The introduction of Digital Tribulations, a series of intellectual interviews on the developments of digital sovereignty in Latin America, can be read here and the entire series can be found here.
I met Juan Martín Graña, a CONICET researcher, one morning in early April in Buenos Aires, at a café called Aromi, near Parque Centenario, in the Caballito neighborhood. We were put in touch by Cecilia Rikap, who recently published the book Teoría de la Dependencia Digital, so far only available in Spanish, and whose book presentation, held at the public School of Economics of the UBA, I attended in a packed room.
Juan and I sat down at a large round table, ordered the inevitable coffee and medialunas, and began talking for about an hour about the political economy of digital dependency in Latin America. Over the course of the long conversation, Juan Martín laid out a framework that challenges mainstream narratives about automation and job loss in the Global South, arguing that the real risk is not mass unemployment but a deepening technological gap that translates into low wages and precarious work.
We discussed how this plays out concretely in the region’s software industry, squeezed between cheaper labor markets and the rise of generative AI, and in local platforms like Mercado Libre, which, despite reaching the technological frontier, remain subordinate to big tech infrastructure such as Amazon Web Services. From there the conversation moved to the political possibilities opening up in the region, from Brazil’s Pix as a model of public digital infrastructure to Argentina’s ARSAT, and to the broader question of how to build a genuinely Latin American path toward digital sovereignty.
Close to the entrance of the bar I found a sentence written on the wall reading Milei nos odia – Milei hates us.
***
Fig 1. At the University of Buenos Aires
What do you do, and what is the subject of your research?
I’m an economist by training. I did my PhD in Economics here in Argentina, at the University of Buenos Aires, and I’m a researcher at CONICET. My workplace is the University of San Martín, at the Center for Studies on Economic Development. Since my doctoral work, I’ve almost always worked on issues related to economic development, with a strong emphasis on the labor market. The general idea behind my research is to connect the production and accumulation conditions of firms and countries with how that translates into employment conditions, wages, and job quality.
In recent years, together with Cecilia Rikap and other colleagues, I started studying how new technologies are affecting that relationship, along two main lines. One has to do with the digital economy in a stricter sense, with the relationship between firms and how that translates into technological dependence. We did a couple of studies on Siemens and its technological dependence on big tech worldwide, and one on Mercado Libre, trying to discuss to what extent the digital economy is a window to development for Latin America or not.
The other line has more to do with the labor market: how these new technologies applied to production are changing the profile and quality of employment, especially in Latin America, because we’re at a fairly weak stage in terms of our own theoretical production. There’s a lot of simply replicating theoretical frameworks that come from the Global North, as if labor markets worked the same way everywhere.
This idea of technological unemployment, the fear that machines will replace workers, doesn’t hold in the same way in Latin America, because the conditions under which labor is hired and the conditions under which firms produce are radically different. From there comes the need to, together with Latin American theory, produce a more applied translation to the territory, one that’s useful for thinking about public policy and union strategies in the face of technological change, which is a reality.
What are the specific conditions in Latin America compared to other places?
The mainstream framework for thinking about the impact of technological change on employment consists of studies claiming that a certain percentage of jobs have automatable tasks and that those jobs will eventually disappear. It’s the idea of polarization, of the hollowing out of the middle class.
That this is based on the idea that, in the economic system of developed countries, when a technology becomes available it’s automatically adopted by firms and automatically reflected in the market. When you look at Latin America, there are two major factors that differentiate how labor markets work in the Global South, and in Latin America in particular, compared to the developed world.
The first has to do with the cost and skill level of the labor force, which in many cases is so low that the cost of machinery or more sophisticated technology ends up being comparatively so high that there’s no economic incentive to adopt it. The idea of replacement that Marx discussed doesn’t operate directly in Argentina or Latin America, because it can genuinely be much cheaper to produce with older, labor-intensive technology than to make the technological leap. That’s why we need to look at the whole world of the informal sector, something we’ve been discussing in Latin America for a very long time.
The second point, which generally goes unnoticed, is that when you do fieldwork with companies, you find that a large share of them don’t even have the capacity to adopt these technologies. Small firms often look for contractors to develop robotics applications or similar tools, and the same technology consultants ask them whether they have any qualified technician who could fix the machines if they break down, or whether they even know what these technologies are for. The answer is usually no. So, beyond what studies say about the risk new technologies pose to the labor market, those effects need to be moderated by how the labor market actually functions and by firms’ real capacity to adopt technology.
Last year we did a study where we built an index to weigh that potential risk, mediated by wage differences between Latin America and the USA or Europe, and by existing technological capabilities. That index showed that, in many of these sectors, even the most advanced countries in the region have wages that are half those of the USA, and that internet connectivity is also about half. Several factors point to the fact that the problem for the labor market in Latin America today isn’t technological unemployment, because not every firm is going to buy the technology and replace workers, but rather a deepening technological gap. USA, Europe, and China are applying these technologies, and the region is going to fall further behind. That’s going to affect the labor market with more poverty and more precarious work, because firms, in order to stay competitive, will have to cut wages.
