State-orchestrated hinterlands? Reworking the digital growth machine from China’s East Data West Computing initiative

Introduction: Reading the digital growth machine from a state-led landscape

Rosen and Alvarez Leon's (2025) mapping of the five hinterland types that sustain the digital growth machine offers a generative vocabulary for an economic geography still strongly centered on the urban core. Their typology of materials, manufacturing, fulfillment, computational, and signaling hinterlands makes visible the far-reaching landscapes through which digital capitalism operates, insisting on the materiality of infrastructures that platform-centric accounts render invisible. Building on earlier elaboration of the digital growth machine as an urban political economy engine (Rosen and Alvarez Leon, 2022), the article tracks how urban coalitions carry accumulation strategies into extra-metropolitan space, producing variegated hinterlands across the sector's global production networks.

This commentary extends that argument through China's East Data West Computing initiative (东数西算, hereafter EDWC). Launched in February 2022 by four central agencies, EDWC designates eight national computing hubs and ten national data center clusters, systematically routing eastern computational demand toward western and interior regions, including Ningxia, Inner Mongolia, Gansu, Guizhou, and Chengdu-Chongqing (Zhang et al., 2025). By the end of June 2024, direct investment across the eight hubs had exceeded 43.5 billion yuan and had leveraged over 200 billion yuan in associated investment (Kollar and Stokols, 2026).

EDWC reveals a mechanism of hinterland production that differs from firm-led siting, local boosterism, and urban growth coalitions. I call this mechanism state orchestration. State-orchestrated hinterlands are hinterlands whose location, sequencing, infrastructural form, and legitimacy claims are authored through national spatial planning. This concept extends Rosen and Alvarez Leon's framework by distinguishing state-enabled from state-authored hinterland production. In EDWC, such authorship works through the designation of hubs and clusters, the coordination of eastern demand with western energy and land capacity, investment guidance, and the public narration of regional redistribution. In this sense, EDWC shifts attention from hinterlands drawn into an urban-centered accumulation machine to hinterlands designed as part of a national infrastructure program.

From firm-led siting to state-authored spatiality

Rosen and Alvarez Leon's discussion of computational hinterlands foregrounds the energy and storage infrastructures required by large-scale computation. They note that modern computational activity requires ‘unprecedented amounts of energy and vast storage infrastructures’ (Rosen and Alvarez Leon, 2025). Their example of Granbury, Texas, is especially useful. There, a Bitcoin mining operation is located alongside existing energy infrastructure, drawing on cheap power and permissive regulation while generating local environmental and health burdens. Firms identify favorable energy landscapes, authorities frame digital infrastructure as development, and affected communities bear uneven costs.

EDWC operates through a different spatial logic. Rather than firms discovering cheap energy and aggregating into computational landscapes, the Chinese state manufactures a national spatial pairing between eastern computational demand and western infrastructural capacity (Kollar and Stokols, 2026; Zhang et al., 2025). Its hub-and-cluster geography is planned through central designation, policy coordination, and regional development strategy. In this process, cloud providers, platform companies, AI enterprises, local governments, and infrastructure providers remain crucial actors. State orchestration works by structuring the spatial field in which these actors invest, negotiate, and build. Firms are not thereby reduced to passive implementers; rather, their calculations are routed through a nationally organized geography of infrastructural opportunity.

EDWC therefore adds a further question to the variegated hinterlands framework: through what mechanism is a given hinterland produced? Firm-led, municipality-enabled, and state-authored hinterlands differ in their institutional drivers, spatial forms, political legitimacy, and possibilities for contestation.

Rosen and Alvarez Leon already attend to the state through governance, regulation, and geoinstitutional differentiation, including China's reform-era transformation of the Pearl River Delta (Rosen and Alvarez Leon, 2025). EDWC deepens this insight by showing how the state can produce the region as a digital hinterland, authoring spatial form rather than only conditioning capital's entry.

What kind of hinterland is at stake also becomes clearer. The western data center hubs are urbanizing and infrastructurally strategic places such as Qingyang, Zhongwei, and Gui’an, the last of which has been read as an infrastructure space whose form derives from state design rather than market aggregation (Oakes, 2023). Their hinterland status follows from their functional position within a national computational network, consistent with Brenner's argument that hinterlands are not fixed by an essential non-urban character (Brenner, 2016). EDWC therefore works within the relational definition of hinterland while shifting attention to a different author of hinterland spatiality.

Redistributive discourse, extractive practice

State orchestration changes the politics of spatial injustice. In market-led computational hinterlands, environmental harms often appear as externalities of corporate siting, permissive regulation, and local competition, and affected residents can contest identifiable firms, facilities, or regulatory failures. EDWC arrives through a redistributive grammar. It is framed as a project of regional coordination, green development, national integration, and efficient allocation of computing resources (Zhang et al., 2025). It promises to relieve infrastructural pressure in eastern cities, absorb western energy capacity, and support less developed regions through digital infrastructure investment.

