Boxtown is a neighborhood in South Memphis established in 1863 by emancipated former slaves. Since 2024, it is also the home to one of the world’s largest supercomputers, property of Elon Musk’s artificial intelligence company xAI. Amidst politicians’ talk about innovation, development, and growth, and boosterism about the US American South’s “data boom,” Boxtown shows how the push to “win the race” of AI is shaping local landscapes in the United States.1 In Boxtown, the environmental effects of a gargantuan computer accrue with geological obstinacy on top of decades of environmental and health hazards, as well as longer legacies of plantation and racial violence, policy, and urban neglect.
The environmental risks and health impacts of this facility have raised the alarms of journalists, scientists, scholars, and citizen groups.2 Methane turbines churning toxic pollutants around the clock; a community railroaded by the well-funded interests of energy-hungry technology companies; the glitzy language of economic opportunity mobilizing acquiescent city administrators to harbor regulation-averse capital flows. And yet, its advance is unimpeded, aided by a variety of financial, rhetorical, and legal instruments. If the rush to build AI infrastructure is a national phenomenon tied to grandiose aspirations of economic and techno-cultural renewal, its local impacts show it is also unevenly transforming the landscape of the United States.
The computer housed in xAI’s facility near Boxtown is named “Colossus”—a name which computing historians will quickly associate with the storied super computer built in northwest London to support the work of cryptanalysts during World War II.3 Instead of decoding German ciphers, however, the Colossus now standing in the Mississippi river’s eastern margin services xAI, the company that Elon Musk, the world’s richest man, established to train AI models. In a caustic twist of history, while the first Colossus was built to decode and disrupt Nazi messages, the second seems inclined to recast and amplify them. xAI’s most famous product, Grok, is a chatbot which in 2025 made international news for praising Hitler, spewing antisemitic rants, and insistently bringing up the absurd idea of a “white genocide” in South Africa, seemingly parroting the views of its owner.4
That the Mississippi Colossus requires a yearly energy expenditure equivalent to that of hundreds of thousands of homes—and that clouds of methane and formaldehyde are dumped into a dangerously warming atmosphere in the process—prompts questions about the nature and convenience of this technology, and about its role as an accelerator of extreme ideology, wealth concentration, and environmental collapse.5 As Rev. Dr. Heber Brown put it: “The emissions from the facility … are the same whether Grok is generating images of rock bands fronted by cats or posting tweets like ‘The white man stands for innovation, grit and not bending to PC nonsense,’ as it did recently. But polluting the nearby neighborhood of Boxtown in order to generate posts like the latter certainly adds insult to injury.”6
Boxtown’s antebellum history is tied to the agricultural economy of the cotton plantation, which flourished on its rich river valley topsoil thanks to the labor of enslaved people of African descent. The neighborhood is located in Tennessee’s Shelby County, specifically on the land of the former Ensley cotton plantation.7 Convict labor extended the agricultural use of this land after the Civil War and until the early twentieth century, informing the postbellum development of African American neighborhoods. With the decline of the cotton industry after the Civil War, steam plants and steel mills found their way into the area.
