1 WYSIWYG: The Uber-Bias

Continuity is boring; nuance is tiring, and gradualism is for the hesitant. To attract any attention, societal change has to be framed in terms of high-caliber notions like revolution, breakthrough or disruption, at the very least. Instead of tracking tedious processes of incremental modifications, historical moments of rupture and transformative change have to be identified: what we are seeing are not legacies of the past, but the advent of a new epochal paradigm. Notwithstanding the critique of such dramatic simplification, such “epochalism” (Du Gay, 2003, p. 664) allows the construction of overarching grand narratives animated by a teleological logic in which history unfolds as a sequence of sharply delineated phases.

One of the more recent epochal shifts has even been dated to a specific year: in 2016, platforms dislodged networks as the key organizational paradigm and chief governance principle of a new era of economic and societal development (Lovink, 2021); the relational governance that had given shape to the network society (Castells, 2004) was dethroned by a novel mode of algorithmic governance that ushered in the new phase of platform capitalism (Srnicek, 2016; see also Kenney & Zysman, 2016; van Dijck et al., 2018). The mantra of disruption left no doubt that this shift was anything but gradual, and the authoritative manual for splitting asunder entire industries carried the befitting title “Platform Revolution” (Parker et al., 2016). Network effects were the propellant of this revolution, and winner-take-all-markets its most radical proposition. Uber and Airbnb were celebrated as the standard-bearers of this revolution that elevated platforms to the status of the “dominant organizational form of the digital age” (Gawer, 2022; see also Davis, 2016) and the “21^st century ideal type of the firm” (Rahman & Thelen, 2019, p. 198), respectively.

This triumphalist portrayal of the epochal shift from network society to platform capitalism, however, is showing more and more cracks. These cracks have not been caused by attacks on the key assumptions of the orthodox account on platforms: they are not wrong as such; rather, they are criticized for being incomplete. The sweeping generalizations on the dominance of platforms in the digital age stand in stark contrast to the selectivity of the research on which these claims are based. Rather than covering the entire spectrum of economic sectors, platform research so far has primarily focused on the consumer- and communication-based internet¹ (Grabher & van Tuijl, 2020, p. 1006; Menon et al., 2020, p. 364; Butollo & Schneidemesser, 2022; Jovanovic et al., 2022, p. 2; Dolata, 2024, p. 10). Amidst the triumphalism of the ever-increasing market valuations of the US Big Tech companies Alphabet, Amazon, Apple, Meta and Microsoft as well as of the more specialized second-tier companies like Airbnb and Uber, the neglect of the vast spectrum of the production-based internet hardly appeared noteworthy — that is, until more recently.

An ever-growing body of research has extended the scope of analysis and accumulated compelling evidence for the distinctive realities of platforms in the realm of production and manufacturing (see, for example, Gawer & Cusumano, 2014; Kenney & Zysman, 2016; Menon et al., 2020; Dolata, 2024; Feike & Rösch, 2024; Lerch et al., 2024; Sauvagerd et al., 2024; Van Dyck et al., 2024). On a most general level, rather than a dramatic event of disruption that inaugurates a new distinct phase of capitalist development, this research reveals rather messy and tenacious processes of transformation; and rather than hitting on radical newness, the focus on production brings path-dependencies to light.² Instead of the complete replacement of incumbent industries, we see their gradual reorganization: hybridization instead of substitution.

Whereas hybrid and fragmentary formations, viewed through the lens of the disruption logic, indicate a deficient state, somewhere between the “no longer” and the “not yet”, the perspective advanced here acknowledges these formations as categories sui generis. Although this analytical strategy foregrounds hybridity and graduality, it does not preclude comparisons between stylized forms, like networks and platforms, for instance. However, rather than treating these forms as “arithmomorphic concepts” that systematically conceal processes of change and transformation in which concepts also contain elements of their opposite (Georgescu-Roegen, 1971, p. 45), they are apprehended as “ideal types” (Weber, 1968/1922): they afford a frame of reference against which actual processes of change and transformation can be perceived and conceptualized. Instead of canonizing ideal types as a normative state, they are pragmatically employed as heuristic devices that guide inquiry and hypothesis formation. Very much in the spirit of this analytical strategy, a critical engagement with the key tenets of the current platform orthodoxy reveals first hints (that later on will be followed up) to the frictions and constraints that preclude an outright disruption by industrial platforms.

