Data plantation: Northern Virginia and the territorialization of digital civilization in “the Internet Capital of the World”

1 Introduction

Northern Virginia (NOVA) is a region in the eastern US, near Washington DC. In February 2022, Google (worth $1.853 trillion, April 2022) announced that its data center community grant program would provide a $50,000 grant to the Friends of the Slave Quarters, a group supporting the site and educational mission of the Arcola Quarters of the Enslaved, built around 1800 and located at what is now 24,837 Evergreen Mills Rd in Sterling, VA (Loudoun County) (LoudounNow 2022). The Quarters are one of the few remaining vestiges of the material history of human enslavement in the area, thanks largely to its unusual stone construction (DHR 2008). They are part of a larger struggle to preserve sites of Black life and history in greater NOVA, where some counties’ populations were as much as 50 % enslaved in 1860 (without accounting for free Black people) (Graham and Hergesheimer 1861). Google’s head of Community Development for the Eastern US, Laurel Brown, remarked on the occasion, “Located across the street from our data center, The Arcola Quarters of the Enslaved is in the heart of our community” (LoudounNow 2022). Google constructed one of its data center facilities across the street from the Quarters (Figure 1).

Figure 1:

Historic Arcola slave quarters (blue circle) site across the street from Google data center (yellow rectangles) in new Loudoun County development. Source: Google Maps.

Over the last half-century or more, NOVA’s landscape has replaced farmland with massive expanses of Internet infrastructure called “data centers” or, by some there, “server farms” (Freed 2016; Mekouar 2020; Shaw 2020; VEDP 1999). CBS News’ Sunday Morning called NOVA “the heart of ‘The Cloud’” in a 2017 feature, through which they and others claim “up to 70 % of the world’s internet traffic passes every day” (see also, Freed 2016; Sunday Morning 2017; Klein 2019). This essay shows why the juxtaposition above is not mere coincidence and argues that we see parts of the internet economy’s infrastructure as data plantations, which “reproduce the basic features of plantation capitalism,” in Woods’ words above. Therefore, we see them as part of the larger epoch of planetary transformation and crisis some have called the “Plantationocene” (Davis et al. 2019; Haraway 2015). That epoch is defined by contemporary crises of economic polarization and destitution, ecological collapse, political violence, and intersecting forms of human and non-human social domination. While pundits wring their hands about the “existential threat” presented by innovations in “artificial intelligence,” they elide that the Plantationocene has already brought us to the brink. Data plantations should be part of that analysis.

NOVA, within the greater Washington, DC metropolitan area (DMV), includes Alexandria city, Arlington County, and three western counties: Fairfax, Prince William, and Loudoun (see Figure 2). Since the mid-late 20th Century, it has gradually become a sprawling region of ‘McMansion’ housing tracts, ‘town centres’ and strip malls (many with remarkable ethnic food from recent-immigrant communities), some of the nation’s worst traffic, and growing air quality concerns (Ceruzzi 2008; Garreau 1992). The area also became home to a booming technology industry specializing in government and military contracting, and land- and energy-intensive server farms (Ceruzzi 2008; Klein 2019).

Figure 2:

Map of region.

Figure 3:

This map illustrates the search results for data centers (blue) “overlaid” on search results for historic antebellum plantation sites (red) in greater Northern VA. Source: Google Maps.

Against the high dollar spectacle of this industry, critics have begun to describe digital capitalism—what has happened to NOVA and what its urbanism has capacitated—as imperialistic, impoverishing, psychologically dangerous, anti-democratic and anti-urban. For instance, Couldry and Mejias explain, “We are concerned here with the annexation to capital of life processes … evidenced when many of the life streams from which data is extracted for value are not seen by those involved as part of any productive activity” (Couldry and Mejias 2019, 30). While “data” and the automation of management processes, first developed on plantations (Rosenthal 2018), are crucial to large-scale server farming, in this place—with its past in slavery and Civil War so close to the surface—NOVA reveals additional important homologies of one system overlaid on another. These repetitions of form are not incidental, but—as Graham Pickren suggests—indicate how data infrastructure are shaped by a variety of intersecting material processes unique to the particular, historied geographies in which they arise (Pickren 2018). As increasing numbers of data centers continue, even today, to replace what was once agricultural and undeveloped land, their accumulation illustrates how old systems and lingering social struggles for life and liberation are forged through a complex interrelation of land, technology, state power, and corporate influence.