Boxtown sits in Memphis’ southern periphery flanked on the west by an industrial zone, which aside from xAI’s facility includes a large steel processing facility for Nucor, one of the country’s largest steel manufacturers. In the North, the neighborhood is limited by the T.O. Fuller State park, another portion of the former Ensley plantation, sold in 1938 to create “the first park open for African Americans east of the Mississippi river.”8 The neighborhood thus has a history of air pollution, health hazards, and resistance that precedes the digital age.9 And yet, in Boxtown, Old South patterns of land use and their associated dynamics of capital accumulation persist and combine today with the interests of a new industrialist class, shaping its urban form.10
AI’s new economy in Boxtown literally dwells in the shell of previous industries. On the east margin of the Mississippi river, xAI’s facility occupies a building previously owned by Electrolux, the Swedish washing machine and dishwasher appliances company. The company sold the facility to xAI in 2024 and transferred its operations to Springfield, Tennessee.11 We might draw a contrast between the affective spheres of these respective technological eras. If powered appliances promised an easier and more enjoyable life by easing the highly gendered toil of domestic labor, the services provided by xAI’s promise to drastically reorganize the market of cognitive labor. But this vision is not accompanied by bucolic or leisurely images of a future without work. Proponents of AI are, for the most part, content with using this technology to put workers on notice.12
The displacement of physical manufacturing in the US by the industrial production of language and images via data and energy-intensive computational processes reveals the country’s broader economic and techno-cultural transition. And yet, from cotton plantations, steam, and steel manufacturing to the current manufacturing of digital data, the neighborhood’s residents have organized. Boxtown has been and continues to be an urbanism of resistance.13
The Environmental Politics of AI in the US
The race to dominate the global economy of AI requires a drastic increase in data storage, hardware manufacturing, and processing capacity. All of this extends the United States’s reliance on power plants using fossil fuels, fundamentally derailing efforts towards a clean energy transition.14 An important portion of these costs are associated with the powering and cooling of data centers.15 A forecast estimates that by 2026, excess energy demands for powering data centers will be comparable to the entire energy expenditure of Japan, up to 1,050 terawatts.16 In addition, a data center’s daily water consumption is in the hundreds of thousands of gallons. Further, material and energy-intensive processes to produce integrated circuits, processors, and semiconductors for AI accentuate problematic labor and mineral extractive dynamics.17 With this complicated picture as a background, wishful voices have emphasized the need to power the data economy with cleaner forms of energy.18 Others have minimized these potential impacts, deferring them, in a form of magical thinking, as something that AI itself might solve in the future.19 However, given the AI industry’s projections of growth, and their unrestricted embrace by the US government, what seems most likely is that “winning the race” of AI simply means relinquishing the goal of reducing carbon emissions and transitioning into clean energy.
xAI’s Southern Colossus should thus be seen in the context of an emerging AI doctrine in the United States. Its relentless march is enabled by a US government dictate to “build, baby, build” computing infrastructure by streamlining the process of issuing permissions for data centers, semiconductor manufacturing facilities, and energy infrastructure by exempting these from environmental regulations.20 Penned by presidential science advisor Michael J. Kratsios, “special advisor for AI and Crypto” David O. Sacks, and the Assistant to the President for National Security Affairs Marco Rubio, the report aligns with the Trump administration’s well documented animosity against environmental protections. Notably, economists have warned that the resulting fast development of AI infrastructure is sustaining the entire US economy.21 Aside from rescuing the coal mining industry, hardware companies supplying the AI industry with microprocessors seem to explain, at least for the moment, US stock market gains, however raising concerns about “circular financing.”22 In addition, with data center construction outpacing office space construction, AI seems to be offering the field of architecture a lifeline in an otherwise declining construction industry.23 Amidst global economic uncertainty, and increasing warnings about a financial bubble, the US seems to have placed its bet firmly—and exclusively—on AI.24
Aside from promising to secure US military, economic, and scientific superiority—as well as an artistic “renaissance” through digital art—the US AI doctrine has important spatial and environmental dimensions. xAI’s unpermitted use of harmful methane gas turbines to power its Memphis facility illustrates that the social and environmental harms of computing infrastructures are not evenly distributed, and disproportionately affect zones inhabited by Black and other underprivileged communities, which experts have described as “sacrificial zones.”25 Alongside these environmental effects, the construction of AI infrastructure entails costs that citizens are already beginning to shoulder. In Virginia, the Carolinas, and Georgia, for example, utilities companies are building natural gas infrastructure aimed at producing more than 20,000 megawatts by 2040, largely for powering data centers in these states.26 If these plans are realized, households in these states will pay even higher electricity rates to subsidize an oversized, fossil fuel–powered grid which heats the atmosphere and largely works to consolidate an increasingly privatized and surveilling data economy. These new land use patterns are rapidly leaving a mark on US landscapes.
