Infastructuring the last mile: State,Chinese tech,and densification of Ghana's rural network

Abstract

China's infrastructure footprint is expanding to rural Africa with a growing number of last-mile initiatives to address the digital divide. Supported by Huawei's RuralStar base station solution and financed by a Chinese loan, Ghana rolled out a large-scale rural telephony project (RTP) to densify its mobile network in 2020. Drawing from interviews and ethnographic fieldwork, this paper elucidates the multi-layered yet contradictory techno-politics of datafying the last mile in Ghana. Viewing RTP as a means to restore its political legitimacy in the twenty-first century, the Ghanaian state politicizes the digital divide to boost rural development. Telecom operators see RTP as a source of value creation and constantly negotiate the location and datafication of the rural sites. To mitigate geopolitical risks, Huawei strategically de-politicizes the project and reduces its local participation. Approaching RTP as data-infrastructure-in-the-making, this paper illustrates how it assembles political, commercial, and ecological forces to connect the last mile, which entails three types of data relations with regard to state, market, and nature. This paper contributes to infrastructure studies and critical data studies by bringing the environmental politics into the theorization of last-mile datafication. It presents an infrastructural lens to observe the dynamics of Africa–China tech engagement and sheds new light on the intricate entanglements between state, infrastructure, and nature.

This article is a part of special theme on Infrastructure, Labour, & Social Change. To see a full list of all articles in this special theme, please click here: https://journals.sagepub.com/page/bds/collections/infrastructure_labour_social_change?pbEditor=true

Introduction

On July 27, 2024, amid Ghana's presidential elections, a group of angry youth from Kofitiakrom, one of the many Ghanaian villages without internet connections, protested over the lack of a mobile network by destroying school desks newly donated by a political candidate. ¹ Claiming that their immediate priority was a mobile network instead of school desks, the youth threatened not to vote for any candidate until their demands were met. Sparking controversy in Ghana, this incident suggested a strong desire among its rural populations to access the digital world, and meanwhile, a persistent infrastructural challenge to connect rural and remote areas. Widely known as the “last mile” problem, the difficulty of delivering the final leg of services (be they logistics or telecommunications) to end users has troubled many societies due to reasons that can be geographical, socio-technical, economic, and/or political (Blevins, 2021; Deng, 2024; Pollio et al., 2023; Rajadhyaksha, 2011). On the one hand, the lack of affordable technical solutions discourages telecom operators from investing in rural Africa despite the growing interest in capitalizing on “the bottom of the pyramid population” through datafication (Arora, 2016; Avle, 2022; Lu, 2020). On the other hand, the popular discourse of “closing the digital divide” from states, international organizations, and big tech companies has contributed to a surge of last-mile projects aiming to connect the villages of the Global South through various data infrastructures, such as drones, balloons, and mini cell towers (Dave, 2021; Lu, 2025a; Satariano, 2018), whose dynamics are yet to be explored.

Since 2017, Chinese tech giant Huawei has been promoting a light-weight base station solution called “RuralStar” (Figure 1), which can provide 2G–4G services for rural and remote areas in the developing world. ² Compared with traditional cell towers, RuralStar claims to cut the deployment cost by 50% and can be powered by 4–9 solar panels (Figure 2) in off-grid villages (Huawei, 2023). First launched in Ghana, this solution supported the Ghanaian government to roll out an ambitious infrastructure initiative called the Rural Telephony and Digital Inclusion Project (RTP hereafter) in 2020. Financed by a €155 million concessional loan from China Eximbank, ³ the project aims to deploy 2006 RuralStar sites in Ghana to extend the national mobile communication coverage from 83% to 95% (The Business & Financial Times, 2020). Since its commercial use in Ghana, RuralStar has been introduced to over 60 countries (Huawei, 2023), becoming a significant infrastructure that is datafying the world's rural populations.

Figure 1.

A grid site in Eastern Region, Ghana, 2023 (Source: the author).

Figure 2.

A solar site in Eastern Region, Ghana, 2024 (Source: the author).

Compared to urban-oriented 5G technologies that often hit headlines amid rising geopolitical tensions (Tang, 2020), little attention is paid to China's infrastructural power in rural Africa, a region Western tech companies have either ignored or failed to engage with (Dave, 2021; Satariano, 2018). ⁴ This raises questions about why Chinese corporate players are able to tap into these markets and how they are reconfiguring Africa's rural last mile technologically, culturally, and geopolitically. In Ghana, the government not only actively embraces the digital economy but also aspires to go beyond foreign aid to search for alternative development (Ghana Beyond Aid, 2019; Issaka, 2023). In such contexts, RTP serves as a crucial case to explore Ghana's infrastructure practices in the digital era and whether it creates an alternative development space with the involvement of Chinese capital and technology. If we consider the commercial use of solar panels as an example of “neoliberalizing nature” (Castree, 2008), RTP also provides an opportunity to examine the planetary expansion of data infrastructure in tropical Africa.

Drawing from interviews and ethnographic fieldwork, this paper examines the techno-politics of datafying the last mile in contemporary Ghana. The rest of the paper proceeds as follows. The second section reviews the literature and proposes a three-component conceptual framework (state–market–nature) to understand the infrastructure politics of the last mile. The third section historicizes Ghana's digital divide since the colonial era to contextualize the emergence of RTP and Chinese corporate actors in the twenty-first century. The fourth section elaborates on the methodology of the paper. The fifth through the seventh sections illustrate the unfolding of RTP in contemporary Ghana. The eighth section discusses the tensions of multiple data relations RTP has entailed with regard to the state, the market, and nature. The last section explains this paper's contribution to infrastructure studies, Africa–China studies, and critical data studies.

The techno-politics of the last mile: state, infrastructure, and nature

The challenge of connecting the last mile is not new but occurs in the development of almost every communication network, including the postal systems (Blevins, 2021), telegraph (Schwoch, 2018), telephone (Deng, 2024), and the internet (Rosa, 2022). Far from being a technical problem, the last mile is closely linked to the “infrastructural power” (Mann, 1984) of states and corporations in their attempts to render last-mile citizens, territories, and resources “legible” (Scott, 1999) and thus governable or profitable. Attending to the techno-politics of the last mile, this section reviews how infrastructure expansion to the margins – in either a geographical or a political sense – creates distinctive types of data relations with regard to the state (politicization) and the market (commodification). Against this backdrop, this section adds an ecological perspective to make sense of last-mile datafication by treating nature as an agentic actor instead of a neutral background.