Arcola reveals the literal overlaying of the server farming economy on the terrain of the former agrarian plantation economy. The preserved Arcola Slave Quarters historic site is now surrounded by new data center construction, along with a Sheetz gas station, a self-storage facility, a school, several housing developments, and other small industrial facilities. And the state they occupy is now fighting one of the country’s most prominent battles against education on the topics of slavery and racial capitalism (see EO-1, 2022, “end the use of inherently divisive concepts, including Critical Race Theory, and to raise academic standards”).

2 Towards the data plantation: context, approach and method

This paper is part of a larger project empirically and theoretically exploring NOVA’s digital urbanization, on which (almost unbelievably) much of contemporary global communications and cultures are made and remade. In some ways its goal is (too) simple: to understand how this region of former plantations, farms, and woods, became a global technology capital. But further, inspired by work like Browne’s (2015) intervention in surveillance studies, it aims to bridge gaps and imagine ways to mutually enrich critical data studies (e.g., Pickren 2018) and analyses of racial capitalism (e.g., Chakravartty and Silva 2012; Ranganathan 2020; Taylor 2021). More so, rooted in innovations from the traditions of Marxist geography (e.g., Massey 1994; Mitchell 2000; Gilmore 2007; Harvey 2013), it draws on political economic analysis and synthesizes literatures on digital capitalism (Andrejevic 2019; Briziarelli and Armano 2017; Browne 2015; Fuchs 2020; Sadowski 2020; Wells et al. 2023; Zuboff 2019), digital urban systems (e.g., Foth et al. 2015), and Black/Abolitionist geographies (e.g., McKittrick and Woods 2007; Gilmore et al. 2022) to develop a theory of digital cities within the “Plantationocene,” and vice versa.

The NOVA/DMV region forms a material basis of a digital urbanization, which is globalizing virtually—extensively and intensively constructing possibilities (see Lefebvre 2003, 17)—and virtualizing globally: electronically refashioning representations of space, representational spaces, and spatial practices (see Lefebvre 1992). As the title of this special issue suggests, we see contemporary digital urbanization as “civilizational.” But, as critics of the so-called “Anthropocene” narrative might: we see it as something more specific than an essential phenomenon of “human” life writ large, or as a catastrophic geological force by its “nature.” Instead, digital global civilization represents a hegemonic (naturalized or “atmospheric” [e.g., Adey et al. 2013; McCormack 2017]) (re-)mediation (see Williams 1978) of a set of urban problematics, which Lefebvre suggests predominate in urban (not merely industrial) society:

How can we reconceptualize the urban phenomenon? How can we formulate, classify, and order the innumerable questions that arise, questions that move, although not without considerable resistance, to the forefront of our awareness? Can we achieve significant progress in theory and practice so that our consciousness can comprehend a reality that overflows it and a possible that flees before its grasp? (Lefebvre 2003, p. 15).

The urban problematic is, in this sense, at the heart of the analytical imperative to understand NOVA and its role in global digital civilization.

3 NOVA and the Plantationocene

Beginning in the 1970s and continuing into the present day, NOVA has established itself as a hub for a variety of critical internet infrastructures. To briefly recount: NOVA was the site of an early ARPANet switching point, and home of both MCI WorldCom and America Online (both since purchased by Verizon). Then-Governor James Gilmore described the region as “the Internet Capital of the World” in 1999. By 2003, boosters claimed over 60 % of global Internet traffic was serviced through what has become known as the Dulles Technology Corridor; that figure is currently touted at over 70 % in Loudoun County alone. Likewise, Arlington is to be the new home of Amazon’s HQ2, and the site is already under active development looking forward toward a completion date in 2030.

