The Platformization of Everything: From the End of the Like Button to AI Infrastructure in Space

Abstract

A decade after the term was coined, “platformization” has evolved from describing the infrastructural expansion of platforms into other domains to capturing a broader transformation in how platforms organize (digital) life. This article traces this shift from the early social web to today’s AI-centered platform models. The retirement of Facebook’s Like button and Google’s “Suncatcher” space-based AI initiative are used as illustrative examples to demonstrate how platforms continually adapt their expansion strategies. Although the concept has been productively adopted across disciplines, its frequent conflation with the term “digitization” has led to conceptual erosion, weakening its analytical precision. To reclaim its explanatory power, this article redefines platformization as a form of platform-specific “transcoding”: a situated process whereby practices and domains are made “platform-ready.”

Nothing illustrates the transformation of the past decade and half better than Facebook’s announcement that it will discontinue its iconic Like button in February 2026. For critical platform researchers, this marks more than the end of a feature: it reflects how platforms are repositioning themselves and the infrastructures through which they organize our lives. While writing my dissertation on the infrastructures underpinning social media in 2011, I observed how platforms used social buttons to integrate their functionalities into external websites and apps, while simultaneously sending user data back to their own servers. The Like button exemplified the infrastructural expansion strategy of platforms. Platforms were building what we called a “data-intensive infrastructure” by decentralizing their features across the web while recentralizing data processing and valorization on their own platforms (Gerlitz & Helmond, 2013). These buttons were a key element of the material infrastructure through which Facebook constructed its “social graph,” a semantic map of the web formed by tracking what people liked, shared, and commented on, both on and off the platform (Bucher, 2012).

The retirement of the Like button marks the end of an era, as one of the defining symbols of the social web gives way to the rise of AI-powered platforms. Meta no longer needs human signals retrieved from external websites and apps as input for content recommendations or targeted advertising. Instead, the company now draws on behavioral data from within its own closed apps, from third-party apps and devices that integrate with Meta services, from its advertising network, and from AI-generated inferences about user preferences. Predictive modeling has rendered the manual labor of clicking Like obsolete. For Meta, discontinuing the Like and Comment plugins is part of their “platform evolution” because the plugins “reflect an earlier era of web development” (Le & Lin, 2025)—an era when platforms extended outward to capture the open web. For critical platform researchers, this also signals the declining importance of the social graph model. In Meta’s new vision of “personal superintelligence,” the platform’s AI knows what we want better than we do ourselves, which makes explicit signals like the Like button redundant.

Platformization: From Concept to Condition

A decade ago, I coined the term “platformization” to describe platforms’ extension beyond their own boundaries into new areas and domains and their growing dominance as the central computational, infrastructural, and economic model of the web (Helmond, 2015). Central to this logic is how platforms strategically use their programmability to decentralize their infrastructures while recentralizing data that is made “platform-ready” for collection, processing, and monetization. Platformization captures a process of expansion, entrenchment, encapsulation, enclosure, and commodification.

Building on Gillespie’s (2010) influential article on the politics of platforms, I was interested in understanding not only the discursive work that platforms do, but especially the material work they do: how they build and use their infrastructures to expand and integrate into new domains. The concept of platformization, both as a theoretical lens and as an empirical approach, signaled a move from studying platforms themselves to the processes they instigate and how these unfold and affect other actors and domains over time. It describes not only the emergence or dominance of platforms but also the ongoing work through which platforms expand and embed themselves across domains. At the time, I was concerned about how platforms were quietly taking over the web and mobile space by extending their infrastructures and imposing their logics on external websites and apps. More broadly, I was concerned that they signaled the end of the open web through their logics of enclosure and commodification.

Since then, the concept has been used productively to study the expansion of platforms into various societal areas such as health, transportation, education, housing, the social sector, and even the military (e.g., Hoijtink & van der Kist, 2025; Kerssens, 2026; van Dijck et al., 2018; van Doorn et al., 2021). It has also been used to analyze the reorganization of cultural production and the cultural industries around platform logics (Nieborg & Poell, 2018; Poell et al., 2021), to show how platforms increasingly resemble infrastructures (Plantin et al., 2018), and in many other ways. A decade later, the platformization of everything has become a socio-technical condition: Platforms and their logics extend into more and more domains of life, reshaping infrastructures, societies, economies, and everyday practices.