In their disproportionate scale, the spatial disclosures of AI infrastructures can accidentally resemble utopian architectures. An image shared by Meta CEO Mark Zuckerberg shows an aerial view of the island of Manhattan with its landmass dominated by a colossal box.27 The image is meant to convey the gargantuan scale of Hyperion, the Meta data center currently under construction in Louisiana. As observed by media scholar Roland Meyer, Buckminster Fuller’s Dome over Manhattan (1959) and Zuckerberg’s Hyperion data center (2025) represent drastically different (albeit similarly megalomaniacal) ecological aspirations.28 While Fuller’s three-kilometer-wide geodesic dome sought to reduce energy consumption, Zuckerberg’s gigantic extruded box is premised on its waste.
Proponents of AI wield maximalist claims about economic growth, intelligence, and geopolitical economic and military dominance in order to gather capital and attention, and to drown concerns about the inconvenient impacts of AI on the environment and labor.29 While ostensibly linked to productivity and competitiveness, it is currently difficult to disentangle AI in the United States from a political project committed to oligarchic power, the surveillance and militarization of civic life, aggressive imperial foreign policy, and racist nation-building fantasies. And all of this is taking place while AI technologies threaten to become the very lens through which we look into and participate in political reality.
The dance of billionaire and state power outlines this mechanics. The picture that has emerged is that of a cadre of obsequious billionaires greasing the wheels of presidential munificence in the form of contracts and tariff exemptions with juicy political contributions and curtsies.30 Despite the critical attention this farcical race has attracted, it seems to be paying off. The same week xAI was forced to issue a public apology for repeatedly bringing up Nazi rhetoric, it announced a $200 million deal with the US Department of Defense (DoD).31 At the same time, the US Chief Digital and Artificial Intelligence Office (CDAO) boasted that similar contracts were procured between the US Department of Defense and other technology giants including Anthropic, Google, and OpenAI.32 And the Department of Government Efficiency (DOGE), a government initiative led by Elon Musk, helped slash around two billion US dollars from the Environmental Protection Agency (EPA), the very government agency in charge of overseeing the AI industry’s environmental impacts.33 Supported by this combination of self-dealing, obsequiousness, disdain for environmental concerns, and technological tunnel vision, the construction of AI infrastructure, and the forms of power that enable it, advances.
Really Large Time-sharing or the Schematic Geography of AI
The spatial configuration of present-day AI infrastructure increasingly distributes computing power away from users, relocating it in privatized, highly polluting super centers such as xAI’s. This recalls an earlier moment in the history of computing, when large mainframe computers provided computing services to numerous terminals distributed across organizations such as university campuses, government offices, or large companies. Known as “time-sharing,” this early form of networked computing aimed at making scarce computing resources available to many users who might not be aware that they are sharing a computer.
Today’s geography of AI, with software services increasingly relying on server-side processing, bears a structural resemblance with the spatial logic of this Cold War–era computing paradigm. However, as we have seen, the distribution of computing resources today takes place at an altogether different scale. The computing superclusters servicing the AI economy are not merely in a different location within a building or campus, but in a different city, country, or continent.34 As a result, users are removed from the material and environmental dimensions of computing, contributing to the obfuscation of its costs and hazards. Extruding a text or an image using an AI service can appear as a smooth, frictionless transaction only because the space, energy, and water these require are removed from users’ experience. Further, by centralizing computing in facilities such as xAI’s Boxtown data center, citizens’ data routinely circulates and is processed in distant jurisdictions, under increasingly privatized conditions, lowering thresholds for surveillance and information manipulation.