Tenet #1: Assets lightness. Whereas the “asset-light” company (Parker et al., 2016) embodied the paragon of consumer platforms, industrial platforms run on asset-heavy “cyber-physical systems” (Menon et al., 2020, p. 363) that adhere to fundamentally different operational principles. Tenet #2: Network effects. Explorations of the disruptive force of consumer platforms revolve around the single most powerful escalating dynamics: network effects. Although the efficacy of this propellant of platform proliferation has been problematized early on (Boudreau, 2012), it is structurally limited in industrial platforms: their growth does not simply follow the arithmetic: more users = more value (Sjödin et al., 2021). Tenet #3: Frictionless scalability. Far from winner-take-all markets, the scope of industrial platforms is structurally limited since they operate in specialized sub-markets in which local and industrial domain knowledge comes at a premium and in which physical assets are subject to country-specific regulations (Sauvagerd et al., 2024, p. 6). Tenet #4: Algorithmic control. Whereas consumer platforms employ generic software architectures designed for standardized high-volume, low latency interactions (Feike & Rösch, 2024), the deployment of industrial platforms implies an intervention into highly interdependent socio-technical systems (Dolata, 2024, p. 33) that warrant lengthy processes of negotiation and trust building.

The paper starts off with a brief review of the epochalist perspective on capitalist development from the putting-out system to the current platform era in which the perception of the core-periphery architecture is largely based on research on consumer platforms (sections 2 and 3). By elucidating the specific limitations to the winner-take-all logic in the industrial realm, the focus then shifts from the replacement of forms to the processes of transformation and hybridization that, typically, are rife with challenges, setbacks and outright failures (section 4). The crucial steps of platformization from stand-alone product over smart and connected products to digital ecosystems will be elaborated by tracing the platformization of the leading agricultural equipment provider, John Deere (section 5). Reflections on the broader implications of platformization will conclude the paper.

2 Epochalism I: The Prehistory of the Platform

By taking advantage of the theoretical authority of the grand historical narrative, the platform — as an analytically distinct category — is stylized as the latest epoch-defining form in the development of modern capitalism. The progression of capitalism, in a sense, can be read as a book comprising four successive chapters, each with a clear beginning and end, and each contributing to the broader narrative, but distinct in its challenges and answers (see Table 1).

By superseding the notoriously unreliable putting-out system (Lazerson, 1995), the factory unleashed the economic dynamics of industrialization: Prometheus broke free of his shackles (Landes, 1969). Although the factory system habitually is associated with the arrival of monstrous machineries, it was the rather unspectacular instrument of the clock that served as the key governance apparatus: its mechanical precision dislodged the ambiguity of natural cycles and enforced the discipline (Thompson, 1967) that was a precondition for the synchronization of interdependent tasks (Mumford, 1934). Despite their technological advancement, though, early factories typically remained rather small-scale and single-site operations whose market power was negligible. Together with low barriers to entry and price-driven competition, the atomistic industry structure of the 19^th century, in fact, resonated with the ideal market world of neoclassical economics (Chandler, 1977).

With the process of modernization (Weber, 2016/1904), the factory got absorbed into corporations with an ever-broader spectrum of planning and marketing functions. The complex structure of departments, divisions and subsidiaries required elaborate schemes to manage the coordination across different units (Drucker, 2017): the ever more ramified configurations of subordination to authority as well as the lines of communication and reporting were specified in the novel governance apparatus of the organizational chart (Mintzberg, 1979). Since corporations, in contrast to single-site operation factories, routinely operated across multiple industries and geographies, their extensive webs of subsidiaries and branch plants afforded increasing possessional market power.

Consolidation into multi-layered bureaucratic organizations held out the prospect of efficiency gains under a certain condition: relative stability. Globalization, however, ushered in a new phase of heightened volatility that laid bare the limitations of rigid corporate hierarchies (Dicken, 1986): in turbulent environments, networks quickly proved superior to both, hierarchies (by providing flexibility), as well as to markets (by furnishing long-term collaboration aimed at sharing costs and risks) (Powell, 1990; Grabher, 1993). To leverage the advantages of relational embeddedness underwritten by contracts, previously fully integrated corporate behemoths morphed into nimble lead firms controlling sprawling global production networks geared towards the exploitation of location-specific comparative advantages (Henderson et al., 2002). The strategic network positioning of the lead firms as well as the specific network configuration engendered a form of relational power that secured a dominant role in increasingly oligopolistic markets.

The, so far, penultimate epochal rupture is associated with far-reaching processes of digitalization that paves the way for the shift from “Nikefication” to “Uberization” (Davis, 2016): from globally-dispersed production networks to ubiquitous digital platforms. Uber epitomizes the ideal of the asset-light platform orchestrator who avoids ownership over corporate assets (Davis, 2016) but, instead, aims at the far-reaching co-optation of assets for which it neither has to bear costs nor responsibilities (Stark & Pais, 2020). Instead of the contractual ties binding global production networks together, platforms are built around a digital governance apparatus (Stark & Broeck, 2024): algorithms regulate the interfaces and interactions through unilaterally imposed “uncontracts” (Zuboff, 2019, pp. 220–221). The algorithmic management of platforms extends their reach far beyond the specific industries in which they operate: by occupying gatekeeping positions as obligatory points of passage, platforms exert infrastructural power that stretches out far into economy and society at large (Yoo et al., 2024; Nieborg & Poell, 2025).