Today, Loudoun County accounts for roughly 80 percent of the data center inventory in NOVA, the largest data center market in the US. The city of Ashburn, along with neighboring Sterling, hosts the most significant concentration of data centers and related activity, including Equinix, AWS, and Verizon. Much before server farms took over the city, Ashburn was home to the Farmwell and Ashburn plantations, surrounded by similar concerns such as the Belmont plantation.

After the Civil War, Ashburn emerged as a major settlement of free Black people. While Black people in western Loudoun County were primarily employed in lime and marble quarries, agriculture provided sustenance in the eastern part of the county, where Ashburn is located. Eugene Scheel notes that, by 1880 in the county,

[a]s the need for farm labor increases and jobs in the cities decrease, blacks move back to the country. They now comprise 31 percent or 7,243 persons of a total population of 23,634. This number will not be surpassed until the mid-1990s. (Scheel n.d.)

Eventually, there developed 27 main Black villages and hamlets, almost always on poor and wooded land.

This period also coincided with the end of Reconstruction and the violent reversal of Black political power and advancement. Nevertheless, the Black population of Loudoun County strove to build facilities for education and business, in the face of emerging Jim Crow laws. Scheel notes that the year 1900 marked the beginning of a slow decline of the Black population of the county relative to its white residents. This happened against the backdrop of periodic lynchings in the county, the Plessy v Ferguson decision in 1896, a new state constitution in 1902 that institutes a literacy test and poll tax, and the unveiling of the Confederate Soldier statue at the county courthouse in 1908. This emigration of Black people is also part of the story of a decline in agriculture in NOVA, which had already become unprofitable after the abolition of slavery. By 1980, Scheel notes, the “[d]ecline of farms, especially dairy farms using black labor, leads to the first single-digit percentage of blacks to total population: 8.8 or 5,018 among 57,427 persons.” However, the absent Black population still maintain psychic and cultural relationships with Loudoun County. Since the 1990s, events such as a reunion day for “Blacks who have moved away” and a reunion for descendants of enslaved and free people who labored at the Oatlands plantation keep this history alive. The 2015 discovery of a historic cemetery for the enslaved in Ashburn, mostly from the Belmont plantation, is yet another note in the echoes of this history.

As land and people emptied out, Ashburn has come to sit on the world’s densest intersection of fiber networks. This was the product of the ARPANET project in the 1960s and the eventual relocation of the internet peering exchange, MAE-East to NOVA (Ceruzzi 2008; Greene 2022). As Ashburn continues its reinvention into Data Center Alley, and as NOVA continues to be the largest data center market in the world, we occasionally observe the juxtaposition of historic plantation farming and today’s server farms, inviting us to ponder the latter’s lineages. The history of Ashburn itself, as described above, provides one such opportunity.

Examples like Ashburn proliferate, forming the hidden bedrock of even NOVA’s most visible infrastructural affordances. For instance, the Dulles Technology Corridor features an array of defense and technology companies which have become especially valuable to national and state policy makers in post-9/11 United States (Ceruzzi 2008). A key infrastructural component of this agglomeration is the Dulles Airport, providing key international connectivity to the region. The airport was built by razing the predominantly Black settlement of Willard in 1958 (Scheel 2002). Though the village was well past its prime by that time, the 87 area landowners were shocked to receive condemnation letters from the federal government. The landowners deeded 9,800 acres to the government, and moved a historic Black church and its cemetery to the town of Conklin.