In ways that resemble previous eras of technological investment, we are witnessing the coupling of US government and industry in the construction of a new socio-technical infrastructure. In the background of the US’s ongoing plan to deploy AI infrastructure is the historical entwining of the US industry and government sectors throughout the twentieth century. Since the decades following the Second World War, US officials saw computers as useful instruments to secure global militaristic and economic dominance, and to realize a worldview centered on economic growth, open markets, and international competitiveness. Vast sums of federal funds were directed to university research and technology transfer programs aimed at reorganizing the US manufacturing industry around computers.35 These investments outpaced those of any of the other industrialized economies, configuring technological and scientific standards that firmed the United States’s role as a global hegemon, and shaped a distinctly US American “innovation” culture of scientific production, industry collaboration, and design.36 In the 1990s, a comparably large wave of financial and cultural capital placed the internet at the center of the country’s economic doctrine, with its own infrastructural demands. During this long postwar period, technologies such as numerical control, computer-aided design, and the internet helped model a postwar international order after US interests, ostensibly on ideas about productivity, growth, international competition, and open markets. By virtue of their role in standardizing industrial production, commerce, and culture, they can be seen as relatively soft instruments of US empire. The Trump administration’s action plan to “win the race” of AI refurbishes the language of these previous waves of financial and cultural fervor about US technology, but places them within a different economic and historical trajectory inflected by openly protectionist and ethno-nationalistic political sensibilities. The postwar period of economic expansion in the US, characterized by government-sponsored university research, social security, environmental protections, public education, and health, is now, after forty years of political and cultural erosion, on the chopping block, making way for private interests to further encyst and exert greater influence over all aspects of US politics and life.37 As we have seen, the concentration of computing power across networks of privatized super centers is one manifestation of this dynamic.
Computing has never been neutral, but the form of mediation proposed by many AI systems suggests deeper impacts into how knowledge is not just shared but also produced and manipulated. That these processes are increasingly under the control of private actors of incalculable wealth and questionable commitments to democracy, should raise questions about what the US AI doctrine portends not just for the environment but also the body politic in the United States and beyond.
Counteractions
The perception of AI as a frictionless, neatly automated process rests on hiding its material and social substrates and environmental effects. Making these visible has been the important work of citizens, activists, scholars, journalists, and independent organizations interested in advancing the causes of environmental justice and democratic governance. In Boxtown, activist organizations including Memphis Community Against Pollution (MCAP) and Protect Our Aquifer have sounded the alarm about the planning process leading to the construction of xAI’s facility and scrutinized the company’s promises regarding air quality and water use, which has led to an independent air monitoring study.38 A recent protest named “Get Out of Memphis” organized by the student activist group Tigers Against Pollution brought attention to these issues, prompting the deployment of federal law enforcement under president Trump’s pledge to “fight crime in the city.”39
Beyond Boxtown and around the US, citizen groups and non-profit organizations work to increase democratic oversight on computing infrastructures by conducting independent research, mobilizing public opinion, and supporting related legislation. For example, in a 2025 Data & Society policy brief, Tamara Kneese and Maia Woluchem draw on journalistic, academic, and industry research to scrutinize common claims of AI as the catalyst of growth, alternative energy development, and national security, noting that “achieving computing abundance requires exhausting the planetary bounds of our ecological resources, energy, and labor, all of which are required to support this industrial development.”40 The Tech Workers Coalition, a New York–based group, supported a state-level act requiring the disclosure, hearings, and annual reports about the fossil fuel consumption, water, and energy, of all data centers in the state of New York, and that they all transition to renewable energy by 2030. The act, which was not passed, was inspired by examples of data center legislation from Virginia and Germany, and included consumer protections such as community discount plans and preventing “backroom deals” between tech companies and utility companies.41, directed by Tech Workers Coalition, 2025, ➝.] And in partnership with researchers at the University of Michigan, the Michigan Environmental Justice Coalition has released policy and citizen action recommendations based on a review of the impacts of data center construction and energy demands in communities across several states, notably also including the adoption of regulation and transparency approaches inspired by German legislation on this issue.42
While academic and industry researchers have pursued the notion of “Responsible AI,” they have overwhelmingly focused on topics such as implicit bias and disinformation, leaving open questions about the environmental effects of AI’s underlying infrastructure.43 In a similar vein, US linguist and natural language processing scholar Emily M. Bender has observed that researchers working under the banner of “AI Safety” have mostly avoided seriously confronting AI’s social and environmental dimensions—at times simply channeling AI industry propaganda as well as white supremacist fantasies.44 More on point, scholars in the social sciences, arts, and humanities have in recent years approached data centers as subjects of empirical analysis and critical scrutiny. For example Dustin Edwards, Zane Griffin, and Mél Hogan have proposed “Critical Data Center Studies” as a new interdisciplinary area of study critically focused on contemporary computing infrastructures.45 The landscape of critical scholarship they chart draws on concepts and methods from fields including anthropology, history, and science and technology studies (STS), and ranges from studies of the physical infrastructure of the internet, real estate and land use aspects of data center construction, to ethnographies of data centers and other AI infrastructures.46 Architecture and other design disciplines might hold clues to open up spatial and environmental dimensions of this phenomenon.