Following work on infrastructure and statecraft (Anand et al., 2018; Hughes, 1993; Larkin, 2008), a growing body of research has examined the techno-politics of the last mile from the perspectives of nation-building, citizenship, and political governance (Blevins, 2021; Deng, 2024; Rajadhyaksha, 2011; Schwoch, 2018). A crucial insight from this line of literature is that neither the state nor the last mile is given but constantly in the making when implementing colonial, neoliberal, or developmental projects. As Blevins (2021) documents in his book Paper Trails, the expansion of US Post across the American West accompanied and facilitated nineteenth-century settler colonialism, which seized indigenous peoples’ land on the one hand while embodying the Progressive Era's impulse to deliver universal public services on the other. Faced with sprawling and fast-changing territories, US Post used “the agency model” to extend its rural postal system, delegating local contractors to carry out tasks on its behalf (Blevins, 2021). Despite this, not all rural spaces were integrated either due to geographical constraints or the “spoils system” of the US electoral politics, in which political parties strategically distributed rural post routes as a form of patronage to serve their electoral needs. In this sense, the uneven geography of the US postal network in the nineteenth century both was shaped by and participated in the making of the “modern American state” (Blevins, 2021).

In postcolonial countries such as India, the last mile has to some extent become “a litmus test” for the state to maintain its authority among marginalized citizens and regions (Rajadhyaksha, 2011; Stone, 2011). For Rajadhyaksha (2011), projects that aim to bring radio and television to rural India function as a mode of techno-democracy, where connectivity is translated into “democratic citizenship” – it is a “right” every citizen should have in a democratic country. Essentially, these projects provide a means by which the state redefines or “refurbishes” itself and simultaneously restructures state–citizen–market relationships. Furthermore, the incompetence of the Indian state in providing public services coincides with neoliberal reforms that advocate privatization as a superior solution to allocate goods and resources. In practice, those techno-utopian projects often fail, as market logics tend to contradict with political goals and connectivity alone cannot address deep-rooted social problems (Stone, 2011).

In China, many last-mile campaigns in the twentieth century appeared as development projects to build socialism. Beginning in the 1950s, the newly established Chinese state made multiple efforts to bring radio and telephone to the countryside, which effectively integrated its rural populations into the socialist governance system (Deng, 2024). By the late 1950s, central policies could be delivered to the village level in one day instead of ten; in Jilin Province, it took the statistical bureau only eight to nine hours to collect and calculate the provincial steel output (Deng, 2024). During the Maoist era, connecting the last mile served multiple purposes, including political propaganda, social mobilization, data collection, and agricultural knowledge dissemination, which rendered rural residents and resources more legible to the Chinese state. This also explained why the routes of China's rural telephone networks significantly overlapped with the geographic distribution of grassroots governing bodies such as people's communes and agricultural cooperatives (Deng, 2024). Unlike the agency model used by the US, Maoist China employed a mass mobilization model to enlist active local participation, such as voluntary labor from farmers and workers in deploying telephone lines (Deng, 2024). Here, data relations created by rural communication networks were politicized by both sides – the Chinese state and its rural citizens – during the mass campaign to build socialism.

By contrast, China's last-mile projects in the reform era tend to adopt the public–private partnership model by mobilizing the participation of large corporations (Ting and Yi, 2012), giving rise to new data relations and power dynamics. Take cuncuntong – a series of flagship projects that aim to bring telephone, satellite television, and the internet to Chinese villages – for example. Since the design stage, cuncuntong has entailed intensive negotiations between state and corporate actors to balance political and commercial interests during China's market reforms (Ting and Yi, 2012). For instance, after telecom operators rejected the idea of setting up a universal service fund, China used a “contract responsibility system” to attract their participation, basically turning cuncuntong into a for-profit project. ⁵ Discursively, the reform-era projects become less radical, if not depoliticized, by employing the languages of “poverty alleviation” or “rural informatization,” where infrastructure building is motivated more by economic development than by political governance (Ting and Yi, 2012). In other words, data relations at the last mile are increasingly commodified by market forces – a significant change that has emerged in the reform era.

To sum up, current scholarship focuses on two major actors – the state and the market – to articulate the techno-politics of the last mile: how it becomes a crucial site of political governance or profit-making as states and corporations attempt to increase its legibility through a variety of data infrastructures. This entails two types of data relations regarding the last mile – politicization and commodification. Despite its valuable insights, the anthropocentric approach to the last mile often ignores the materiality of infrastructure due to a lack of an ecological perspective. Although nature is not always absent, it is treated either as a neutral background or as an obstacle in “implementing” certain infrastructures. However, as scholars from multiple disciplines demonstrate, agency is often performed by an array of human and non-human actors since infrastructure expansion is also a process of “territorial engineering” (Mukerji, 2009) by confronting the unruly forces of rivers, mountains, forests, and extreme weather, which may disrupt human intentions to inscribe state or corporate power into nature (Bakker, 2004; Schwoch, 2018).

Drawing on science and technology studies, political ecology, and critical data studies, I contend that nature should be included as an actor equally important with the state and market when theorizing infrastructure-making at the last mile. This is necessary because data infrastructure has become a significant site where contemporary environmental politics unfold (Nost and Goldstein, 2022). Examples include data centers that generate tremendous water, energy, and carbon footprints (Brodie, 2023; Hogan, 2015), lithium extraction to manufacture batteries that power AI-driven devices (Lehuedé, 2025), and solar-based drones or balloons that aim to beam the internet from the sky (Dave, 2021; Lu, 2025a; Satariano, 2018). A series of questions thus arises over the trajectory of RTP: How does its techno-politics unfold in the Africa–China context? What kind of data relations are created or negotiated in relation to the state, the market, and nature? What are the promises and perils of the tropical climate?