West of the Dulles Airport is Arcola, where again we are faced with a juxtaposition of plantation and server farming. The Arcola slave quarters, with its close proximity to Google and Amazon data centers, provides this opportunity. South of Ashburn, and thus away from Data Center Alley proper, the emergence of these data centers indicates the crunch for land in NOVA that can be developed for server farming. As Ashburn and Sterling fill up fast, companies are forced to either lobby for zoning changes to build higher or look outwards to other parts of Loudoun County. The latter is how Google and Amazon have come to rub shoulders with the historic slave quarters.

The Arcola area has already emerged as an outgrowth of Data Center Alley (Figure 3). The hunger for server farming land will continue to bring data centers into direct contact with the plantation history of Loudoun County, especially as it builds over or coexists with specific historical Black settlements, or land once worked by the county’s Black population. Facilitating that contact is also zoning practices in the county, especially as the county Board of Supervisors moves to restrict new data centers in Loudoun (Loudoun County Board of Supervisors 2022).

Constraints in power supply (chiefly handled by Dominion Power currently) and access to water for cooling purposes are two major brakes on server farm expansion. Added to that is growing concern from residents about the exterior appearance of server farms, noise from equipment, and the impact of rising land prices on affordable housing. As a result, the county has moved to no longer allow data center construction, by right, in suburban mixed use (887 acres of vacant land in Loudoun County are currently zoned thus) and urban transit center zoned areas, among others. Instead, data centers will now be restricted to suburban industrial/mineral extraction, transition light industrial, and transition industrial/mineral extraction zones. As we briefly saw in the history of Ashburn, the classification of places for such industrial use can often come about through the emigration of people and the decline of settlements.

Arcola provides a similar example. Its heyday as an agricultural establishment ended with the abolition of slavery. In more contemporary times, the County aimed to develop this area as a mixed-use, town center-style development, under the name, Arcola Center. With this vision remaining unrealized for more than a decade, Google stepped in to purchase the land and change its designated use. That process means that more of the Arcola area has been rezoned to Transition Industrial/Mineral Extraction, featuring some tensions between residents and incoming server farmers (Greene 2018; Lawyers 2014).

The term “Plantationocene” is relatively new and emerges as an experimental alternative to the way that “Anthropocene” and “Capitalocene” tend to omit analytics of race, anti-Blackness, and white supremacy from the study of the planetary effects of anthropogenic processes developed to the point of ecological catastrophe. “It’s more than climate change” explains Donna Haraway of this catastrophe, “it’s also extraordinary burdens of toxic chemistry, mining, depletion of lakes and rivers under and above ground, ecosystem simplification, vast genocides of people and other critters, etc., in systemically linked patterns that threaten major system collapse after major system collapse after major system collapse.” (Haraway 2015, p. 159) She summarizes the conversation that gave rise to including “the plantation” in this analysis nicely in a footnote:

In a recorded conversation for Ethnos at the University of Aarhus in October, 2014, the participants collectively generated the name Plantationocene for the devastating transformation of diverse kinds of human-tended farms, pastures, and forests into extractive and enclosed plantations, relying on slave labor and other forms of exploited, alienated, and usually spatially transported labor. (Haraway 2015, p. 162)

The point is that our specific current catastrophe is connected to specific systemic, political-ecological processes, within which the racialization, control, and exploitation of labor feature centrally. But the concept also ties these processes directly to “the global circulation of people and plants, the simplification of plantation landscapes, and the role of long-distance capital investments in such processes of homogenization and control” as well as forms of extraction and circulation of substances (Davis et al. 2019, p. 4), “scalability and interchangeability” (Tsing 2015, pp. 38–39) describing “the proficiency through which the plantation was able to expand using an established blueprint—the decimation of local peoples and plants, installation of plantation infrastructure on cleared lands, and importation of foreign people and crops” as well as “the ability to exchange one species for another, evident in the plantation practice of substituting cane stock for enslaved people” (Davis et al. 2019, p. 4). Woods (2007, p. 56) notes that permutations of the plantation now characterize “enclosures and reserves; industrial estates and mill villages; free-trade and export zones; enterprise and empowerment zones; ghettos and gated communities; suburbanization and gentrification; game preserves and tourist resorts; pine plantations and mines; and migratory and prison labor.” For Woods, these institutions and modes of production reproduce “the basic features of plantation capitalism: resource monopoly; extreme ethnic, class, racial, and gender polarization; an export orientation; and the intense regulation of work, family, speech, and thought” (ibid). (Davis et al. 2019, p. 6)

