Digital Infrastructures

The Internet is more than its physical self. It is also everything it touches, and everything it rests upon. While it may have begun in earnest as a couple of mammoth mainframes and a handful of wayward cables draped across an arid stretch of the California coast, what has come to be known as The Internet has burrowed itself into rock, bodies, and atmospheres in manners and scales we do not yet fully comprehend. It seems omniscient and omnipresent, an inescapable feature of contemporary life. The decarbonization of the Internet requires better understanding of its physical composition. The material footprint of the Internet extends far wider than just cables, data centers, and cell phones. The Internet is also the energy grids that support it, the global shipping and logistics spaces through which digital devices are manufactured, and, perhaps most importantly, the vast expanses of raw resource extraction that feed it, as digital living would be rendered impossible without ubiquitous access to critical metals like cobalt, tantalum, and rare earth minerals. 

In the last few years, we have heard Google, Apple, Amazon, Facebook, Microsoft, and countless other tech companies wax poetically about how their operations are either already carbon neutral, or will be by 2030. These are noble goals and will no doubt improve sustainability practices within these companies. However, corporate proclamations and a reliance upon benevolent billionaires are not how we decarbonize the Internet—and they certainly aren’t how we might attempt to do so justly. Rather, this approach to zeroing out The Internet’s carbon emissions will continue to reify and reproduce relationships and infrastructures for profit and the endless expansion of capital, instead of for the general public good.

Tech companies attempt to control the rhetoric of decarbonization and sustainability in two big ways. First, these corporate sustainability initiatives reflect a reliance on our understanding of a carbon neutral Internet as an outcome dependent on big tech companies, and their willingness to sacrifice shareholder earnings on the path to rapid decarbonization. This is a strategic framing, rooted firmly in neoliberal politics, meant to inure trust in our tech titans as the proper stewards of climate futures and to continue diminishing the role of the public sector in regulating and investing in such climate and infrastructural matters. A just decarbonization entails a reconsideration of what the Internet is (materially speaking), how it is held together, where its waste comes from, and where that waste goes. It may also require a reframing of the Internet as a public good, operated like a public utility, with democratic ownership driving its decarbonized future.

Second, for decades, tech companies have incubated strategies to frame e-waste as solely a post-consumer problem, for which responsibility principally lies with consumers to properly discard devices. While tremendous amounts of waste and carbon are produced and released at all points in the tech supply chain, not just on the post-consumer side, none of these waste streams have ever been officially treated as e-waste, but rather as problems for someone else, some other industry, discrete from what the Internet claims to be. This is an object-focused framing, which individualizes the problem, and treats electronic waste as a game of hot potato, shifting responsibility constantly downstream, positing consumers as the central arbiters of the dilemma. However, e-waste is far more than collections of discrete objects (like old laptops, phones, and routers). It is a globally distributed process, diffused across many different industrial sectors.

Understanding e-waste as only or even primarily a problem of too many objects obscures the vast amounts of waste and carbon emissions generated during mining and manufacturing processes that elude most accounting measures. While companies like Apple and Google like to cite vast improvements in energy efficiency and carbon neutrality in the operation of their devices and data centers, these alleged gains have not so much eliminated the waste, just displaced it. A 2020 study illustrates how, in the last decade, the overall carbon output and waste profile of the tech industry has predominantly moved from operational spaces like data centers, to hardware manufacturing and system infrastructure, and that that as much as 86% of the life-cycle carbon emissions from end-user devices such as an iPhone 11 occur during the manufacturing process. How and if this waste gets accounted for and incorporated into sustainability profiles of downstream companies like Apple and Google remains an open question not sufficiently covered in corporate sustainability reports. One clear example of the opacity of these waste streams is the case of the 3.5-inch enterprise hard disk drive (the fundamental building block for cloud storage worldwide). Google may claim its datacenters are 100% carbon neutral, but what does that actually mean? What about the thousands of hard drives in those data centers? As data centers cannot properly operate as data centers without ubiquitous access to enterprise hard disk drives, more attention should be focused on the manufacturing of hard disks, not just their operation in a given data center. For example, internal analyses of process chemistry at hard disk manufacturer Seagate revealed that the volume of process chemicals in their manufacturing stream account for as much as four times the amount that actually end up in finished products. As Seagate maintains a forty percent market share for the entire world’s digital storage needs, it is imperative that this waste stream be addressed and accounted for in discussions about a possible decarbonized Internet. 