Examples? By controlling access to a massive customer base, Amazon exerts highly asymmetrical gate-keeping power that significantly conditions opportunities and practices of entrepreneurship. Although Amazon can lower barriers to entry and grant easier access to resources and infrastructural services such as payment systems (Yu & Sekiguchi, 2024), the platform orchestrator at the same time unilaterally imposes rules and policies that circumscribe the corridors of entrepreneurial operations. This form of “platform-dependent entrepreneurship” (Cutolo & Kenney, 2021) is further sustained by the practices of venture capital funds who avert financing potential competitors in the “kill zone” in which new ventures are immediately absorbed by incumbent platforms (Kamepalli et al., 2022).

Although Facebook, with already more than 3.065 bn active monthly users (Statista, 2024a), reaches more than a third of the world population, its influence in fact extends even further (Lovink, 2021). Rather than through an ostentatious display of raw corporate power and imperative directives, social behavior is influenced through affordances of the digital rendering and quantification of social relations culminating in a “sociometric subjectivation of actors” (Cardon, 2020, p. 214): actors understand themselves as nodes in elaborate relational topographies who incorporate novel metrics of (digital) reputation and influence into their behavioral calculus (Grabher & König, 2017). Through ceaseless monitoring and alerting, these “vanity metrics” act as “nudges” (Sadeghian & Otarkhani, 2023) for ever more self-conscious strategies to optimize the own social world. The social world, more generally, is reconfigured in a fashion that assures continuous relational labor and, hence, the perpetual production of relational data that fuels the Herculean engine of social media platforms: advertising (Beauvisage et al., 2024).

3 Epochalism II: The Architecture of Platforms

Even if the notion of platforms is pragmatically imagined as heuristic device of a Weberian ideal type, its categorization is confronted with formidable conceptual challenges. Despite their comparatively short history, platforms already have branched off in multiple trajectories and various manifestations as the variegated spectrum of platform typologies testifies. Each of these typologies is construed from a specific disciplinary angle and foregrounds particular platform dimensions respectively: whereas the economic perspective reveals market configurations and the dynamics of network effects (Evans & Schmalensee, 2016), the strategic management perspective is instructive for understanding the orchestration of ecosystems and the co-optation of complementors (Gawer & Cusumano, 2014); viewed from the information systems and technology angle, the digital architecture, data flows and integration through boundary resources come into view (Tiwana, 2013); and whereas the organizational and sociological perspectives focus on governance, collaboration and social (capital) dynamics (Stark & Pais, 2020), the industrial organization perspective provides insights into the integration of digital and physical dimensions as well as value chain transformation (Porter & Heppelmann, 2015).

Each of these disciplinary angles yields specific insights and helps to resolve particular disciplinary puzzles; and yet, as integrative approaches (Parker et al., 2016; Cusumano et al., 2019) have convincingly demonstrated, a combination of different approaches, that is attentive to the different conceptual assumptions, furnishes additional heuristic value: an integration uncovers the intricate interdependencies between various platform dimensions. The transformation of traditional physical assets into smart and connected devices (industrial organization perspective), for example, has direct consequences for the design of governance mechanisms (organizational and sociological perspective), the value co-creation with complementors (strategic management perspective) as well as the configuration of boundary resources (information systems and technology perspective). Very much in the spirit of elucidating interdependencies, the categorization of platforms starts off from an organizational and sociological angle that subsequently will selectively be complemented with other perspectives — whenever integration holds out the prospect of additional insights.

In the organizational strand of inquiry, the most basic and consistent design principles have rather early on been conceptualized as the architecture of platforms³. Viewed through this architectural optic, “the system is partitioned into a set of ‘core’ components with low variety and a complimentary set of ‘peripheral’ components with high variety” (Baldwin & Woodward, 2009, p. 19). The interrelations between core and periphery are governed by interfaces that specify the conditions for the access to the technical infrastructure (Dolata, 2024). This tripartite core/periphery/interface-design also affords a conceptual template for a systematic differentiation of the specific organizational manifestations of this configuration: core and periphery comprise the different “sides” of platforms and networks, and the respective interfaces are built around distinct governance modes as well. Global production networks and social media platforms provide illustrations of network and platform architectures respectively (see Figure 1).

At the core of global production networks, “lead firms” (Henderson et al., 2002) set strategies for production, design and marketing and capture the highest value in the network through branding, innovation and control of intellectual property. The periphery, typically, is made up of a two-sided constellation in which the supply-side faces the customer market. The relationships between core and periphery are governed by a broad spectrum of interface design variants that ranges from market governance through classical arm’s-length transactions with minimal coordination to captive governance with high levels of relation-specific investments and dependence on the side of supplier (Yeung & Coe, 2015). Whereas relations to the market side, in Hirschman’s (1970) terms, are governed by “exit” and, to lower degree, by (brand) “loyalty”, supplier relations are sustained through various practices of “voice” aimed at repairing rather than abandoning ties.