There are myriad ways that this concept applies to server farms, media infrastructures, data colonialism, and communication empires. For example, Zane Griffin Talley Cooper has documented how the rare earth mining practices for permanent magnet (amongst a variety of other digital components) manufacturing rely on “peripheral” geographies of toxic extraction and reformulations of global resource flows that service big data’s central, though largely invisible, material apparatuses (Cooper 2021). Similarly, as Parks (2012) indicates, the use of nature metaphors—“the cloud,” data “mining,”—works to obscure the significant environmental costs of data infrastructure; such a critique certainly extends to “farming” servers. Although these scholars do not use the language of the Plantationocene—we would point out that analytic has not yet been used in critical data studies, and remains underutilized in digital geography and related work on internet infrastructure—their projects clearly link transnational histories of race, labor, and resource production/extraction to the production of expropriative economies of both land and data.

3.1 State investment and “resource monopolies”

“Who owns the internet?,” asks Daniel Greene. “As a complex stack of different technologies, protocols, and politics, there is not one answer,” he concedes. But his fascinating work traces the development of the fixed and terrestrial storage, switching, and transmission architectures and infrastructures into an economy of landlordism: private equity funds, financial trusts, and corporations who own and compete for profitability over and through “physical foundations supporting the rest of the internet.” It is “not software developers in control, but firms like Equinix and Digital Realty,” which he calls “internet landlords” keeping real estate at the foundation of the digital dimensions of digital civilization (Greene 2022, 905). Further, Greene explains:

As a class, internet landlords were birthed by the US property state, which tasked private actors with housing the commercializing internet’s network connections in the 1990s. Telecommunications executives and more traditional real estate investors poured into the industry during the dotcom boom. The latter came to dominate after the bubble burst, as private equity cash poured into the market. As Web 2.0 drove up the demand for data storage, the market consolidated, and internet landlords transformed their firm and its cables, servers, and warehouses into financial assets themselves – [REITs.] (Greene 2022, p. 909)

As a set of governmental institutions called the “property state” (Greene 2022; Haila 2000) privatized public infrastructure assets, they formed the institutional foundation for a private internet real estate market.

The early urban development of the Internet in the 1970s revolved around a military funded (though still partnered with research institutions and defense contractors) project called ARPANET, which was composed mainly of clustered sites in three metropolitan areas: Boston, San Francisco, and Los Angeles (Stough 2000; Townsend 2001a, 2001b, 2003). Yet, thanks in large part to federal investment, the Washington, DC metropolitan area soon became a key hub in that network (Gorman et al. 2004; Greene 2022). As Townsend explains: “During the 1970s, growth was rapid and by 1980 ARPANET had a fourth major metropolitan cluster in the Washington, DC area, a region densely populated by military bases, defense contractors, and government information technology contractors” (Townsend 2003). Some of this growth in the 1980s was founded on the Reagan administration’s unprecedented deficit investment in defense growth as well as its emphasis on federal outsourcing. Stough has noted how this spending “focused increasingly on the technical and software attributes of arms and arms systems (electronics, design, systems management) rather than on the armaments themselves …” (Stough 2000, 115). It was through these forms of governmental subsidy to local economies that MCI WorldCom and AOL grew to become Fortune 1000 companies in Northern Virginia (Stough 2000). In the 2000s, Virginia was ranked third of the 50 states in the allocation of federal research and development (R&D) funds (AeA (now CompTIA) 2006a; AeA (now CompTIA) 2006b; Davis 2007; PR Newswire 2006; VEDP 2006).