As a state-led development project targeting the last mile in the tropics, RTP has the potential to extend current scholarship in two aspects. For one thing, the tropical climate it aims to mobilize presents new opportunities and uncertainties for infrastructure construction. In popular narratives, the abundant solar resource in tropical zones provides not only a flexible energy solution for regions that are not fully electrified but also an opportunity to leapfrog to a green future (Cross, 2019; Günel, 2021). But as scholars have demonstrated, elements of nature, be they wind or water, are often “uncooperative commodities” (Bakker, 2004) that defy easy control. This raises questions over the materiality of infrastructure in the tropics, or what Benitez and Lundberg (2022) call “tropical materiality,” and how it may disrupt the datafication efforts. Hence, unpacking the trajectories of RTP will shed new light on the complexity and contingency of infrastructure-making in the tropics.

For another, the complex state–market relations RTP entails will enrich our understanding of Africa–China engagement. Over the past two decades, China has become Africa's major infrastructure builder and financer, reshaping the development landscape previously dominated by Western donors and international organizations (Gagliardone, 2016). When Chinese companies are tasked to deploy Africa's digital infrastructures, be they fiber optic cable or satellite television, it often triggers debate over whether China is exporting its development model to Africa, which not only essentializes China's infrastructure approaches but also ignores African agency (Corkin, 2016; Gagliardone, 2016; Jureńczyk, 2020). Far from being passive, African countries constantly imagine their own development visions. Ghana, for example, has popularized the slogan of “Ghana Beyond Aid” ⁶ since 2019 to envision a “prosperous and self-reliant” nation that can transcend its historical dependence on foreign aid. Similar to the Indian case, Ghanaian discourses exhibit a great sense of techno-optimism, articulating data infrastructure as a prerequisite for national development in order to galvanize support for the state's digital initiatives (Issaka, 2023). How these digital projects play out in local contexts remains unclear given the lack of empirical research.

Approaching RTP as data-infrastructure-in-the-making, this paper explores how it assembles political, commercial, and ecological forces to develop the last mile. Particularly, I foreground the state–market–nature dynamics in creating data relations and reshaping the infrastructure politics. To better contextualize the emergence of RTP in the contemporary era, the next section historizes Ghana's rural last mile from the colonial era to the recent rise of Chinese actors.

Configuring the last mile in Ghana: A historical perspective

Ghana's telecommunication networks can be traced to the colonial era when the first telegraph line was installed in 1881 between Cape Coast Castle and Elmina Castle, two important stops on the transatlantic slave trade routes where colonial governors resided (Okyere et al., 2018). Since then, more lines were extended to coastal cities, and later, the middle and northern parts of the country. Their geographical distribution displayed a strong urban bias toward colonial settlements with the aim of facilitating colonial administration. Along with the expansion of the British Empire in the nineteenth century, the telegraph last mile was gradually pushed into inland areas, which was not a smooth process but entangled with wars and indigenous resistance. In Ashanti Region, telegraph was used by British colonizers to subdue the Ashantis during the Ashanti Uprising – a series of conflicts between the United Kingdom and the Ashanti Kingdom from 1900 to 1901 (Allotey and Akorli, 1999). Meanwhile, local people often cut down telegraph cables, convinced that they were “magic” lines being used by Europeans to win wars (Allotey and Akorli, 1999).

Being deeply politicized during colonial governance, the telegraph last mile served as the “infrastructures of empire” (Aouragh and Chakravartty, 2016), which, along with roads and railways, facilitated the exploitation and extraction of the then-colony known as the Gold Coast. Furthermore, since the deployment of telegraph lines heavily depended on the availability of roads, there emerged a close parallel between transportation and telecommunication infrastructures under colonial rule (Okyere et al., 2018). This pattern continued in the introduction of telephone lines and fiber optic cables to Ghana in the twentieth century (Okyere et al., 2018). When the telephone trunk routes were built in the 1920s, they followed an important transport corridor, known as the Golden Triangle, linking three major cities – Accra, Kumasi, and Takoradi – while bypassing the majority of rural areas (Allotey and Akorli, 1999). As the colonial state repeatedly embedded its spatial bias into infrastructure planning, it contributed to the plurality of the last mile by converging multiple infrastructural frontiers.

After its independence in 1957, Ghana initiated various infrastructure projects to provide basic services, such as electricity and telecommunications, to its citizens, which often politicized the last mile in the name of modernity and development (Miescher, 2014). Many of those projects, however, failed to achieve their goals. For decades and until the 1990s, Ghana's telecommunication industry was administered by the state-owned Ghana Posts and Telecommunications Corporation (GP&T). Although GP&T rolled out several projects to modernize Ghana's telecom networks, they only generated a modest impact due to poor management, conflicting regulations, and tensions inherent in tied-aid investments (Haggarty et al., 2002). In 1988, about 75% of Ghana's telephone lines were concentrated in ten regional capitals (45% in Accra), which accounted for 10% of the country's population (Frempong, 1996). When the Ghana Cocoa Board requested more rural telephone lines to facilitate its commercial activities, GP&T was unable to respond, forcing the Cocoa Board to set up its own radio station using obsolete equipment (Frempong, 1996). The lack of rural telecom services not only hindered the country's agricultural economy but also challenged the political legitimacy of the postcolonial state, a situation that worsened during subsequent neoliberal reforms.

Ghana started to implement Structural Adjustment Programmes in 1983 and privatized its telecom sector in the 1990s. In 1995, GP&T was split into Ghana Post and Ghana Telecom as separate entities. Ghana Telecom was rebranded as Vodafone Ghana in 2008 and then Telecel Ghana in 2023 when Vodafone and Telecel successively acquired a 70% stake in the company, with the Ghanaian government retaining a 30% stake (Frempong, 2010). To streamline the governing institutions, Ghana established the National Communications Authority (NCA) in 1996 to regulate communications activities and services in the country. NCA was granted the authority to collect 1% of operators’ turnover to set up a universal service fund, which was named the Ghana Investment Fund for Electronic Communications (GIFEC), in 2008. With legal backing from the Electronic Communications Act 2008, GIFEC is mandated to “facilitate the provision of universal access to basic telephony for the unserved and underserved communities in the country,” and it later became the major designer and implementing agency of RTP.