The emblematic type of interface design on the supplier side, phrased differently, is based on relational governance that galvanizes around trustful and long-term relations and affords the social infrastructure for joint innovation, technical support and sharing proprietary knowledge (Smith-Doerr & Powell, 2005). Relational, of course, must not be confused with egalitarian: power relations are an essential ingredient in global production networks (Henderson et al., 2002) and networks more generally (Grabher, 1993, p. 11–12). However, at least in the upper echelons of the supplier pyramids, power asymmetries are flattened out in favor of interfirm partnerships since “these key suppliers are not envisioned as mere satellites orbiting a dominant but benevolent patron, dependent and beholden” (Whitford, 2005, p. 17; emphasis in original).

In a manner analogous to the “lead firm” at the center of the global production network, the “orchestrator” governs the architecture of social media platforms. From its central position, the orchestrator defines and enforces rules, standards and policies that regulate behavior of platform members (Parker et al., 2016), and seeks to incentivize innovation while at same time capturing value through fees, commissions and, crucially, the monetization of data (Gawer, 2014). And here the obvious similarities between network and platform architectures end. Although some consumer platforms with generic offerings, like Airbnb or Uber, for example, also operate two-sided configurations of their periphery, platforms in their archetypical configuration not only incorporate three sides, but also couple different types of social collectivities (see, for example, Kozinets et al., 2008; Grabher & Ibert, 2014): whereas the user-side of social media platforms is populated by communities of laterally related users who share practices, concerns, interests or beliefs (Faraj et al., 2011), the complementor side mobilizes more temporary and relationally rather thin crowds who provide critical contributions to the platform (Malhotra & Majchrzak, 2019) (see Figure 1). To categorize the customers of social media platforms as periphery, although correct from an architectural perspective, however, somewhat obscures the central role of the digital advertising market that, in fact, is funding “much of today’s everyday digital world” (MacKenzie et al., 2023, p. 555; see also Nieborg & Poell, 2025): in 2023, Meta, for example, generated 97.5% of the total revenue of $116.6 bn from advertising (Statista, 2024b).

The decisive difference between network and platform architectures, though, resides in the mediatory elements between core and periphery — as the combination of the organizational and sociological with the industrial organization perspective reveals (see Figure 2). Whereas the interface in network configurations involves relational contracting, core and periphery in the platform context are intertwined through algorithmic interfaces (Gawer, 2021, p. 3–4; Stark & Broeck, 2024). Algorithmic interfaces are not simply passive fasteners that, once in place, remain fixed: rather, they are critical “boundary resources” (Ghazawneh & Henfridsson, 2013), as the jargon has it, that can flexibly be readjusted at the orchestrators’ discretion: whereas relational contracting across network interfaces concedes at least some latitude for negotiations and “voice” to the periphery, the total control of boundary resources by the center leave the peripheries of platform architectures with the terse alternatives of brusque “exit” or wholesale “comply” (see also Dolata, 2024, p. 15–16).

Algorithmic governance, more generally, has shifted from top-down designs, in which the rulesets of algorithms are conceived by human programmers, to bottom-up machine-learning designs in which the algorithm is given a learning rule and trained on large data-sets in order to develop its own rules for decisions and categorizations (Srivastava, 2023, p. 991). With the increasing deployment of AI, algorithmic interfaces acquire agentic capacities: they are no longer limited to endlessly repeating the same processes which are “constant and automatic” (Bourdieu, 2018, p. 37), but are empowered to make independent decisions. The principal accomplishment of algorithmic interfaces, then, is no longer automation but autonomy, and agentic performance supersedes repetitive prescription (Filosa et al., 2025, p. 3)⁴.

Since the overall value of platforms for users (and thus for customers) critically depends on the contribution of complementors⁵, the design of the respective interface is of highest priority for orchestrators: the algorithmic interface has to incentivize innovative contributions by complementors — without relinquishing command which could compromise quality and value capture (Ghazawneh & Henfridsson, 2013; see also Rilinger, 2024). Balancing the conflicting imperatives of generativity and control is the key function of Application Programming Interfaces (APIs) that serve as the central building blocks of the platform architecture (Gawer, 2021, p. 4; see Figure 2): Control over APIs “amounts to control over the platform and its evolution” (Tiwana et al., 2010, p. 680). Construction principles of APIs are provided by Software Development Kits (SDKs) that facilitate the app development process with developer libraries, code samples, testing and deployment tools as well as guides and tutorials. APIs serve as the lingua franca (van der Vlist et al., 2022, p. 1) for the exchange of data and services and, hence, are much more than microscopic technical objects: they have evolved from simple programming interfaces into complex interconnected governance arrangements that exert infrastructural power (Yoo et al., 2024) and, indeed, articulate the processes of the assetization of data more generally (Birch & Muniesa, 2020; Hackfort et al., 2024, p. 2; Sauvagerd et al., 2024, p. 6).