Science and technology scholar Langdon Winner proposes that certain technologies, by virtue of their scale and configuration, can be seen to enact a politics. Nuclear reactors, he notes, require centralized forms of power for their deployment and operation that might be antithetical to democratic governance.47 As the deployment of the xAI Colossus onto Boxtown suggests, the US AI doctrine and the nation-wide development of data-intensive computing infrastructure both requires and expresses a worldview organized around undemocratic concentrations of capital, space, and political agency. The existence of democracy in the United States and of breathable air in Boxtown are in this way intimately intertwined.
The US South’s “Data Boom” is outlined in Southern Environmental Law Center, “Data Centers Across the South,” Southern Environmental Law Center, 2025, ➝.
See, for example Willy Blackmore, “This Is What They’re Polluting a Black Neighborhood in Memphis For,” Word In Black, July 14, 2025, ➝; Andrew R. Chow, “Inside the Memphis Community Battling Elon Musk’s xAI,” TIME, August 13, 2025, ➝; Dustin Edwards et al., “The Making of Critical Data Center Studies,” Convergence 31, no. 2 (2025): 429–46.
B. Jack Copeland, “Colossus: Its Origins and Originators,” IEEE Annals of the History of Computing 26, no. 4 (October–December 2004) 38–45.
Rev. Dr. Heber Brown, “Artificial Intelligence Furthers Environmental Racism in Black America,” Word In Black, July 28, 2025, ➝.
Emily R. Schwimmer, “A Tale of Two Plantations: The Comparative Development of the ENsley and Davis Plantations in Shelby County, Tennessee and the Museums That Interpret Them” (masters thesis, University of Memphis, 2011), ➝.
“T.O. Fuller State Park,” Tennessee State Parks, n.d., ➝.
“T.O. Fuller State Park.” See also: Chunrong Jia and Jeffery Foran, “Air Toxics Concentrations, Source Identification, and Health Risks: An Air Pollution Hot Spot in Southwest Memphis, TN,” Atmospheric Environment 81 (December 2013): 112–16.
Breaking from previous scholarship, historians trace continuities between enslaver and industrialist economies in the pre– and post–Civil War US American South. Musk’s facility illustrates how these might extend into the digital age. See: James Cobb, “Beyond Planters and Industrialists: A New Perspective on the New South,” The Journal of Southern History 54, no. 1 (1988): 12. See also: Forensic Architecture, “Environmental Racism In Death Alley, Louisiana,” Forensic Architecture, June 28, 2021, ➝.
Sophie Arnius, “Electrolux Group to Divest Memphis Factory,” Electrolux Group, September 21, 2023, ➝.
Analyzing AI through the lens of class, cognitive scientist Hagen Blix proposes we see AI as a “wage depression tool rather than a productivity tool” Hagen Blix, “‘AI Is an Attack from above on Wages’: An Interview with Cognitive Scientist Hagen Blix,” interview by Brian Merchant, October 1, 2025, Substack, ➝.
See, for example, Chow, “Inside the Memphis Community Battling Elon Musk’s xAI.” On community responses, see for example Matthew Gooding, “Fury from Campaigners as Elon Musk’s xAI Gets 150MW for Colossus Supercomputer in Memphis,” Data Center Dynamics, November 8, 2024, ➝; Bracey Harris et al., “Up against Musk’s Colossus Supercomputer, a Memphis Neighborhood Fights for Clean Air,” NBC News, May 16, 2025, ➝. For an outline of Boxtown’s citizens’ historical struggles see: University of Memphis Graduates, “Boxtown: The Land of Broken Promises,” StoryBoard Memphis, September 16, 2019, ➝.
