Introduction

A vendor in Lagos who once sold only to those who walked past her stall now ships to towns she has never visited. A driver in Jakarta who had no reliable source of income last year now earns a living through a platform. Both are, according to a growing body of scholarship and an increasing number of governments, victims of extraction.

Extraction has become the main topic of contention in the global discussion about Big Tech in developing countries. Critics argue that platforms arrive and take money out of the country, data to the Global North, and market share from local entrepreneurs. The paradigm has moral force and has created its own policy agenda to match it: digital service taxes, data localization mandates, and antitrust lawsuits across four continents.

The diagnosis needs investigation because the policy remedy depends on getting the diagnosis right. This series began with Amazon in the United States, where, although the extraction premise should have been strongest, it was found not to hold. Amazon accumulated wealth through value creation, not by extracting value from consumers or employees; the losers in the process were Amazon’s competitors. Extraction and competitive exclusion are different things and require different approaches.

The straightforward objection is that the framework breaks down elsewhere. As good as it works in Seattle, the argument goes, the calculation is different in Lagos and Jakarta, where there is a stronger power imbalance and states with weaker regulatory capacity.

The objection does not hold when confronted with evidence from these platforms. Jumia, Grab, and MercadoLibre, chosen because they represent leading digital platforms across Africa, Southeast Asia, and Latin America, demonstrate how platforms can create genuine surplus for consumers and small merchants rather than simply extracting value from developing countries. But the extraction narrative obscures two harms that are real, structural, and urgent: data sovereignty and gig worker precarity.

The Surplus is Real

The numbers are strikingly clear. The MercadoLibre ecosystem serves over 9.5 million entrepreneurs and small businesses in Latin America. In Argentina, such businesses constitute 9.8 percent of GDP, and for over 50 percent of those businesses, sales through the platform are the primary source of their income. This is not a platform capturing a fixed pool of commerce; it is commerce that previously took place through informal channels, reached only local customers, or never occurred because buyers and sellers could not efficiently find one another. 

The situation with Jumia is even clearer. In 2021, 27 percent of the delivery services provided by Jumia reached remote regions where, according to the company itself, the consumers’ access to the products was severely restricted. These regions are accessed through partnerships with over 600 logistics companies, ranging from entrepreneurs to well-established companies. Thus, a person in a secondary Nigerian city who previously had no practical way to purchase a particular product can now do so.  platform reduces the cost of matching buyers and sellers, it expands the number of available suppliers, and even makes delivery economically feasible, often offering lower prices than local alternatives. The difference between “unavailable” and “available” cannot be considered as a simple exchange from anybody else’s account: it represents newly created surplus, a portion of which goes to the consumer through greater choice, lower prices, and increased access.

Additionally, according to a Grab report, roughly 45 percent of the company’s partners had no income before becoming partners of Grab. The information is provided by Grab’s own reporting and should be weighed accordingly; however, even discounted, it points to something the extraction narrative cannot accommodate: income cannot be extracted from someone who had none.

Critics are right that someone loses, but they are wrong about whom. When Jumia reaches a town previously unserved by retailers, the losing party is not the consumer, who is clearly better off; rather, it is the informal middleman who could previously charge a premium for scarcity. That is competitive displacement, and it is what functioning markets do.

Data Sovereignty is a Real Harm, and it is Not Extraction

Here is where the critics have an argument to make, and it deserves to be made carefully rather than folded into a general indictment.

As Nick Couldry and Ulises Mejias explain in their piece, “making data colonialism liveable: how might data’s social order be regulated?” user data collected in the Global South is analyzed, monetized, and stored on the servers owned by the companies of the Global North, all while there is very little benefit for the country where such data is collected. The asymmetry is structural and does not require intentional misconduct by platforms to produce unequal outcomes. It also does not imply that the platform in question is foreign. Gojek and Grab both originated in Southeast Asia, yet both operate within a global capital framework supported by foreign investors, and data extracted from Indonesian or Vietnamese users is often stored and monetized beyond the jurisdiction of national regulation. Local ownership does not guarantee local data sovereignty, nor does it ensure that the citizens of the Global South benefit from the value created by the data they generate.

Infrastructure economics make this point concrete: constructing a data center may create over 1,500 jobs, but operating and maintaining one generates very few lasting jobs, around 100, meaning the value captured in-country is minimal even when the physical hardware sits on national soil. Additionally, the data itself is a valuable asset whose economic returns are typically captured by platform owners rather than the citizens who generate it.

