Store
This module focuses on the materiality of data, building upon the earlier concept of data capture and exploring the physical realities of data storage, media materialism, and the geology of media. This module challenges conventional understandings of data as purely abstract or software-based, emphasizing its physical, environmental, and geopolitical implications.
The tangible realm of data: where is the ‘cloud’?
Data centres are the physical locations where most data ‘lives’ today, making the abstract concept of the ‘cloud’ really tangible. Globally, data centres are concentrated in specific regions, notably North America and Europe, with some presence in China and Southeast Asia, but very few in the rest of Asia, Africa, or South America.Interestingly, the size of data centres is often measured by their electricity consumption in Megawatts, rather than square footage. This is because servers can be stacked vertically, computers require energy to run, and, crucially, data itself is stored in the form of electricity within modern hard drives and memory chips. Northern Virginia, for example, is a major data centre hub, not Silicon Valley. This is attributed to factors such as available land, a reliable power grid, and having more fibre optic cables than anywhere else in the world, largely due to its history as home to early telecom companies that built the internet. Its proximity to undersea cables, large East Coast cities, and transportation infrastructure, coupled with its relative safety from major natural disasters and favourable tax incentives, also contribute to its strategic importance.
Unpacking data’s materiality with theory: Kittler’s Media Materialism
To understand the physical nature of data and the ‘cloud’, we can turn to Friedrich Kittler’s theory of media materialism. Kittler (1997) argued that media studies should extend beyond content or cultural analysis to include the “stuff of the technology that makes media”. All media forms, from ancient stone age rocks to modern smart TVs, internet cables, and data centres, rely on material components to circulate. Kittler stressed the importance of understanding the science and engineering that shape media. For the study of digital media, this means comprehending what happens at the level of the circuit, or hardware, as these are integral to digital media.
Kittler provocatively declared, “There is no software”. He viewed the distinction between software and hardware as an illusion that obscures the true materiality of media.
According to Kittler, all computer code operations, despite our human-friendly language metaphors, ultimately boil down to “signifiers of voltage differences.” This concept draws upon Saussure’s semiotics, where a sign comprises a signifier (the material, physical form) and a signified (the concept). In digital media, programming language acts as the signifier – the code syntax we use to communicate with the computer. The signified, which we cannot directly perceive without instruments, consists of the voltage differences or states of electrical charge in tiny electrical components. By separating software and hardware, we fail to grasp the machine’s true materiality as these electrical charges. Kittler’s core point is that software cannot exist independently from the machine and the materials that create these electrical charges; they are inextricably linked, much like a linguistic sign.
The physicality of binary code and a computer’s electrical components
Computers operate using a binary language, often represented as 0s and 1s. However, these 0s and 1s are not the actual computer language; rather, they are a high-level language humans use to represent the underlying electrical charges within components. Binary acts as an intermediary layer of translation: complex information like words and characters is converted into binary code (0s and 1s), which is then translated into electrical charges. Each character or number, for instance, has a reserved binary code, which can vary depending on the language and hardware. These 0s and 1s are referred to as bits, a type of encoding, not numerical values themselves. For example, each character of a name translates into a combination of 8 bits. Ultimately, these bits are interpreted as electrical charges within the machine’s electrical components, with the specific mechanism depending on the technology.
Different technologies employ various components and materials for data storage. This illustrates that data requires a diversity of minerals and technologies to exist and be stored:
- Transistors: In modern disks, six transistors are needed to hold one bit of memory, representing a charged (1) or uncharged (0) state. While older transistors might be visible, those in modern SSDs are microscopic and typically made of Silicon, or sometimes Germanium.
- Capacitors: Conventional RAM (dynamic memory) stores bits in capacitors, which hold either a charged or uncharged state. Modern DRAM (Dynamic Random-Access Memory) uses tiny capacitors made of silicon nitride, only visible through microscopes.
- Magnetic Hard Drives: Approximately a decade ago, hard drives used magnetic technology, where 0s and 1s represented magnetic attraction. These drives were made of materials like cobalt and aluminum, differing from the silicon-based components of modern storage.
The Geology of Media
Jussi Parikka (2015) introduced the field of Geology of Media. Geology is the scientific study of Earth’s materials, processes, rocks, minerals, fossils, and landforms, helping us understand the planet’s past, present, and future. Applying this to media, Parikka examines how the Earth’ bears the weight’ of our media culture through the metals, minerals, and waste involved in the production and consumption of all forms of media.
Parikka’s Geology of Media extends Kittler’s approach by including not only the technology itself but also the minerals and components that make that technology possible. This makes it a more radical form of media analysis, encompassing extraction, storage, energy consumption, and waste. Crucially, it links these aspects of media to the geopolitics and political economy of Earth materials, highlighting the strategic importance of resources and mineral extraction for countries and governments in relation to data.
Parikka positions Geology of Media as an “anti-McLuhan” theory: Marshall McLuhan (1994) famously proposed that media acts as an “extension of Man,” amplifying human capabilities (e.g., writing extends memory, the telephone extends voice). Parikka challenges this anthropocentric view by centring the Earth as the main unit of analysis, rather than humans. Thus, for Parikka, media is an “extension of the Earth” because it is physically constructed from the Earth’s materials.
Examples of how this perspective informs our understanding include examining the environmental impact of mining and the creation of ‘ghost towns’ - as discussed by Crawford (2021)- the communities and jobs built around data centres, and individual or industry responsibilities in balancing AI development with environmental sustainability (Crawford, 2021; Tung-Hui, 2015).
Conclusion
Data centres and AI technologies demand significant energy and specific minerals that conduct or store electricity:
- Key minerals include Silicon (Si), Silicon Nitride, Germanium, and Lithium (used in batteries). These chemical elements are inseparable from what we call data, as data is fundamentally electricity, and its storage necessitates the mining of these conductive or storage-capable minerals.
- Data is an environmental issue. While technology advances towards greater efficiency and microscopic components, there is a concurrent and powerful trend towards the demand for more data for everything, especially with the rise of AI. This increased need for data and computing power transforms data into a crucial resource, intensifying its relationship with these fundamental minerals, shaping the Earth’s surface in numerous ways.
- This has significant geopolitical and environmental implications, not only for the protection of resources but also for the militarisation of data centres, which are seen as valuable and protected assets.
- This entire process is embedded within a cycle of extraction and growth, characteristic of capitalism. As companies require more data for automated processes like search engines or video recommendations, or AI, the demand for natural resources – the minerals that power this infrastructure – escalates. Consequently, AI, through its reliance on data, is fundamentally tied to this capitalist cycle of growth, where more powerful AI models necessitate more data, equipment, and ultimately, more resources.
Sources and Readings
Crawford, K. (2021). Atlas of AI: Power, politics, and the planetary costs of artificial intelligence. Yale University Press.
Kittler, F. A. (1997). There is no Software. In Literature, Media, Information Systems. Routledge.
McLuhan, M. (1994). Understanding media: The extensions of man. MIT Press.
Parikka, J. (2015). A geology of media. University of Minnesota Press.
Tung-Hui, H. (2015). A Prehistory of the Cloud. The MIT Press.