The algorithmic interface between orchestrators and customers is based on a computational adaptation of “mechanism design”. With its origins in (Nobel prize awarded) economic and game theory, mechanism design has become an authoritative science for engineering individual choice and optimizing value distribution in society (Viljoen et al., 2021, p. 2). In contrast to the Hayekian self-organizing market, mechanism design is about the creation of market mechanisms that reveal the preferences of actors through the respective design of “choice architectures like bespoke auctions, markets or voting systems” (Birch, 2023). In its algorithmic version, mechanism design governs the orchestrator-customer interface through “real time bidding” where an “impression” is auctioned in the 200–300 ms between the moment a user navigates to a webpage or opens a mobile app and the moment the advertisement loads on the user’s device (McGuigan, 2019) (see Figure 2). The dominant advertising exchange, Google’s AdX (that conveniently is integrated with Google’s ad server DFP (DoubleClick for Publishers)) runs tens of billions of such algorithmic auctions a day (MacKenzie et al., 2023, p. 563). The bidding for discrete impressions heralds the shift from targeting larger audience segments defined by demographic features towards a “post-demographic” categorization by machine-learning algorithms that afford unprecedented levels of precision targeting (see also Zook & Grote, 2025, p. 6)⁶.

Turning to the orchestrator-user interface, the strategies of orchestrators to maintain near-total control over the interface to users and to preempt any allegations of lawlessness has engendered an entire genre of Terms-of-Service (ToS) agreements that specify the rules on (un)acceptable user conduct or speech, serve as framework contracts establishing the conditions of litigation, and contain IP licenses (Palka, 2025, p. 2). These “uncontracts” (Zuboff, 2019, pp. 220–221) invoke the notion of a public domain that underwrites legal privileges to take (purportedly) raw data, to subject them to processing, and to impose the idiosyncratic understandings of legibility of each individual platform orchestrator.⁷ ToS agreements, hence, “are performative acts of consummation. Together with the technical protocols that structure interactions […] they work to leverage ad hoc and contingent trade secrecy entitlements into de facto property arrangements” (Cohen, 2019, p. 242). More generally, the legal design of ToS is not intended to sideline the law, but rather to catalyze shifts in legal accountability and to create new zones of immunity by advancing novel understandings of legality.

4 Gradualism: Platformization as Process

Inaugurating a new epoch of capitalist development, the design principles of ubiquitous consumer platforms, that were ennobled with regular appearances in movies and tv shows early on, in fact, have morphed into the architectural blueprint of platforms per se. The critique on this new orthodoxy produced by a structural bias of platform research (Menon et al., 2020, p. 364; Butollo & Schneidemesser, 2022; Jovanovic et al., 2022, p. 2; Dolata, 2024, p. 10) drew attention to a strand of research that, initially in relative isolation from the consumer platform research, had elucidated the distinctive features of industrial platforms already (Kenney & Zysman, 2016; Menon et al., 2020; Dolata, 2024; Lerch et al., 2024).

In terms of design principles, the architecture of industrial platforms resembles the basic configuration of consumer platforms, at least at a cursory glance: in a three-partite configuration of sides, a platform orchestrator at the center controls the peripheries comprising users, complementors and customers. The key differences in architectural terms, however, are again to be found in the interfaces between core and peripheries: whereas the core elements of network architectures are integrated through relational contracting, and algorithmic interfaces govern consumer platforms, industrial platforms rely on intricate amalgamations of relational and algorithmic networks (see, for example, Sjödin et al., 2021, p. 16; Butollo & Schneidemesser, 2022, p. 15; Dolata, 2024, p. 37; Madanaguli et al., 2023, p. 11).

A comparison of static architectures makes sense in an epochalist view in which change is perceived as a complete replacement of the existing formation by a new configuration — such as the substitution of networks by consumer platforms. Viewing history through the optics of a diachronic slide show, however, conceals the very dynamics that sets industrial platformization apart from the establishment of consumer platforms. Whereas research on consumer platforms was preoccupied with the disruption of existing markets, the exploration of industrial platforms primarily foregrounds processes of metamorphosis (Van Dyck et al., 2024), framed in terms of “incremental platforming” (Cusumano et al., 2019), “platform co-evolution” (Tiwana et al., 2010), “platform layering” (Thomas & Autio, 2019) or “platform grafting” (Björkdahl et al., 2024).

Instead of aiming at the penultimate representation of the distinctive platform architecture, the pertinent research, then, seeks to conceive platform evolution as an evolutionary process traversing various stages (Lerch et al., 2024) from (1) traditional product provider over (2) digitalized product provider and (3) digital business model provider⁸ to (4) digital platform provider. The usual conception of a smooth transition from step to step, however, should not disguise the fact that these transformations involve lengthy and resource-heavy phases of experimentation (Filosa et al., 2025), tensions and setbacks (Duguma & Bai, 2024, p. 62–63) and, of course, outright failures (Wlcek et al., 2023, p. 4).⁹ A critical engagement with the key tenets of the current orthodoxy (derived from research on consumer platforms) reveals the frictions and constraints in processes of platformization.