As an “institutional fix” to the digital divide, GIFEC was born in the paradoxes of Ghana's neoliberal reforms. On the one hand, deregulating the market through neoliberal reforms attracted private operators such as MTN, which improved the overall quality and efficiency of Ghana's telecom services. On the other hand, resorting to market forces to address the connectivity gap is a false hope given that commercial operators are reluctant to go to rural areas. In 2002, Accra still accounted for 67.7% of fixed-line subscribers (Akakpo, 2008), suggesting a widening instead of narrowing divide after the neoliberal reforms. In response, NCA mandates that commercial operators deploy base stations in district capitals (around 217 in 2023) as part of the requirements for obtaining a license. But beyond district capitals, operators are free to decide whether to go. Given the lack of a legal mandate for and commercial interest in rural areas, GIFEC becomes one of the few institutional means that can intervene into the provision of telecom services in Ghana, with RTP as its most recent and ambitious initiative. Seen in this light, what GIFEC represents is a series of reregulation efforts by the Ghanaian state after it realized the market flaws in the early 2000s.

Interestingly, it was also in the early 2000s that many Chinese tech companies started entering Ghana to tap into its connectivity gap (Avle, 2022; Lu, 2025b). Huawei entered Ghana in 2001 as a marginal player, and it survived by providing more affordable services and focusing on market segments ignored by bigger players (Personal Interview, July 2024). Furthermore, Huawei stands out for its engagement with state projects, making it more entangled with Chinese capital and state actors. In 2006, when Ghana launched the E-Government Project to extend its national fiber backbone, the project was implemented by Huawei and financed by a US$30 million concessionary loan from China Eximbank, a pattern that repeats in the case of RTP. In many ways, the engagement of Huawei and RTP in Ghana is not a coincidence but a historical product of two forces emerging in the twenty-first century – a new wave of reregulation efforts by the Ghanian state and the deepening Africa–China relations. This also makes RTP a critical case for unpacking the intricate techno-politics of the last mile in contemporary Ghana.

Research methods

This study draws from in-depth interviews and four-month ethnographic research in Ghana. Given that RTP is a telecom network connecting numerous villages (1000 by 2023), researching it in a single field site would be impossible. To explore how this network has emerged and expanded, I adopted the “following-the-thing” strategy widely used in multi-sited ethnography (Marcus, 1995). From 2023 to 2024, I conducted two rounds of fieldwork in eight field sites, including the capital city Accra and seven villages where RuralStar is deployed. Integrating different field sites allowed me to map the actor-network of RTP and trace its complex relations in various settings.

In the first stage (June to July 2023), I followed several major institutions involved with RTP, including GIFEC and its local contractor Ascend, Huawei, MTN, and Telecel. Benefiting from my research experience in Ghana since 2019, I recruited interview participants from these institutions through snowball sampling, most of whom were senior managers or engineers. Through these interviews, I identified the key actors assembled by RTP and accumulated a list of villages (around 20) with RuralStar sites. I managed to visit five of them, including four in Eastern Region and one in Volta Region. In the second stage (July to August 2024), I used purposive sampling to recruit lower-level employees from these institutions by considering their different roles in implementing RTP. I visited two more villages in Eastern Region by following the daily work of two local engineers.

In total, I conducted 18 in-depth interviews with multiple stakeholders in Accra, including regulators, contractors, telecom operators, equipment providers, and government officials. Each interview lasted 40–90 min. The seven villages were selected based on two considerations. First is how they are related to my research questions. Second is their travel distance and whether I have informant access. To explore infrastructural materiality in diverse scenarios, I intentionally selected villages with different locations and types of RuralStar sites. Among the seven selected sites, four are solar-based, two are grid-based, and one uses a wooden instead of a metal pole; the latter was also the first RuralStar site constructed in Ghana. Six sites are located in Eastern Region due to their acceptable travel distance (six hours by car, round trip), with one in Volta Region (eight hours by car, round trip). ⁷ In the villages, I observed the operation of the infrastructure and interviewed villagers about their experience before and after the site was deployed. With the help of local guides, I recruited 27 villagers through purposive sampling by considering their gender (14 male and 13 female), position and occupation (two chiefs, three traders, and 22 farmers), and the distance between their homes and RuralStar sites. Each interview lasted 15–30 min.

Politicizing the digital divide in Ghana: GIFEC and the “third model”

Since its establishment, GIFEC has tried different methods to extend Ghana's rural telecommunications network. According to senior Telecel engineer Anthony, the significance of RTP lies in that it introduced a “third model” for extending rural network under the universal service fund. Previously, there were two feasible models for telecom operators in Ghana. First, the government builds the mast, while operators provide equipment and service. Second, the government builds a complete cell site (mast and equipment) and uses the spectrum of an operator to provide service. Data in this model are directly routed to the relevant operator for management, be it MTN or Telecel. While operators have participated in these two models for years to fulfill their corporate social responsibility (CSR), there are few such sites due to their slim profitability. In the third model, the government builds a complete cell site based on the RuralStar solution, connects it to an existing “macro site” (also called a “donor site”) of any available operator in a nearby town, and then makes the network available for all operators. Unlike the second model, data flowing to the third model are first directed to GIFEC's core network in Accra before being rerouted to respective operators. In Anthony's words, this is essentially “a roaming network” that every operator can use to reach rural users no matter whether they have a macro site available. For instance, with a RuralStar site, an MTN user in the village can make calls by connecting to a Telecel macro site nearby, and vice versa. For operators, this greatly minimizes their investment, as the government now builds, monitors, and maintains the infrastructure. Although the third model still needs to use operators’ spectrum and macro sites, they do not incur a lot of new investment costs but mostly reappropriate their existing resources. As Anthony puts it,

For us [Telecel], whatever investment needed [for the third model] is largely a sunk cost. We’ve already bought the spectrum. It was paid for 50 years. We already have the backhaul [network] and macro sites. We’re using them for our own traffic anyway… The government is trying to incentive us [to go to rural areas]. That's why they build everything in the third model. And you don’t have any choice to say “No” [in that situation].

