepackaged with dashboarding capability to, among other things, monitor overall equipment effectiveness. As hyperscalers cocreate IoT solutions with customers, they may start offering their own IoT-based enterprise applications. An AWS executive observed at a recent developer conference that a “long list of product ideas” came from the joint AWS-Volkswagen Industrial Cloud partnership. Whether the hyperscalers or the vertical-specific industrial IoT platform providers ultimately prevail is an open question based on the degree to which the hyperscalers choose to compete with some of their largest industrial customers.

The authors are grateful to Patrick Su, Claire Wu, Tazia Middleton, Moira Scanlon, Natnael Kassaw, Bernhard Siegert, and Parth Tripti for research and analysis supporting this publication. They also thank David Duffy for his help in writing this report and Katherine Andrews, Kim Friedman, Abby Garland, Adam Giordano, Frank Müller-Pierstorff, Shannon Nardi, and Ron Welter for their assistance in editing, design, and production.