5 Agriculture: From Product to Ecosystem

In contrast to the canonical accounts on disruption by heroic new market entrants, the agricultural sector represents a case par excellence of platformization driven by an oligopoly of long-established incumbents (Grabher, 2020, pp. 254–256). Instead of being challenged by a new breed of digital-native start-ups, companies like John Deere, CNH, Bayer or ChemChina that since long dominate global markets for agricultural equipment and inputs (like seeds, fertilizers and pesticides), are able to entrench themselves ever deeper in the sector through “oligopolistic platformization” (Sauvagerd et al., 2024).

And an entire range of barriers-to-entry fortifies the central position of incumbent manufacturers. While the development of complex machinery (like a combine harvester, for example) incurs classical capital barriers-to-entry of considerable upfront investments, data network effects are even more effective defenses against new market entrants (Hagiu & Wright, 2023): the installed base of equipment grants exclusive access to data that are transmitted to the cloud-based platform infrastructure in order to improve proprietary algorithmic models (on seed productivity or weed identification, for example) as well as machinery (in terms of fuel efficiency, wear and tear and reliability) (Hackfort et al., 2024, p. 10).

Moreover, the dominant incumbent manufacturers command over the financial muscle that allows to cross-subsidize the “freemium” pricing strategies that, so far, hardly cover the operating costs of the industrial platforms. Whereas the prospect of ultimate market dominance (and price setting authority) of consumer platforms mobilizes venture capital even across extended loss-making periods,¹¹ the fragmentation of the agricultural sector frustrates any such the winner-eventually-will-take-all aspirations. Currently, and for the foreseeable future, “no platform market leader exists in the ag sector as each sectoral platform promotes its company’s core business, such as selling agrochemicals or machinery” (Hackfort et al., 2024, p. 10). Since the investment motive of market dominance is absent, financing by the incumbent manufacturers has become paramount (Van Dijck et al., 2024, p. 5): John Deere, for example, selects each year a cohort of start-ups to test innovative technologies in real-world customer environments through its Startup Collaborator Program (Deere, 2024); and Leaps by Bayer (Bayer, 2024) is the company’s strategic impact investment unit that has invested in more than 17 startups that provide specific solutions for sensing and improving soil and crop health (Bayer, 2025).

Incumbents, then, command a privileged position in setting up digital platforms and entrenching them ever deeper in the agricultural sector. Agricultural platformization, in fact, is less driven by the escalating dynamics of network effects than by the latitude afforded by the strategic network position that allows for an opportunistic pursuit of tertius gaudens and tertius iungens strategies respectively (on the tertius strategies, see Obstfeld, 2005).¹² Resonating with the phase models of industrial platformization (Sjödin et al., 2021; Lerch et al., 2024; Filosa et al., 2025), the transformation of the agricultural sector has been chronicled as a stepwise process — definitely not devoid of hurdles and backlashes (Sauvagerd et al., 2024; Van Dyck et al., 2024; see also Beinert et al., 2020; Hackfort et al., 2024). John Deere, “the Apple of ag tech” (Klaehne, 2023) due to both its technological leadership as well as its stringently guarded architecture, exemplifies this metamorphosis through the sequence of strategic pivots laid out by Porter and Heppelmann (2015).

The long history of John Deere, dating back to 1837, reads across vast stretches as a chronicle of a provider of innovative, yet stand-alone products (until 1995) that began with the self-scouring plough and culminated in a broad portfolio of equipment, primarily for agriculture, forestry and construction at the beginning of the 21^st century (Deere, 2025a). The shift towards smart products (1996–2005) in which digital components like sensors, controls, microprocessors, and embedded operating systems amplify the capabilities of physical products (Porter & Heppelmann, 2015, p. 5), began at John Deere with the installation of Yield Monitors (1996) that scan the crop of combine harvesters in real-time; it continued with the deployment of StarFire GPS (1998) that enabled sub-inch accuracy for the location and navigation of agricultural machinery for critical tasks like planting and tillage (Deere, 2025b).

The critical steps towards smart and connected products (2006–2012) were taken with the roll-out of the telematics system JDLink (2006) that allows remote monitoring of machinery performance and diagnostics, Machine Sync (2011) for the wireless synchronization of combine harvesters and grain carts, and of the field monitoring system Field Connect (2013) that employs moisture sensors to optimize irrigation (Deere, 2025c; see also Van Dijck et al., 2024, pp. 5–6). These expansions of the digital solutions portfolio were the result of a diversification of collaborative relations: on the one hand, the collaboration with Monsanto (now: Bayer) that aimed at enhancing the compatibility between John Deere planting equipment and Monsanto seed prescriptions widened the range of complementors with incumbent input providers. On the other hand, and more consequential, the collaboration with Microsoft Azure who provided cloud solutions for remote monitoring and diagnostics from 2010 onwards, extended the relational architecture around John Deere with Big Tech.