The only path to a just, carbon-neutral digital life is through the cultivation of a coordinated, inter-scalar understanding of what, where, and how the Internet is actually comprised and maintained. This means, in part, broadening our concept of the Internet beyond just its connective tissue–all the cables, devices, data centers, and the like. We must develop better methods of understanding how infrastructures of data, energy production, resource extraction, logistics, and labor are entangled nonlinearly across multiple scales–and then we need to build solidarity through these scales. From the mining of crucial minerals like cobalt, tin, and rare earths, to silicon foundries and widespread manufacturing infrastructures, to the massive energy regimes supporting the global data center industry, all the way to the unevenly regulated post-consumer e-waste disposal, reclamation, and recycling industry–all of these processes are deeply bound up with one another.

The Internet is a “transnational infrastructure” of extraction from and for which democratic forms of solidarity must be built. One should see the plight of indigenous activists pushing against a proposed rare earth mine in South Greenland as connected to the climate activism currently fomenting amongst Amazon workers, and we should see both of these as necessarily connected to growing post-consumer e-waste issues in places like Ghana and Zimbabwe. While, at first glance, these disparate struggles seem separated by geographies, industries, and supply chains, they are in reality inextricably entangled. A just decarbonization of the Internet entails upacking these too often unacknowledged connections and build and cultivate solidarity through the material supply relations that uphold and maintain digital life. Thea Riofrancos calls this “supply chain solidarity”, the building of social power by opening and mapping the black box of tech and resource supply chains. More than this, there is a real need to move beyond the pale image of the supply chain (itself a concept invented and championed by management science). This concept limits how one is able to see and organize globally distributed workforces toward climate justice. It promotes a linear, directional understanding of material, capital, and labor flows that does not accurately reflect the heterogeneous, uneven, and often unintended peripheral effects of the world’s logistical relations. To that end, this project attempts to visualize supply relations more volumetrically, where technologies, resource flows, and political economies are all co-produced in a non-linear “matrix” of action and “interplay”. These heterodox supply topologies (be they matrices, spheres, rhizomes, etc.) have no beginning nor end points, but rather illustrate how tying ideas to linear spatio-temporal concepts (like chains) about how stuff moves around the world, embed and naturalize specific, largely incomplete ideas about where and when large socio-technical systems like the Internet are, how they operate, and on what they depend.

As an example, maps of the global data center industry are becoming increasingly common, and the data center has become a popular site of analysis for scholars and journalists grappling with the material footprint of the Internet. However, data centers are not static objects. They are persistently maintained, upgraded, and downgraded, as server racks, hard disk drives, cooling units, and all sorts of other mechanical paraphernalia flow endlessly in and out of their doors. In the case of hard drives, about every two or three years, any given data center’s storage infrastructure is completely new. The metals for hard drives (including rare earth metals, see map) are mined, smelted, and machined into storage units that live short, high-intensity lives, and then become discards, living second and third lives as recycled scrap metal, leaving all sorts of peripheral, toxic traces in their wake. Nanna Bonde Thylstrup refers to this phenomenon as “data out of place”, but perhaps data only appears out of place because we are looking at it in the wrong way. Building and maintaining the infrastructure of the Internet enrolls multiple industries, geographies, geologies, and societies in uneven, unintended ways that the metaphor of the supply chain cannot properly account for. 

An international Green New Deal for the Internet requires a foundational reassessment of what it means to move information through the world, and that begins with breaking with linear metaphors, and embracing the dynamic, volumetric “interplay” of the Internet’s transnational infrastructure. Decarbonization is not just a material process, but an epistemological one as well. It requires a reimagining of how stuff moves around the Earth, and for whom it moves. This applies to digital data, as well as to the materials and people that must be enrolled in maintaining the ubiquitous movement of information on which the world has come to depend.