In 1992, MFS Inc established MAE East through its distributed Virginia Ethernet facilities, providing Internet connectivity to several locally based internet services providers, Alternet, PSI, and Sprint-ICM (Gross 2005). MFS/MAE East was then awarded a grant in 1993 by the National Science Foundation, to which control of the development and shape of the Internet had been handed (dubbed, “NSFNet”). This established the facility as one of four original Network Access points (NAP) to the Internet. In 1994, and through collaboration with NASA, MFS Inc built MAE West in Silicon Valley and MAE products continued to grow through the 1990s (Gross 2005). MAE facilities were eventually consolidated into three key MAE Internet Exchange Points: MAE East (Washington, DC metro area; New York, NY), MAE Central (Dallas, TX; Chicago, IL), and MAE West (San Jose, CA). MFS Inc also continued the operation of MAE LA (Los Angeles, CA), MAE Paris, and MAE Frankfurt (Gross 2005).

Illustrating the functional importance of the morphological forms themselves in the development of these communications networks, MFS Inc and its facilities were acquired by WorldCom and later by MCI, subsequently acquired by Verizon (in 2006), who accrue rents through them for global Internet traffic. With MAE East’s facilities in Virginia distributed between Herndon, Reston, and Vienna, these transformations—allowing direct connections to Amsterdam, Frankfurt, London, and Paris—would ensure the ongoing centrality of Northern Virginia and the DC metro area’s cyber-landscapes in the global flows of information and global cyber-culture. More than half of the world’s Internet traffic flows through the infrastructural landscapes of Northern Virginia.

3.1.1 State and County roles in the production of the “Internet Capital of the World”

In addition to federal investment, individual state governments have also played key roles in regional differentiation and the production of infrastructural landscapes. States compete for the relocation of businesses through various incentives, generally in the form of tax breaks (facilitated by the state’s Economic Development Partnership in Virginia). Very often, the state is understood to itself represent a partnership between public offices and corporations. For instance, Delegate John J “Jack” Rust, Jr. (Republican from Fairfax) was known by his colleagues as “the delegate from AOL” until his electoral defeat in 2001 (Timberg 2000). The Northern Virginia Technology Council’s Political Action Committee (NVTC TechPAC) supported his unsuccessful campaign to reclaim his seat in 2003. James Gilmore, Virginia’s former governor (and later head of the Republican National Committee under George W. Bush), appointed a former executive of Litton/PRC (a public sector IT solutions of the defense IT conglomerate Northrop Grumman) as his “Tech Tzar” to ensure the region’s development as a technology center.

The Virginia state assembly, which considered only 4 bills including the word “Internet” in 1995, debated 79 in 2000. This assembly in 1999 also provided AOL with an $18 million tax break on computer equipment and an additional annual tax savings of $4 million, in a measure forwarded by Del. Rust Jr (Bredemeier 1999). AOL’s original construction plans for a server-farm facility located in the Manassas Battlefield Business Park (see Figure 4) included $400 million in computer hardware. Then-Governor Gilmore provided $500,000 to help Prince William County with “site preparation,” a number matched by county grants (VEDP 1999). Later in 1999, however, Intel, MCI WorldCom, Global Crossing, PSINet, Cable & Wireless, Equinix, and UUNet appealed to the Virginia General Assembly to grant them similar tax breaks. These firms were at the time in the planning stages of their own “server farms” and high-tech campuses, and wanted relief from the 4.5 % state sales tax on their proposed spending of over $500 million (Bredemeier 1999). Referring to the AOL facility, VA Secretary of Technology, Don Upson, at the time exclaimed: “This expansion further cements Virginia’s role as the Internet Capital of the World” (VEDP 1999). Thus, state taxes and the ability to waive them for favored corporations make state government a key variable in the production of Internet infrastructure. Similar initiatives are available at the local level.