Advocated by GIFEC, the third model has become the most popular approach to densifying the rural network in Ghana, with 1000 sites deployed by 2023. Operators are effectively mobilized to join this model for several reasons. First is due to political pressure. As explained earlier, the long history of boosting digital connectivity through state-led initiatives in Ghana has cleared legal and institutional barriers for regulators to mandate the major stakeholders. Although my interlocutors did not disclose the details, they suggested the negotiation around RTP was anything but easy. One GIFEC official even claimed that the project would not be possible without the head of the Ministry of Communications (MoC), who, as a female lawyer-turned-politician, played an instrumental role in negotiating with commercial operators. More importantly, when the rural network can be used as a campaign strategy to win votes, the importance of RTP goes beyond CSR to be further politicized by Ghana's electoral politics. That is why the inauguration of rural telephony sites in major towns often hit headlines, usually accompanied by speeches from and group photos with politicians. ⁸ When the first site was deployed in Ashanti Region on November 4, 2020, it was unveiled by the then-President Nana Akufo-Addo (Ghana News Agency, 2020). Sometimes, politicization of the project comes from local communities, who strategically use it to demand digital connectivity, as shown by the Kofitiakrom youth protest mentioned earlier.

Technically speaking, the third model is acceptable (if not attractive) because RuralStar makes rural network investment economically feasible. In fact, the first institution to utilize this solution was not GIFEC but MTN. A major background factor is that, as the largest operator in Ghana, MTN frequently receives complaints over the lack of rural coverage from politicians, chiefs, and users. To protect its corporate image, MTN would build a few rural sites every year through its contractor Huawei, but it was unable to scale this up due to the meager profit. Troubled by this problem for years, the Huawei team decided in 2016 to search for a more sustainable solution. After multiple rounds of prototyping and testing with MTN, a turning point occurred when Huawei proposed using a new technique for backhaul transmission and meanwhile replacing the traditional metal poles with wooden poles, which greatly reduced the cost. In 2017, MTN signed a commercial contract with Huawei to deploy 400 rural sites in Ghana before the name RuralStar was invented by Huawei's PR department. Surprisingly, the return on investment was higher than expected, and it took MTN 2.5 years to recoup the investment in the 400 sites – only half of the predicted time. Encouraged by the success of the MTN project, GIFEC decided to appropriate this technology nationwide, which finally led to the launch of RTP. This move marked not only the scaling up of last-mile datafication in Ghana but also a significant change of data relations as the driving force shifted from the market (i.e., private capital from MTN) to the state (i.e., GIFEC). Although commodifying the data flows is still important to attract telecom operators, political considerations begin to take priority.

Given the large amount of investment needed, GIFEC secured a €155 million loan from China, and the financing agreement was officially signed by GIFEC, MoC, and China Eximbank in September 2020. Huawei was not one of the contracting parties but played two major roles in the process: providing detailed reports to justify the economic feasibility of RuralStar, which helped GIFEC secure the loan, and serving as an equipment provider after the project was rolled out. In 2012, Huawei caused controversy when Ghanaian media alleged it provided sponsorship to the then-ruling party National Democratic Congress in exchange for implementing the E-Government Project, which contravened Act 574 of Ghana. ⁹ During our conversations about RTP, my Huawei interlocutors never mentioned this allegation, but a few times, they emphasized that Huawei wanted to keep “neutral” in Ghana. When I asked Wang, a Huawei Chinese engineer, whether villagers know “this is a project done by Huawei,” he corrected me:

This is not a Huawei project. It's a government project. GIFEC contracted the installation to a local company [called Ascend]. So, it's local guys who are installing the sites at the village level. Huawei Chinese engineers will not show up in the villages. It's not good for us…We’re just an equipment provider. We have to be seen as neutral. We don’t want to show preference to any political party. There is a risk there, as you don’t know which party will win in the next election.

Claiming a neutral stance in a state-led project sounds self-contradictory, but it signifies Huawei's desire to strike a delicate balance between its commercial and political interests in the face of increasing scrutiny from both Western and African states. Discursively, Huawei appropriates Ghanaian state narratives such as “the digital divide,” “Ghana Beyond Aid,” and “ICT for Accelerated Development” to articulate RTP, while downplaying Chinese state narratives that often link the project with the Belt and Road Initiative. During my interview with Liang, Huawei's Chinese top manager in Ghana, he claimed that “RuralStar has nothing to do with the Belt and Road Initiative, and we don’t want to be associated with it.” By emphasizing the “commercial” nature of the project, Liang tried to keep a distance from highly politicized Chinese state discourses. Practically, Huawei mitigates the (geo)political risks by limiting its role to equipment provider, but its low level of participation increases the difficulty of maintaining the infrastructure, which will be explained in detail later.

Locating the sites: selective densification and datafication

Instead of building a brand-new network, the core idea of RTP is to densify existing network by constructing more rural sites and then connecting them with the closest macro sites. In 2021, Ghana Statistical Services reported a coverage gap in almost 12,000 communities. ¹⁰ Selecting and locating 2006 sites across the country thus becomes a challenging task for GIFEC that entails considerable coordination and negotiation. Generally, GIFEC identifies suitable communities for RTP based on several criteria. The top priority, according to GIFEC official Ahaji, goes to so-called “green fields”, areas that have never been connected before. To some extent, green fields are blind spots of Ghana's neoliberal reforms, whose urban bias is being reversed by GIFEC through its re-politicization and reappropriation of the last mile.

The second factor is population size. Despite GIFEC wanting to prioritize villages with larger populations in the hope of getting more data traffic, there are not many such villages left after multiple rounds of market-oriented selections over the past years. When MTN conducted the 400-site project in 2017, many villages it selected were large or medium-sized. An MTN manager called Raymond joyfully mentioned to me that around 12% of the sites quickly became congested, which prompted MTN to upgrade them into macro sites in order to capitalize on the rising data traffic. By contrast, many villages left for GIFEC in 2020 were small or medium-sized, with a population size below 1000. Some of the villages I visited only had 200 to 500 residents.

Furthermore, planning a rural site depends on the availability of electricity, roads, and macro sites. The interdependency of multiple utility infrastructures makes the last mile not a single point but itself a network. Repeating patterns since the colonial era, GIFEC prefers locations with access to main roads, which can lower transportation and maintenance costs. Since telecom equipment requires an uninterrupted power supply, GIFEC prefers connecting the site to the national grid. Only when the community is off-grid will a solar site be deployed. By July 2024, around 40% of the completed sites in Ghana were solar-based, usually located in more remote and isolated spots than grid sites.