Amazon Web Services (AWS), Google Cloud and Microsoft Azure, in fact, “form the backbone of the digital infrastructure in digital agriculture” (Sauvagerd et al., 2024, p. 10) serving the major incumbents, start-ups as well as public institutions like the United States Department of Agriculture (USDA): an agricultural version of the “Big Techification of Everything” (Hendrikse et al., 2021), to phrase it in the histrionic lexicon of the disruption debate. AWS has entwined itself in the agricultural sector by, amongst others, creating the curated digital catalog AWS Marketplace that allows agricultural firms to purchase and deploy third-party software and services to build solutions on the AWS platform (AWS Marketplace, 2025). Since Big Tech companies do not have access to the content of the data harvested from the installed base of equipment of the incumbents (Hackfort et al., 2024, p. 12), Google and Microsoft have set up test farms and developed specialized hardware to refine their industry-specific data analytics and algorithms (Sauvagerd et al., 2024, p. 10) — corroborating once again the critical role of domain knowledge.

With the opening up of its Operations Center for third-party developers, John Deere in 2013 took the final step towards integration into one of the largest agricultural system of systems — or ecosystems in contemporary parlance. Collaborations and alliances provided the relational base for extending the spectrum of hard- and software solutions accessible through the John Deere Operations Center (Sauvagerd et al., 2024, p. 5; Deere, 2025a & 2025b). In the domain of software complementors, the alliance with The Weather Corporation (since 2015), for example, provided the preconditions for integrating agronomic data and weather insights into operative prescriptions uploaded to John Deere equipment; the collaboration with Blue River Technology (acquired in 2017) secured access to machine learning and computer vision for precision spraying; and the partnership with Sentera (since 2020) augments crop performance monitoring with AI-driven aerial imagery.

Whereas the software complementor domain is populated by startups that, after years of relational contracting, might be vertically integrated into John Deere (Saroniemi et al., 2022), the segment of agricultural input complementors as well as of infrastructure providers is dominated by the aristocracy of reigning incumbents (see Figure 3):¹³ as to the former, the alliances with Bayer (Monsanto), ChemChina (Syngenta) and DuPont (Corteva) advance agronomic data integration for seed selection, planting and crop protection; as to the latter AWS, Google Cloud and Microsoft Azure provide the cloud infrastructures for large-scale data storage, processing and analytics (Hackfort et al., 2024). In January 2024, John Deere entered a strategic partnership with Starlink (Tita & Maidenberg, 2024), a division of SpaceX (run by Elon Musk, the notorious uncrowned king of the new aristocracy), to extend the reach of its platform services into areas with “rural connectivity challenges” (Deere, 2025d) which comprise around 30% of US farmland and 60% of farmland in Brazil — where the satellite-based services have been rolled out first (Jewett, 2024).

The extensions and diversifications of alliances and collaborative agreements produced a complex platform architecture that (at the end of 2024) interconnected the offerings of more than 200 companies in the device, network and application layer of this ecosystem (Sauvagerd et al., 2024, p. 5; Deere, 2025a & 2025b). If the ecosystem is the engine, data are its fuel, and the assetization of data is its chief construction principle. For the actual implementation of the imperative of data assetization, the technology of “field mapping” plays a critical role (Miles, 2019, p. 4; Grabher, 2020, p. 255). In the John Deere ecosystem, tractors, combine harvesters and drones equipped with GPS receivers and spatial sensors collect data that are synthesized into “boundary maps” that provide the geospatial reference for machinery operations. The further layer of “performance maps” is computed from data collected through the spectroscopic sensors of combine harvesters, these “roving, metal algorithms” (Miles, 2019, p. 8): by monitoring crop yield variables like starch, moisture, protein and fiber, they identify zones of varying crop yield and crop health and thus, put more colloquially, pinpoint problem areas that need particular attention.¹⁴

Stacked upon each other and augmented with weather data (historical and real-time weather conditions) as well as seed and chemical recommendations (based on crop type and agronomic practices), “prescription maps” specify the application of inputs (like seeds, fertilizers or pesticides) for the various zones of the field through “variable rate technology (VRT)”. John Deere’s See & Spray application exemplifies the workings and the performance capability of VRT (Gullickson, 2022). With a massive, and by the installed base of connected equipment permanently updated, database of images, this technology employs advanced computer vision and machine learning to identify and calibrate the difference between weeds and plants and activates the specific nozzle that best correlates with the weed geolocation — approximately 20 times per second while the sprayer moves at an average speed of 12 mph through the field (Dulaney et al., 2023). While traditionally decisions about the dosage of pesticides and herbicides have been made on a field-by-field basis, VRT calibrates dosage to the sub-field level of the individual plant: precision farming indeed (Grabher, 2020, p. 255).