This redistributive narrative functions as a political technology that shapes how extraction becomes legible. Western hubs absorb land, water, energy, and infrastructural burdens for computational demand generated elsewhere. Meanwhile, high-value technical functions may remain tied to eastern firms and metropolitan labor markets. Lower-paid data annotation and AI training work may connect interior regions to coastal AI industries without shifting control over value capture (Wu et al., 2025). What emerges is subordinate incorporation into the digital economy. Such incorporation supports national computing capacity, yet it can leave western regions with the burdens of hosting infrastructure while higher-value design, ownership, and command functions remain concentrated elsewhere.

This incorporation differs from Neel's account of hinterland populations as ‘surplus’ yet integral to accumulation (Neel, 2018). In EDWC, western populations are politically incorporated as the named beneficiaries of national development while remaining economically marginal to the high-value circuits that the infrastructure supports. Their regions are inside the project, assigned to provide energy, land, data infrastructures, and lower-value labor for computational demand concentrated elsewhere.

The politics of contestation are therefore reorganized. Under state orchestration, local costs can be absorbed into national developmental legitimacy. Opposition may be recoded as resistance to regional revitalization, ecological transition, or technological upgrading. The hinterland is used and narrated as uplifted, a narrative central to how state-orchestrated hinterlands work. This is why state orchestration matters analytically: it shapes both the material distribution of infrastructural burdens and the political language through which those burdens are justified.

A state-authored spatial fix

The variegated hinterlands argument invokes the spatial fix to theorize how digital capitalism drives accumulation into new geographies, noting that virtual spaces ‘allow for new sites of accumulation while remaining interdependent with the physical world’ (Rosen and Alvarez Leon, 2025). Following Harvey (1990), this formulation understands the spatial fix as capital's search for new geographies that can absorb surplus value and sustain growth. EDWC retains this logic of geographical displacement but changes its authorship.

EDWC addresses several pressures at once: eastern China's growing computational demand, western energy underutilization, and persistent regional inequality. The fix is designed, sequenced, and institutionalized through national planning, with the state linking computational demand, energy geography, and regional development into a single infrastructural program. EDWC is therefore more than an infrastructure relocation project. It is a territorial solution to the uneven geography of computing capacity, energy availability, and developmental legitimacy.

Beyond its immediate logic, the state-authored spatial fix is layered onto older territorial projects. EDWC echoes earlier west–east infrastructure transfers and development strategies, including the West–East Gas Pipeline, the West–East Electricity Transfer Project, and the Great Western Development program of 2000. It adds a computational layer to existing energy and circulation landscapes. Harlan and Baka's notion of ‘stacked energyscapes’ is useful here as a vocabulary for infrastructural layering: new systems are added to older fossil fuel, energy-transfer, and territorial-development geographies rather than replacing them cleanly (Harlan and Baka, 2024). EDWC similarly stacks digital infrastructure onto prior state projects of energy transfer, regional development, and territorial integration.

Reworking variegation

The broader implication is that variegation should be understood through both hinterland types and hinterland-producing mechanisms. Rosen and Alvarez Leon's typology asks what digital capitalism needs from different landscapes: minerals, manufacturing, logistics, computation, and signaling. EDWC adds another question: who authors the spatial arrangement through which those needs are met?

That question opens onto debates on variegated capitalism (Peck, 2023; Peck and Theodore, 2007) and Chinese state-capital configurations (Wu et al., 2024; Zhang and Peck, 2016). State-orchestrated hinterlands are one possible mode of digital hinterland production within an increasingly fragmented planetary digital economy. The Digital Silk Road, semiconductor industrial policy, AI infrastructure planning, and cloud sovereignty all suggest that digital capitalism is not converging toward a single platform-led geography (Kollar and Stokols, 2026). Multiple state-capital configurations are producing different digital hinterlands.

State orchestration is most likely where central states hold strong planning authority, control over land and energy systems, and mandates for regional integration (Wu et al., 2024). Such capacities enable the pairing of computational demand with infrastructural supply at the national scale, something firm-led siting cannot replicate. Where they are weaker, hinterland production remains more municipality-enabled or firm-led. This distinction is not a rigid typology, but a way to compare how institutional configurations organize the material geographies of digital capitalism. Similar data centers may perform the same computational function while producing different relations among territory, value, and legitimacy. The geography of digital capitalism is planetary in reach but plural in authorship.

Conclusion

The variegated hinterlands typology proposed by Rosen and Alvarez Leon (2025) is a powerful instrument for mapping the material geographies of digital capitalism. EDWC does not invalidate that framework but sharpens it by foregrounding state orchestration, shifting the analytical lens from state-enabled accumulation to state-authored spatial form and reframing spatial injustice in the process. As AI demand accelerates and industrial policy returns to the center of digital infrastructure politics, the geography of digital capitalism will increasingly be shaped by multiple modes of hinterland production. The task is not only to map the hinterlands of the digital growth machine but to ask how different machines produce different hinterlands and with what consequences for territory, legitimacy, and justice.