When Everything Becomes Platformized

Over time, however, the term has also become a general descriptor for how platforms mediate and is sometimes used even more broadly to describe a contemporary form of digitization. While digitization emphasizes the conversion of analog to digital, from “atoms to bits” (Negroponte, 1995), it is also widely understood as the adaptation of digital technologies within organizations or domains. Similarly, platformization is sometimes used as simply using a platform within a specific practice, context, or domain. In this sense, anyone, anything, or any action involving a platform has now become platformized, sometimes even when the mechanisms are not platform-based.

However, similar to Vaidhyanathan’s (2011) earlier notion of “Googlization” which critiqued how Google expanded its technologies, logics, and aesthetics into various cultural and informational domains such as the library, platformization should also be understood in relation to platform-specific logics. This means acknowledging that not every platform is the same, nor does it follow the same trajectory or share the same logics, nor does the same platform operate similarly within different regions, domains, or industries. As it becomes a shorthand for general digital transformation, we risk losing its explanatory power if everything is platformized. The “platformization of everything” narrative can also sound technologically deterministic, as if the platform model is an inevitable, unstoppable force destined to consume all economic and social activity, which also minimizes the possibility of alternative digital infrastructures, resistance, or regulatory intervention.

Platform Expansion as Sphere Transgression and Intellectual Monopolization

Understanding platform expansion also benefits from related perspectives that highlight different aspects of how Big Tech gains influence beyond its original domain. Sharon and Gellert’s (2024) notion of “sphere transgression” describes how Big Tech’s advantages (e.g. engineering expertise, data-processing capacity, infrastructure, and capital) gained in one societal sphere are carried over into another, potentially violating the boundaries that protect each sphere’s integrity and giving them influence in areas where they lack domain expertise, democratic legitimacy, or accountability. Another way of understanding Big Tech’s expansionary strategies is through a process that Rikap (2023) calls “intellectual monopolization”, which focuses on how these firms accumulate and monopolize intangible assets such as proprietary algorithms, patents, and unique datasets. From this perspective, Big Tech does not expand simply because it owns and operates key infrastructures, but because its private control over knowledge and data gives it a structural advantage that can be carried into new sectors.

These frameworks help us understand the economic and political dimensions of platform expansion, but we also need conceptual tools and empirical observational techniques to grasp the technical work of making things “platform-ready.” Here, Lev Manovich’s (2001) notion of “transcoding” remains instructive. In his seminal work The Language of New Media, Manovich defines transcoding as one of the five principles of new media: the process through which cultural forms and practices are translated into the logic and language of the computer. When cultural forms enter the computer’s domain, they must follow the “established conventions of the computer’s organization of data” (Manovich, 2001, p. 45). That is, when individuals, companies, sectors, and professional and everyday practices become entangled with platform infrastructures, they are re-expressed in terms of database schemas, metrics, classifications, ranking systems, or data points that fit the underlying platform architecture and business model. Thinking about platformization as a situated, platform-specific form of transcoding helps us move away from broad claims that everything is platformized and toward careful analysis of how particular domains are being made platform-ready, with very concrete technical, organizational, cultural, and economic consequences.

Platformization in Practice: The Work of Third Parties

We therefore need to move beyond general claims to study what van Doorn et al. (2021) call “actually existing platformization”. Their work shows how platformization takes shape differently in sectors such as food delivery, short-term housing rental, and the voluntary sector, where integration depends on local partnerships and situated organizational arrangements. Kerssens (2026) demonstrates how this works in the Dutch education sector, showing how Microsoft collaborates with local intermediaries to embed its services into classrooms. These geographical and sectoral variations are central to de Kloet et al.’s (2019) analysis of the platformization of Chinese society, which shows that “[. . .] platformization is not a uniform process, but follows different trajectories along the vectors of infrastructure, governance, and practice” (p. 254). Poell et al. (2021) and Steinberg et al. (2025) further build on this by contending that platformization manifests through multiple configurations shaped by specific state-market-culture relations rather than following a single, universal model.

This is why, as we argue in our forthcoming book, Platforms: A Critical Introduction (Helmond & van der Vlist, 2026), we need to trace platformization both empirically and historically: examining how specific platforms operationalize their expansion within distinct domains, at specific moments, and through concrete technical and organizational arrangements. Only such grounded analysis allows us to move beyond sweeping claims that everything is being platformized and understand the mechanisms, variable outcomes, and consequences of platform expansion.