Although increasingly invisible (and deliberately so), the consequences of these most advanced versions of weaving digital dataflows seamlessly¹⁵ into the physical fabric of the equipment, are hard to overlook. These data streams from the sensing and GPS-devices over the edge computing appliances and the modems for data-sharing and analytics in the John Deere Operations Center to the uploading of prescriptions on to the activators on the machines, rely on proprietary software that is subject to license agreements and IP protection (Hackfort et al., 2024, p. 6): data-flows are increasingly black-boxed, and the farmer is increasingly bypassed. “Digital locks” limit how and who can access, repair, and modify the machinery and, thus, hollow out the notion of ownership of the equipment.¹⁶ Rather than owning, farmers instead have to license critical infrastructures and are obliged to share critical data for free in order to leverage the benefits that the equipment providers propagate (Miles, 2019, p. 7; see also Zook & Grote, 2025, p. 102). Resonating with the more recent managerial idée fixe of the inevitable shift from the product to a service logic, this trajectory towards Farming-as-a-Service (FaaS) holds the prospect of increased eco-efficiency — at the price of hollowing out the managerial authority of the farmer and undermining the role of the farm as basic business unit of agricultural operations.

6 Summary and Conclusion: For Boring Nuances

What, then, is the point of the John Deere story? The proliferation of platforms, up until more recently, has been narrated as an epochal shift: the relational governance that had given shape to the network society (Castells, 2004) was dethroned by a novel mode of algorithmic governance that ushered in the new phase of platform capitalism (Srnicek, 2016). The celebration of disruption left no doubt that this shift was anything but gradual, and the authoritative manual for splitting asunder incumbent industries carried the befitting title “Platform Revolution” (Parker et al., 2016). The sweeping generalizations of this epochal shift, however, stand in stark contrast to the selectivity of the research on which these claims are based that, so far, has primarily focused on the consumer- and communication-based internet (Butollo & Schneidemesser, 2022; Jovanovic et al., 2022, p. 2; Dolata, 2024, p. 10).

And here John Deere enters the picture. The present account seeks to contribute to the ever-growing body of research that accumulates compelling evidence for the distinctive realities of platforms in the realm of production and manufacturing (see, for example, Kenney & Zysman, 2016; Menon et al., 2020; Dolata, 2024; Feike & Rösch, 2024; Lerch et al., 2024). Rather than the overnight and wholesale imposition of a new organizational form by a novel breed of digital-native disruptors that rapidly scale to monopolists, the development of platforms in the industrial realm is driven by incumbent manufacturing oligopolists (Sauvagerd et al., 2024; Van Dyck et al. 2024); and instead of a distinct new phase of capitalist development, this research reveals rather messy and tenacious processes of hybridization and transformation. By engaging with the key tenets of platform research (on consumer platforms), the tracing of John Deere’s trajectory from the provider of stand-alone physical products to the orchestrator of a platform ecosystem revealed distinctive features that preclude an outright disruption.

Materiality. Whereas consumption platforms are built around software apps, industrial platformization involves the entanglement of hardware and software into cyber-physical systems (Menon et al., 2020, p. 363) to which neither network effects nor the zero-marginal-cost principle (Rifkin, 2000) apply. Heterogeneity. Industrial platforms continue to operate in specialized sub-markets in which domain knowledge comes at a premium (Sturgeon, 2019; Meier et al., 2024). Market fragmentation limits the number of complementors who, in turn, have a greater say in the development of industrial platforms than in consumption platforms (Dolata, 2024, p. 37). Complexity. Industry platform solutions imply interventions into complex intra-organizational processes as well as multi-layered technical architectures and legacy-systems. Whereas downloading a consumer app is a matter of a couple of seconds, onboarding onto industrial platforms involves lengthy procedures of negotiation and trust-building (Sjödin et al., 2021). Security. To reduce the risk of dependence from a single orchestrator and losing strategic information to complementors, extensive data-sharing agreements have to be negotiated and, in addition to sophisticated NDAs and blockchain technologies, traditional trust has to be built (Mancuso et al., 2024).

For the foreseeable future (if there is such a thing), these features will continue to provide robust barriers-to-entry and barriers-to-scale that fortify the position of incumbent oligopolies of manufacturers on the one hand, and big tech on the other, in shaping processes of industrial platformization. Although Big Tech companies will increase their infrastructural power as obligatory points of passage in data collection and analytics, the “Big Techification of Everything” (Hendrikse et al., 2021) seems more a threat (or a promise), at least for now. The totalizing perceptions of platformization as the unrestrained ravaging of an all-consuming Leviathan might make for a thrilling (and chilling) read. Yet, if an in-depth examination of industrial platformization (as exemplified with the case of John Deere) conveys a more general conclusion, it is this: transformation might unfold in a rather unspectacular fashion of piecemeal adaptations, sporadic frictions and partial drawbacks and failures. True, this sounds like boring “nuts and bolts”; but, then, nuts and bolts are indispensable.