Figure 4:

Gates and guard of a server farm, Manassas Battlefield Business Park, by author, 2000.

Prince William County, housing the city of Manassas, has spent considerable effort since the late 1990s to attract technology businesses to settle within its borders. In late 2000, after bringing their first server farm to Manassas six months earlier, AOL was persuaded to locate a second in Gainesville, a town roughly 10 miles west of Manassas on 1–66, bordering Fauquier County. State and county leaders enticed AOL with a package of incentives including tax breaks, job training, and fast-tracking the numerous local permits that were required (Rein and Joyce 2000; VEDP 1999). The local incentives provided for the Battlefield Business Park Site in 1999 were very similar: $557,500 in site preparation/land acquisition/infrastructure improvement grants, workforce training by the Virginia Department of Business Assistance, breaks on business personal property tax, and fee reductions on building inspections (VEDP 1999).

In Gainesville, AOL’s state and county incentives included $1 million in land acquisition and costs for site preparation (37-acres on Linton Hall Road). The county agreed to waive $62,000 in local planning and permitting fees. Likewise, the county also lowered its tax on computer equipment from $1.50 to $1.25 per $100 (all in all saving AOL nearly $400,000 per year) (PWC Department of Economic Development 2000; Rein and Joyce 2000). And, like the $18 million tax break described earlier, which was allotted by the state’s general assembly for AOL’s first server farm, they received a similar break for the second, in addition to a $500 per employee reimbursement for training costs (Rein and Joyce 2000). These combined incentives from state and county governments produced a landscape with a markedly higher rent potential than the surrounding regions, both attracting businesses and creating conditions favorable for relationships of monopoly rent.

In some ways a product of these political interventions, the Linton Hall Road server farm (see Figure 5) is located down what used to be a backwoods side-road in Gainesville. The server farm business does not require a great number of employees: when built in the early 2000s, each employ on average about 125 people (mainly engineers and security staff). However, over-accumulation became salient again within the local server farm market not long after construction was finished, as the digital storage capacity of the computer machinery within the structures no longer required the acreage for physical storage that it once did. Because AOL was able to store and process the same amount of information within a smaller space, it subsequently sold the Linton Road facility to an internet landlord who rents portions of its capacity to various clients. It can be assumed, however, that the remaining benefits of server farms to counties are 1) a boost to the commercial tax base and more importantly, 2) ensuring the infrastructural inertia necessary to bring more technology and telecommunications business to the area. Prince William County illustrates the regionally competitive nature of this economy, which states and counties have managed through competitively providing tax breaks, permit fast-tracks and other location-conditional incentives.

Figure 5:

Speculative build data center on Linton Hall Rd., Gainesville, VA. By author, 2001.

4 Conclusion: abolition and the digital Plantationocene

Besides partially mapping the physical, processual, and conceptutal overlap of digital infrastructure with the plantation, hypothesizing data plantations also helps us reimagine the urban problematic and with it the usefulness of the ‘right to the city’ in this era. Digital capitalism’s well-documented data accumulation by dispossession, unpaid digital labor, debt and digital rentier economies, digital scientific management and labor surveillance, and algorithmic policing are rooted in parcels of land and transformations of the landscape. They become simply problematics of privacy, intellectual property, search and seizure rights, and so on as long as other urban relations of property and dehumanization are taken for granted. Perhaps, in response to the data plantation, abolitionism provides another perspective on digital urban rights: a global end to slavery and its afterlives and a corollary transformation of property rights. Perhaps we imagine struggles for the digital urbanism of the Plantationocene inspired by a speech by Bayley Wyat to a gathering of his fellow Freedmen on December 15, 1866 in Yorktown, VA, recorded (in dialect) the Superintendent of Friends’ Freedmen’s Schools: “I may state to all our friends, and to all our enemies, that we has a right to the land where we are located. For why? I tell you. Our wives, our children, our husbands, has been sold over and over again to purchase the lands we now locates upon; for that reason we have a divine right to the land.” Wyat illustrates in stunning concreteness and poetics simultaneously a rendering of post-Bellum urban relations and the social morphologies that form the plantation. He invokes not simply the relationship between his own personal labor and his proximate plantation economy, but further, the much broader circulation of capital and its parasitic dependence on the fungibility of Black human bodies.