A second point we develop in the book concerns how platformization unfolds through collaboration and integration rather than through purely top-down imposition. The term may suggest that platforms simply initiate and enforce the process. However, as we have shown in previous studies, platformization depends on webmasters and app developers incorporating platform features into their websites and apps. It also requires collaboration with all kinds of partners to integrate data, infrastructure, and services into new domains (Gerlitz & Helmond, 2013; Helmond et al., 2019; Helmond & van der Vlist, 2026; van der Vlist & Helmond, 2021). Platformization, we argue, is not merely a technological process of extending platform infrastructure and logics but also an organizational process that unfolds with the help of third parties who are drawn into building out the platform under the promise of mutual benefit.

Big Techification and the AI Turn

Hendrikse et al.’s (2022) idea of the “Big Techification of everything” can be interpreted as intensification of the platformization condition from a political-economic perspective. They describe a future in which a small number of major technology companies become so infrastructural and integral to the economy, society, and the state that they function as a central “Sun” around which other actors orbit and depend (p. 2). That future is already here.

Google’s (2025) “Project Suncatcher” illustrates the scale of Big Tech’s infrastructural ambitions. The project envisions an AI infrastructure that is powered directly by the Sun and uses solar-powered satellites equipped with Tensor Processing Unit (TPU) chips to scale machine learning in space. This is a direct response to the fact that AI now requires so much computational power and energy that current data centers, energy grids, and water supplies are reaching their limits globally. Although this may sound like science fiction, Google plans to test-launch two prototype satellites in 2027. This demonstrates how Big Tech is moving from “platform earth” (Riemens, 2025) to space as the next frontier to sustain the continuous expansion drift of “platform capitalism” (Srnicek, 2016).

This example illustrates how platformization is entering a new phase, as Big Tech companies are focusing their platform models on AI. Their AI platforms operate as integrated ecosystems of tools, infrastructure, and services to develop, train, and run machine learning models. As core “hyperscalers” (Narayan, 2022), the cloud infrastructure and AI platforms of Big Tech have become so central to AI development and use in general that we are now witnessing the era of what we have called “Big AI” (van der Vlist et al., 2024). Even Meta, formerly the social network Facebook, has gradually evolved into an advertising platform and “platform-as-infrastructure” supporting large parts of online social and economic life (Helmond et al., 2019). The company is now repositioning itself around AI to realize its vision of “personal superintelligence” (Le & Lin, 2025) and improve ad targeting and engagement across its apps, by spending hundreds of millions on AI talent and infrastructure.

The retirement of the Like button must be understood in this context. It was one of the key elements that built Facebook’s (now Meta’s) social graph, playing a central role in the News Feed algorithm that ranks posts and recommends people to connect with and groups to join. The decision to discontinue it signals that Meta is gradually decoupling itself from the social graph paradigm and turning toward content-recommendation models similar to those used by TikTok. It is also moving toward AI-powered personal assistants and agents. This transition indicates a broader shift from social networking to AI-powered communication, personalization, and content production and consumption (Helmond & van der Vlist, 2026). The end of the Like button also shows the decreasing importance of the open web as a strategic site for capturing behavioral signals for recommendations and advertising.

The Next Decade of Platformization

Understanding platform evolution is thus crucial because of the way platform companies are continuously reinventing themselves and repositioning their business models. As we outline in our book, the future of platform studies should look into both the present and the future. For example, it should examine the converging relationships between Big Tech and the state, the role of platforms in the military and energy sectors, new and emerging platform configurations such as super apps, and the platformization of AI. Platformization is a process that unfolds over time. To understand how platforms expand, entrench themselves, and acquire power, detailed empirical and historical research is required into how these processes emerge, stabilize, and change over time (Helmond and van der Vlist, 2019, 2026).

Why is this important? The next decade of platformization will not simply be a repetition of the last one. Despite the current dominance of Big Tech, new actors, domains, and regions will be involved. Consider, for example, Nvidia, originally a company focused on GPU chips for video gaming, which is now repositioning itself as a platform company that develops software, hardware, AI models, and services to create an ecosystem for AI and accelerated computing. Consider also the rise of the platform model in the military sector, where alongside traditional defense contractors such as Lockheed Martin or Thales, commercial Big Tech firms and new defense-focused tech startups such as Anduril, Palantir, and Shield AI are positioning their AI-driven software platforms as the new backbone of military decision-making (Hoijtink & van der Kist, 2025). To understand platform power today, we argue (Helmond & van der Vlist, 2026), it is essential to trace these distinct trajectories of platformization with empirical precision and avoid overly general or universalizing accounts.