After a village is shortlisted, the local contractor Ascend will conduct a “drive test” to access the technical feasibility of deploying a RuralStar site. This step requires a lot of data collection and calculation, such as of the distance to macro sites and estimated data traffic. Only after the simulation indicates the proposed location is both technically and economically feasible does GIFEC move to the next step – acquiring the land permit for the site. In most cases, GIFEC does not buy the land but only seeks the endorsement by landlords, mostly chiefs. For this step, GIFEC must work with district assembly members, i.e., political heads for each district who can help communicate or negotiate with the chiefs. Aware of land disputes that happened in previous projects, GIFEC carefully designed an agreement for chiefs to endorse before Ascend could kick off the construction.

Unlike the local hostility toward telegraph lines in the colonial era, base stations are desired last-mile infrastructure in contemporary Ghana. The tension is less about whether to deploy a site or not than about which village should be prioritized. ¹¹ Since the coordination involves chiefs and various levels of government agencies, it is inextricably intertwined with the country's electoral and tribal politics. Several GIFEC officials revealed that they received requests from chiefs and politicians to deploy a site in their rural hometown. While the details were not disclosed, my observation suggested the process depended on the bargaining power of the different sides. An Ascend manager called Marcus once said some chiefs would “give the land for free in exchange for the network.” But in Eastern Region, one chief I interviewed admitted he received “a small amount of money” from the government by agreeing to the deployment of a grid site in his village. When I visited a solar site in Volta Region, people praised the chief for bringing the site to the village. Well respected, the chief has a great influence on village affairs even if he mainly lives in Accra. Generally speaking, chiefs benefit more from RTP than ordinary residents, but what they seek varies from financial gain to cultural and political influence. It is also worth noting that three of the seven sites I visited are located near the village schools, and school land, according to rural residents I talked to, is owned by the government instead of chiefs. ¹² This is likely to be a negotiation strategy aimed at reducing GIFEC's reliance on local opinion leaders such as chiefs. At the national level, RTP is more aligned with the interest of the ruling party as its signature project. In November 2023, the ruling party was criticized for allowing RTP to “spend more than its stated budget” by a member of parliament from another party. ¹³ Political disputes like this offered a glimpse of the unequal influence over and benefit from RTP.

Sometimes the “technical” criteria GIFEC openly discussed with me can also conflict with each other. For instance, prioritizing green fields often means a lack of road or electricity infrastructure. When a solar site was deployed in Okoda (see Figure 2), a small village in an Eastern Region forest, there was no road or grid there, which also explained why it used solar panels. By the time (July 2024) I visited with local Huawei engineer Kofi, the Electricity Company of Ghana (ECG) had just extended the national grid to the village, electrifying a few households. “Here, you have telecom first and then the light,” Kofi commented. That the internet came to the village before the light testified to GIFEC's achievements in “connecting the unconnected.” Densification of the rural network, however, does not lead to intensive datafication and then decent revenue for operators. One important reason is the low population density in these regions, a long-term challenge exacerbated by Ghana's rapid urbanization in recent years. Walking around Okoda, I only found a few kids and elderly adults, as many young people had gone to bigger towns or cities for work – a common scene during my visits to Ghana's rural areas.

Unlike the 400 MTN sites, the overall revenue from GIFEC sites is growing slowly and unevenly. According to Peng, a Chinese Huawei manager, in some regions only 20%–30% of the sites’ network capacity was used in 2024. Since rural sites must hook up with macro sites for data transmission, GIFEC promises to share the revenue with operators to mobilize their participation, which entails another round of selection and negotiation. When operators receive the village list from GIFEC, they will select the most promising ones, and the revenue-sharing agreement varies from site to site depending on their negotiation with GIFEC. Relatively speaking, MTN has larger negotiating power than Telecel due to its dominance in Ghana's telecom sector, which accounted for 74% of mobile voice subscriptions and 50% of mobile data subscriptions in 2023. ¹⁴ In addition, Telecel is more burdened with political obligations to cover remote areas, as the government holds 30% of the company's stake.

Unsurprisingly, Telecel is more troubled by the revenue challenge than MTN despite its high tolerance for a low return rate. Danso, a senior Telecel manager, admitted that “the data is not picking up as expected.” From the user perspective, their embrace of datafication opportunities is very selective and cautious. An older man I met in Eastern Region upgraded his feature phone to a smartphone after the site was installed in his village. He quickly found his phone bill increased for reasons he could not understand. “MTN eats my money! I may click on the wrong button, and then, my money is gone,” he complained. Since then, he often turns off mobile data unless he must use it. When it comes to mobile voice traffic, many rural residents prefer receiving calls rather than making them themselves to save money. All these factors contribute to the meager revenue from rural sites, posing great financial challenges for GIFEC. As a GIFEC official called Richard remarked,

You see the small money gained from the rural traffic. You must use the majority of it to pay for the power [i.e., electricity]. For RuralStar to work, we need to connect to a macro tower [that operators rent from tower companies]. So, we share profit with operators. They [operators] take a certain percentage of the revenue to pay for the power. We take a certain percentage to pay the rent [for deploying our equipment in the tower]. Every month, we have to pay.

It should be noted that in 2022, after years of market reforms, the majority (79%) of telecom towers in Ghana were owned and managed by TowerCo Ghana, a joint venture between American Tower Corporation (51% share) and MTN (49% share). The unequal access to tower infrastructure between MTN and Telecel is likely to make a difference in their revenue generation from the rural sites. Through selective densification and datafication, the structural inequality generated by Ghana's neoliberal reforms is maintained, if not amplified, during the implementation of RTP.

Maintaining the infrastructural materiality: the tropical challenges

In its PR materials, Huawei repeatedly tells the story that when a pilot RuralStar site installed on a wooden pole brought the internet to a Ghanaian village, the community described it as “magic wood” and volunteered to protect the site (Huawei, 2018). The story might be true, but it significantly mystifies the technology by obscuring its everyday materiality. When I visited the wooden pilot site in Eastern Region, almost everyone complained about the network: “It's not stable. It's on and off.” After being deployed in 2017, there was little maintenance of it. Mistaking me for a technician, the chiefs even asked me to report the problem to MTN or Huawei to urge them to solve the problem. Browsing local media coverage about RTP, I also found reporting about two abandoned sites, one in Oti Region and the other in Ahafo Region. Speaking of the abandoned site in Ahafo Region, the chief put the blame on “the government's neglect,” and angry comments on the news website called for a few politicians to step down, including the head of MoC (GhanaWeb, 2024). Demystifying the “magic wood” narrative, these stories point to the everyday failure of datafication of the rural sites.