Cathy Kunkel, Data Centers Drive Build Out of Gas Power Plants and Pipelines in the Southeast (Institute for Energy Economics and Financial Analysis, 2025). See also Emma Foehringer Merchant / Floodlight, “Coal Was on Its Way out. But Surging Electricity Demand Is Keeping It Alive — Costing Customers and the Planet,” Floodlight, January 8, 2025, ➝.
Adam Zewe, “Explained: Generative AI’s Environmental Impact,” MIT News | Massachusetts Institute of Technology, January 17, 2025, ➝.
International Energy Agency, Electricity 2024: Analysis and Forecast to 2026 (International Energy Agency, 2024), 170, ➝.
Mark Kidd, “Energy and E-Waste: The AI Tsunamis,” Data Centrre Dynamics, October 24, 2023, ➝.
David Gelles, “Climate: A.I.’s Insatiable Appetite for Energy,” The New York Times, July 11, 2024, ➝.
The White House 2025 Action Plan for AI exempts AI infrastructure from environmental regulations under the framework of the National Environmental Policy Act (NEPA), the Clean Air Act, the Clean Water Act, the Comprehensive Environmental Response, Compensation, and Liability Act, “and other relevant related laws.” See Donald J. Trump, “Winning the Race: America’s AI Action Plan,” The White House, July 2025, ➝.
For example, see: Nick Lichtenberg, “Without Data Centers, GDP Growth Was 0.1% in the First Half of 2025, Harvard Economist Says,” Fortune, October 7, 2025, ➝.
Emily Forgash, “Nvidia’s Massive OpenAI Deal Fuels ‘Circular’ Financing Concerns,” Technology AI, Bloomberg, September 23, 2025, ➝.
An August 2025 Bloomberg report uses census data to observe that data center construction is on track to outpace the construction of office spaces in the United States. See: Ben Holland, “AI Takeoff Turns Data Centers Into America’s New Building Boom,” Bloomberg, November 8, 2024, ➝. This dynamic seems to be offering, for the moment, a lifeline to many architectural firms. See, for example, Kriston Capps, “Can Data Centers Help Keep Architecture Firms Afloat?,” Bloomberg, August 3, 2025, ➝.
See Frick, “AI Is Everywhere But the Jobs Data.” However, as of October 2025, China’s stock market growth shows signs to be also sustained by AI. See Meaghan Tobin, “A.I. Is Driving a Stock Market Rally in China, Too,” Business, The New York Times, October 3, 2025, ➝.
Adam Mahoney, “America’s Digital Demand Threatens Black Communities with More Pollution,” Capital B News, February 25, 2025, ➝. Ben Green, “Growth of Data Centers Requires New Policies to Mitigate Local Community Impacts,” Gerald R. Ford School of Public Policy, Ford School, University of Michigan, July 17, 2025, ➝.
Kunkel, Data Centers Drive Build Out of Gas Power Plants and Pipelines in the Southeast.
Kriston Capps, “Why Meta’s Hyperion Data Center Looks Like It’s Crushing New York City,” Bloomberg, August 24, 2025, ➝.
Roland Meyer (@bildoperationen.bsky.social), “Megalomaniacal Plans,” Bluesky, July 16, 2025, ➝.
The perception of AI as automated helps to obscure the labor that sustains it, often under precarious conditions. See, for example, Billy Perrigo, “Exclusive: OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic,” TIME, January 18, 2023, ➝.
See Andrew Ross Hirsch et al., “The Billionaires’ Row at the Inauguration,” Business, The New York Times, January 21, 2025, ➝; Stephanie Kelly, “Trump’s Inauguration Billionaires, CEOs: Ambani, Zuckerberg, Bezos Attend Church, Ceremony,” Reuters, January 21, 2025, ➝; “Trump’s Inaugural Brings the World’s Billionaire Elites En Masse to DC,” Reuters, January 21, 2025, ➝.