Internet infrastructure does not inherently lend itself to a kind of “mortal urgency”^[1] on the surface: it may even seem banal or trivial when compared to other topics, and we will not try to compare. But perhaps we can still take the Plantationocene seriously, and within that context we can also reimagine the attention and war economies on which so much of NOVA’s wealth is built, including their schools, parks, retirements, healthcare, and so on. There is no such thing as subsistence server farming and the large-scale farming of data has drawn nearly every social practice into its exchanges. The infrastructural landscape of Northern Virginia is at once an idealized symbolization of modern capitalist political economy and an inadvertent memoir of the logistical demands of that productive force built onto the physical terrain. While the (digital) farmlands of Dulles, Reston, Manassas, and Gainesville, VA are not generally thought of in relation to the assumed placelessness of the internet, they are critical to its growth and character as they underpin its function. The clandestine nature and physical structure of Northern Virginia’s server farms, organized by powerful social actors, undermines rooting digital urbanism within a sense of place and historical development that includes the racial capitalist history of the plantation. The purposeful invisibility and enclosure of these infrastructural landscapes work to conceal the spatial class strategies that formed them.

This in mind, the landscapes of the Internet need to be understood on several related levels. First, they are the context and outcome of political and class struggle (Mitchell 1996; Zukin 1993). In NOVA, as in other places, they are managed by capital-allied governmental bodies, built by labor, and inhabited by social actors whose contending interests for accumulation, development, and social reproduction shape and are shaped by the landscapes themselves. Following from this, the militarized shape of the morphological aspects of this infrastructural terrain—and its server farms in particular—also prevents public access to these spaces and preserves its relations of monopoly rent (Graham 2002; Graham and Marvin 2001). Any informational right to the city (Shaw and Graham 2017) must address the issue of landand property as part of the circulation of capital. Finally, while landscape is considered a “medium of exchange” (Mitchell 1994), this aspect of the infrastructural landscape is perhaps amplified. Not only do the fiber optic cables, investment banks, technology production plants, and server farms in NOVA and the DMV area facilitate the accelerated exchange of information, commodities, and capital across space and between places. They also, in their networking and expansion incorporate new spaces into their material system of technological dependency, individuated social interaction, and commodity exchange. But this exchange is built, literally, on the long history of insurgent anti-abolitionism, on the afterlives of slavery in a color-blind context, and in the naturalization of plantation capitalism by the erasure of Black life and anti-blackness in digital urban space. It is only in this context that NOVA can build regional “data center operations” curricula in one instance (Gustin 2023) and, in the next, become embroiled in the fight for banning “critical race theory” and other “equity-minded” approaches in Loudoun public schools.

The period of development this essay discusses continues to have profound effects on the economy and terrain of Northern Virginia. Amazon’s recent decision to locate half of its new headquarters (HQ2) in Arlington, VA, was motivated both by the terrain produced via the processes described here and a more recent history of similar tax credits and job training incentives offered by state and local governments. The region’s infrastructural accumulation and its corresponding “enrichment” of the local labor force have been credited for that decision—as a recent Forbes article (Levick 2019) describes, the area’s “transformation started long before efforts to attract Amazon. The ground had to be tilled first” (Levick 2019). Though Levick’s discussion of the “ground” is largely metaphorical, we suggest instead that consideration of the terrain is a critical element of reproduction of plantation capitalism, and the planetary catastrophe of the Plantationocene.