Maintaining the last mile in the tropics can be difficult. The first challenge comes down to the shifting tropical materiality of the base station infrastructure, including its solar panels, battery, and electrical meters. Unlike corporate narratives that hail solar energy as a cheaper and greener solution for off-grid areas (Günel, 2021), my interviewees often highlighted its unstable power supply shaped by the changing tropical climates. During the rainy season from May to September, frequent rain will slow the conversion of sunlight into electricity and interrupt the data transmission between rural and macro sites. This is consistent with villagers’ complaint that the internet signal often becomes weak or totally disappears when it rains. If the battery is short-circuited by a lightning strike, the whole site will be shut down. The dry season also has its problem due to the Harmattan, a dusty season appearing in West Africa between November and March. Characterized by trade wind that blows from the Sahara Desert into the Gulf of Guinea, the Harmattan brings a heavy amount of dust, which, coupled with the lowest sun of the year, reduces the efficiency of solar panels in capturing sunlight and hence the performance of the base station.

Sometimes, damage comes from humans instead of nature. In one village, four solar panels were broken when kids threw stones at them during gameplay. Occasionally, batteries are stolen due to a lack of inspection. According to David, the leader of Ascend, one of the biggest challenges in implementing the project is “quality assurance”:

With each site, you must do it up to the standard. You must constantly check and correct things that are not done properly. We ask technicians to do regular visits every three months. When the site is down, you have to go and repair it. But the problem is they are very far. And the road is bad. Sometimes you plan to visit four sites [in a day] but end up with only two… Because we have some issues with the solar sites, GIFEC is considering converting some of them into grid ones.

In practice, Ascend further outsources the maintenance work to a smaller local company called Remi. In total, Remi has 15 teams providing maintenance for the project, with each having two to three “field engineers.” I interviewed two Remi field engineers in July 2024, who joked that they were given this name because they were “always in the field” due to the nature of their job. Moving between different “fields” that are distant and dispersed entails an intensive workload. While the two engineers had some working experience, neither had received systematic training for RTP. When we visited two sites in Eastern Region together, they could not explain or solve the technical problems from time to time. As the equipment provider, Huawei provides some technical support, e.g., the local Huawei engineer, Kojo, often conducts “on-site trouble shooting” for the project, but its overall role is very limited due to a lack of manpower and the multi-stakeholder structure of the project. According to Kojo, Huawei has no more than five Ghanaian engineers working on RTP. Koji himself is tasked with multiple Huawei projects and could not spare much time for it.

In response, both GIFEC and Ascend have attempted to mobilize the participation of rural communities. As David puts it, “we are trying to involve the town folks as much as possible. Tell them that this network is saving you. So, you have to be careful about whatever is happening.” For some sites, they managed to find a “security guard” in the village. But the challenge is that “when a problem arises, people may channel their complaints to the person who is looking after the site,” says David. Tensions also exist within rural communities. In Eastern Region, a group of women I met believed that the base station would “make people sick,” referring to a rumor widely circulated during the pandemic that 5G technology could cause COVID-19 even though the site in their village only supported 2G and 3G. “We told the chief about the risk. But he still endorsed it. Then we said the government should give us insurance in case we get sick,” one woman called Gina said, half-jokingly. In another village with a grid site, a store owner living nearby complained to me about the noise generated by the base station. Things get more complicated if the grid site uses a villager's electrical meter. This happened to some communities when Ascend had to complete the site before ECG came to install a separate meter for RTP. As one GIFEC official, Samuel, recalled about this scenario,

For the grid site, the power is always on. 24 h. 365 days. [If it connects to your meter], all of a sudden, you will notice that your power is running quickly. If the [pre-paid] meter is not topped up immediately, ECG will disconnect you. You’ll be unhappy. And then, the site will also be shut down.

In this story, the inconsistent pace between the plural last miles – rural telephony and rural electrification – became a source of dispute between GIFEC and villagers. For GIFEC, the rush to install the sites is not merely motivated by political considerations. Due to project delay caused by COVID-19, GIFEC has stored the equipment exported from China in a warehouse for a while. “The batteries are going down,” Samuel says, “We have to use it very quickly,” which again points to the fragile materiality of data infrastructure in the tropics.

Discussion: the tensions of multiple data relations

Supported by Chinese capital and technology, RTP represents an ambitious infrastructure effort to bridge the digital divide in contemporary Ghana, one which transforms the rural last mile into a frontier of datafication and statecraft. This is not a smooth process but subject to failures and negotiations. To explain the intricate mechanisms of RTP, this paper illustrates three types of data relations with regard to the state, the market, and nature, which intersect and sometimes contradict with each other. I argue that the difficulty of infrastructuring the last mile in Ghana can be partially explained by the inherent tensions between the three types of data relations.

For the Ghanaian state, datafying the last mile is both a political responsibility to deliver basic telecom services to its rural citizens and a campaign strategy to attract rural votes. While the state effectively mobilizes telecom operators, it cannot fully control where and how rural sites are being commodified in the market logics of maximizing the profit. The state–market tensions create two competing forces that prioritize the politicization and commodification of data relations. These tensions are not new, but they intensified in the early 2000s with a new wave of institutional efforts to reregulate the market by remedying (if not reversing) its urban bias. In practice, transmitting data between rural sites and macro sites depends on successful mobilization of the tropical land and climate in deploying or powering the base station infrastructure. From an ecological perspective, this is a process of building a data relation with nature by assembling various elements such as dust, rain, and sunlight into the actor–network of infrastructuring, which either facilitate or resist datafication. As I demonstrate through RTP, this process is very unstable given the shifting tropical materiality and the lack of regular maintenance. Failing to stabilize this data relation frequently disrupts the politicization and commodification of the last mile.