In line with this effort is a new line of “Grok For Government” tools purported to “accelerate America—from making everyday government services faster and more efficient to using AI to address unsolved problems in fundamental science and technology.” See: “Announcing xAI for Government | xAI,” News Release, xAI News, July 14, 2025, ➝.
Their aims are “to accelerate DoD adoption of data, analytics, and AI from the boardroom to the battlefield. CDAO exercised its organic acquisition authority to issue the awards announced today, demonstrating that DoD acquisition can move at the speed of emerging technology and operational necessity.” “CDAO Announces Partnerships with Frontier AI Companies to Address National Security Missio,” Chief Digital and Artificial Intelligence Office, July 14, 2025, ➝.
“EPA Administrator Lee Zeldin Cancels 400+ Grants in 4th Round of Cuts with DOGE, Saving Americans More than $1.7B,” News Release, United States Environmental Protection Agency EPA, March 10, 2025, ➝.
“Data Centers Database,” Data Center Map, 2025, ➝.
See, for example, Paul N. Edwards, The Closed World: Computers and the Politics of Discourse in Cold War America (Cambridge, MA: The MIT Press, 1997); David F. Noble, Forces of Production: A Social History of Industrial Automation (Oxford University Press, 1986).
For a discussion of the roots of design technologies in cold war era politics, see: Daniel Cardoso Llach, “Computer-Aided Design in the United States, 1949–1984: Designing in a Closed World,” IEEE Annals of the History of Computing 46, no. 04 (2024): 10–26. For an in-depth study of innovation in the US, see Matthew Wisnioski, Every American an Innovator: How Innovation Became a Way of Life (Cambridge, MA: The MIT Press, 2025).
For a political history of this period see, for example, Wendy Brown, Undoing the Demos: Neoliberalism’s Stealth Revolution (New York: Zone Books, 2017).
“xAI Campaign,” Memphis Community Against Pollution (MCAP), 2025, ➝.
Austin Johnson, Trinity Williams, the Community Engagement Organizer with the Equity..., October 4, 2025, Photography, ➝.
Tamara Kneese and Maia Woluchem, Data Centers Aren’t the Future of American Prosperity (Data & Society, 2025), ➝.
Introducing Data Centers, Climate, & Labor [Webinar
Terry Nguyen and Ben Green, What Happens When Data Comes to Town, Science Technology and Public Policy (University of Michigan, 2025), ➝.
The energy costs of training AI models are sometimes considered. See, for example, Yolanda Gil and Raymond Perrault, Artificial Intelligence Index Report 2025 (Stanford University Human-Centerd Artificial Intelligence, 2025), 73, ➝.
Emily M. Bender, “Talking about a ‘Schism’ Is Ahistorical,” Medium (blog), August 25, 2025, ➝, offers a useful overview of recent critical work on AI. See also Emily M. Bender and Alex Hanna, The AI Con: How to Fight Big Tech’s Hype and Create the Future We Want (New York: Harper, 2025).
Edwards et al., “The Making of Critical Data Center Studies.”
Ingrid Burrington, Networks of New York: An Illustrated Field Guide to Urban Internet Infrastructure (New York: Melville House, 2016). Daniel Greene, “Landlords of the Internet: Big Data and Big Real Estate,” Social Studies of Science 52, no. 6 (2022): 904–27. Steven Gonzalez, “Cloud Ecologies: An Environmental Ethnography of Data Centers” (PhD diss., Massachusetts Institute of Technology, 2023), ➝. For important, critical perspectives on the materiality of data and the informational dimension of urban infrastructure see: Shannon Mattern, Code and Clay, Data and Dirt: Five Thousand Years of Urban Media (Minneapolis: University of Minnesota press, 2017); Paul Dourish, The Stuff of Bits: An Essay on the Materialities of Information (Cambridge, MA: The MIT Press, 2017); Jean-François Blanchette, “A Material History of Bits,” Journal of the American Society for Information Science and Technology 62, no. 6 (2011): 1042–57.
Langdon Winner, “Do Artifacts Have Politics?,” in The Whale and the Reactor: A Search for Limits in an Age of High Technology (University Of Chicago Press, 1986).