This is not that consumers are being overcharged, nor are workers being underpaid. This is a completely different issue, with its own way of functioning, and needs to be considered separately. Otherwise, the argument gets weakened by folding it into a general extraction narrative, joining the real argument to a fallacious one.

So is Labor Precocity, and it too is Structurally Distinct

According to the International Labor Organization, in 2023 more than 28 percent of the youth population in Southeast Asia and the Pacific were in temporary employment lacking financial security and social protections, a trend rising since the early 2000s. Around 40 million gig economy workers across the region remain dangerously exposed to automation without adequate protections.

The extraction narrative reads these numbers as evidence of platform exploitation. But these platforms function as risky substitutes for traditional financial institutions. In 2024, Grab’s loan disbursals rose 46 percent year-on-year to $2.2 billion, and one in three active driver-partners took loans; these workers are those whom banks turn away, by Grab’s own account, for lacking fixed incomes or credit histories.

A platform extending credit where banks will not be extracting money from workers. Rather, it is stepping into a place the welfare state left empty while operating on incentives built on flexible labor. 

The problem lies in this gap. Grab did not create the absence of social protection for informal workers in Southeast Asia. However, the platform model can reinforce that absence by allowing companies to rely on flexible, independent labor without assuming the responsibilities traditionally associated with employment, such as stable wages, benefits, and long-term protections. It simply took advantage of this existing problem, the absence, and built a business on top of it. The precarity is real and demands attention, but the policy that addresses it is worker classification reform and social protection extension, not a tax on platform revenue. 

The Counter-Argument

Critics will say this analysis is too generous toward digital platforms. Platforms displace local sellers, extract money for their operations abroad, and concentrate market power in structures that mimic colonial history. The scholars who take this approach view the expansion of American tech firms as accompanied by continuous extraction of local user data for corporate profit generation that ensures economic and cultural dependency.

When Temu entered Africa in 2024, it almost immediately took the lead position as the most downloaded app in Nigeria due to its extremely low prices. Temu is not Jumia, as it does not rely on local vendors listing products on the platform. competes against them directly, thanks to capital investments Jumia cannot match. That is a competitive threat of a different order, and the concern is legitimate. However, the criticism falls flat because it conflates three different dynamics: consumer harm, which evidence does not support; competitive displacement, which is real but is what markets do; and data and labor extraction, which is real and which the narrative obscures by lumping it in with the others.

Three different problems require three different solutions, and different platforms contribute to different combinations of these problems. A marketplace platform that enables local vendors, a foreign retailer that directly competes with them, and a ride-hailing platform that relies on gig labor do not create the same harms or require the same policy response. Treating them as a single category obscures the specific mechanisms through which harm occurs. Here, we have just one solution for the problems, and it is not the right one.

The Diagnosis Determines The Cure

The consequence of getting the distinction wrong is expensive.

If platforms extract from consumers, the response is regulation that prevents excessive pricing or unfair charges. If the harms are competitive displacement, the focus should be on maintaining competitive markets rather than treating every instance of market loss as extraction. If the harms are data flows and labor classification, the solutions are data sovereignty frameworks and expanded gig worker protections. These are different policies targeting different threats and yielding different consequences for their subjects.

MercadoLibre’s Mercado Pago illustrates one way a platform can create genuine surplus in a market where large shares of the population remain outside traditional credit systems. In Argentina, where over 60 percent of adults do not have credit cards, Mercado Pago functions as a form of financial infrastructure where conventional institutions have long failed to reach enough people. That is a clear example of surplus creation: financial infrastructure where there was previously no access. The sustainability of the surplus is determined by one thing only: whether the data that is produced by those millions of transactions will be managed in the interest of its producers. That is not solved by taxing the company’s profits; that problem can be solved through data governance, which is a different policy instrument.

The Lagos vendor and the Jakarta driver have it better than before. That is not a praise of the platforms but rather an empirically verifiable statement. Both are vulnerable: the first to offshore monetization of every transaction they make, and the second to automation without a safety net.

These are the harms: specific, addressable, and totally different from those described by the extraction narrative.

The narrative is comforting. It is also inaccurate, and inaccuracy at this level is not just a rhetorical error but a policy failure that costs people their protections. The following piece in this series, in short, describes how the misdiagnosis of the problem costs governments in taxes on their citizens and harmful localizations of the platforms.

Leave a Reply

Trending