Huawei entered the picture when it proved that datafying the last mile was both technically and economically feasible through the 400 MTN sites. Its RuralStar solution not only laid out the technical architecture of datafication but also scaled it up with the mobilization of GIFEC. Instead of being a sudden innovation, the birth of RuralStar benefited from Huawei's long-term experience of tinkering with rural networks as a marginal player in the market. In this light, global China is not about implementing a top-down globalization strategy but often times is a structural response to local needs and markets as a latecomer. Putting RTP in the broader context of Africa–China engagement, it exemplifies a cooperation model of “African state project – Chinese loan – Chinese technology,” which is increasingly used to expand Africa's digital infrastructure recently. Compared to the infrastructure-for-resources model much discussed in previous literature (Corkin, 2016; Jureńczyk, 2020), China's motivation here is not to access natural resources but more technologically and economically oriented. In the RTP case, Huawei frames itself as a neutral and rational market participant that merely sells telecom equipment. Equipment, however, is not neutral but can hold strategic importance for the global tech race, for instance, by affecting future standard-setting. While RTP mainly focuses on 2G to 4G at the current stage, the project raises an important question regarding whether it will enhance the dependence on Chinese technology when Ghana upgrades its rural network to 5G or even 6G in the future. For African countries, this is not an easy question, as it concerns both the state–market relations and broader geopolitics over next-generation internet technologies.

In Ghana, both dreams of bridging the digital divide and going beyond aid have encountered setbacks. Facing a debt crisis during the COVID-19 pandemic, Ghana once again turned to international donors to negotiate debt restructuring programs, including a US$3 billion loan secured from IMF in 2023 (Sauvage, 2023). The shift to international aid is a severe blow to the “Ghana Beyond Aid” vision and raises new questions about its future positioning in China–West competitions. In December 2024, a few months after my fieldwork, Ghana's ruling party New Patriotic Party lost the presidential election in a landslide defeat. Two of my interlocutors at GIFEC were removed from their positions after the new president took office, casting uncertainties over the future of RTP, which is already struggling with manpower and maintenance.

Conclusion

Through a case study of a large-scale rural telephony project in Ghana, this paper elucidates the intricate entanglements between state, infrastructure, and nature in connecting the last mile in the tropics. It offers several contributions.

First, it illustrates the multi-layered yet contradictory techno-politics around Sino-African infrastructure engagement. For the Ghanaian state, connecting the last mile is a means to restore its political legitimacy after decades of neoliberal reforms that marginalized its rural populations. Hence, it re-politicizes the digital divide to enlist the participation of large operators. While this is not a new phenomenon in Ghana, the support of Chinese capital and technology makes it possible to scale the project up. In this sense, the arrival of Chinese actors creates new space for Ghana to pursue its own development agendas given Western tech's obsession with urban markets and Western donors’ preferences for education and health over infrastructure. However, this space is highly structured. Seeing the last mile as a site of value extraction, commercial operators frequently negotiate the location and revenue-sharing of rural sites, which continues, if not widens, the structural inequality in Ghana's telecom industry.

To mitigate its geopolitical risks, Huawei considers RTP as an opportunity to sell equipment, and it de-politicizes the project by distancing itself from the Chinese state while reducing its local participation. Paradoxically, Huawei's local participation is not encouraged by the Ghanaian side due to the motivation to create more local jobs through construction and maintenance. Hence, contrasting with the close interactions with local stakeholders at the design stage (Lu, 2025a), Huawei is less obliged to transfer technical know-how at the maintenance stage.

Consistent with Gagliardone's (2016) research on Ethiopia, this paper reveals the ambivalent techno-politics of digital infrastructure in Ghana. This challenges the narrative of China-importing-its-development-model-to-Africa that essentializes both sides and calls for more nuanced understandings of Africa–China engagement. Historically speaking, the questions RTP raises are not entirely new. To some extent, the depoliticization narrative used by Huawei resembles the twentieth-century development discourse that viewed technologies as neutral development tools, or what Ferguson (1990) calls “the anti-politics machines”. The design–maintenance gap observed in RTP also repeats the failures of many development aid projects, which offers an important lesson to African policymakers, no matter if they are using Chinese technology or not.

Theoretically, this paper contributes to infrastructure studies and critical data studies by bringing the environmental politics into the theorization of last-mile datafication. It expands our understanding of how datafication intersects with nature by examining an understudied data infrastructure (base stations) in a challenging climate zone (the tropics). Attending to the trajectories of batteries, electrical meters, and solar panels in the tropical climate, this paper demonstrates that infrastructuring the last mile is also a process of confronting and maintaining tropical materiality, whose unstable nature constantly disrupts the datafication efforts. Therefore, I contend that infrastructure-building should not be understood as an anthropocentric story of mastering nature but as an opportunity to rethink the ontology of nature and infrastructure as they increasingly intersect with datafication. As Lehuedé (2022, 2025) reminds, the expansion of data infrastructure in the capitalist world has been guided by “an assetized ontology” that reduces nature into resources. The obsession with one-way extraction often forecloses the possibility of living with or caring for nature. Regarding RTP, the abandoned sites are likely to generate a great deal of e-waste, including lithium-ion batteries, telecom equipment, and solar panels that contain toxic metals such as lead and cadmium. Without care and repair, the recent solar boom across the world will lead to a global “solar e-waste” (Munro et al., 2023), which will challenge the green credentials associated with solar energy. Echoing recent literature on data infrastructure maintenance (Jackson et al., 2024), this study argues that maintenance is necessary both for stabilizing the infrastructural materiality and for caring for nature.

Finally, this paper sheds new light on the ontology of the last mile by revealing its multiplicity and instability – what it is varies from context to context. As I demonstrate, the last mile is not a singular line or point but an assemblage of the human, nature, and infrastructure. Physically, it often converges multiple infrastructural frontiers, such as road, electricity, and telecom networks, whose interdependency enhances existing geographical bias and whose inconsistency creates tensions between stakeholders. Discursively and politically, the last mile entails multiple forces that compete in defining its techno-politics and data relations. Shifting state–market relations, coupled with unruly natural forces, make the infrastructure highly contested and unstable. This instability also means there is no single best way to develop the last mile, as is exemplified by the agency model, the mass mobilization model, and more recently, RTP. How it unfolds depends on how different actors are assembled and mobilized. Seen in this light, a sustainable approach to addressing the digital divide requires a delicate balance between political, commercial, and ecological forces, which offers important implications for future last-mile projects in Africa and beyond.