I think the logic that fits Latin America when it comes to new technologies has more to do with the old discussion about a development project, a project of technological planning and research and development, than with the warning that we’re going to lose our jobs. The risk is more one of deepening precarity than of outright unemployment.
And this threat of robots coming for our jobs usually functions as a form of discipline. But I wanted to ask about the latest wave of generative AI applied to programming; it seems to me it has quite an impact on IT workers.
Regarding the specific labor market for programmers, there are a couple of particular issues. Argentina and Brazil developed a software export industry based on knowledge services, heavily concentrated in the intermediate links of the value chain: neither small shops doing simple, low-wage work, nor large, complex systems. The competitive advantage of that sector in Argentina has essentially to do with lower wages: we do the same of a software engineer in the United States, butat half the salary. Argentina also has a competitive edge because of good English proficiency and similar time zones, which favored outsourcing many of these services.
The problem is that, being in that intermediate link, there’s competition from below, from countries with even lower wages, such as India, Kenya, or Southeast Asia, where there’s a very low wage ceiling and no room to progress. That also prevents moving up toward more complex links in the chain. We’re caught in the middle of a sandwich where generative AI hits directly, precisely because this segment did neither the most complex nor the simplest work.
This is just starting, but it goes quite fast. What’s being observed is that the strongest impact is on entry-level programming positions, because AI can write code, but you always need a senior programmer to review and check it. There are several somewhat funny stories of companies that gave AI access to the base code of their platforms and ended up crashing the whole system. From that came the idea that you shouldn’t let AI control everything, because it makes mistakes just like anyone else, and that some human oversight should be kept.
So the main problem arises for all the young people who, over the last ten years in Argentina and Latin America, thought that learning to program guaranteed them a job. That’s exactly where the impact is strongest.
On top of that, there are two problems specific to Argentina. The first is that Globant, one of the major multi-Latin software companies, is going through a very complicated economic period. Globant is an Argentine software company with offices in many countries, which ten years ago went through a process of acquiring numerous other companies, worked as a contractor for Disney and other large multinationals, and became known for a very strict, aggressive labor regime with low wages, although it promised a certain career path given that it was a large company.
Over the last two or three years it has had very poor results: it’s listed on the New York Stock Exchange and its stock has plummeted. That has also spilled over into the Argentine system, since many programmers who had moved to these large companies ended up in migration-type positions, working on specific contracts for the United States or Europe, which makes them even more fragile in the face of automation or AI: if you work alone, within a team you don’t control, and based in another country, you’re the first exposed to automation.
So several dimensions come together around software work: some structural to Latin America, like low wages and a position in an intermediate link of the chain; some specific, like the large number of young people who went into programming, which is where AI is hitting hardest; and a particularly bad moment for major Argentine software companies, whether Globant or Mercado Libre.
In Argentina we have a serious problem of household debt and, above all, late loan payments, which has hit Mercado Libre hard, since Mercado Pago is one of the leading fintechs in Latin America. The whole sector is being hit both by technology and by specific economic issues. It’s a complicated moment, and we’ll have to see over the next six months to a year how much of this promise that AI will replace tasks is actually put into practice.
We’ll especially need to watch the segment of companies we call “software factories,” which produced simple, somewhat customized programs for medium and small businesses. I think automation there is going to be brutal, especially since they compete on cost: once the base code is created, it just gets replicated. That would be a serious problem for Argentina, because it was one of the few segments of the labor market that was working relatively well, generating employment despite the crisis, and offering relatively good wages compared to the average. It’s bad news.
Speaking of these Argentine companies like Globant or Mercado Libre: Cecilia’s shows that they end up working for U.S. tech companies, both because of the appropriation of intellectual property and because they pay to use their computing infrastructure, Do you agree with that?
That idea is part of the work we started discussing with Cecilia and another colleague, Sebastián Fernández Franco, on Mercado Libre. At first, it was an update of dependency theory, which posits two very sharply defined poles: multinationals and subordinate domestic entrepreneurs.
We did a series of interviews at Mercado Libre with programmers and staff from different areas, such as customer surveys and user experience, and a consistent answer came up: in technological terms, anyone who passes through there and ends up working can go work anywhere, because there’s no technological gap at all, they’re at the frontier.
So we asked ourselves why Mercado Libre doesn’t consider some business scheme to become independent, especially thinking about defending its own corporate interests. Mercado Libre uses Amazon Web Services to run its algorithms and manage its database. In Mexico and Brazil, both Amazon and Mercado Libre compete: who are you giving your data to? You might protect it, but you’re handing it to your competitor. Netflix, for example, states in its annual reports to the New York Stock Exchange that one of its main risks is that its systems run on Amazon Web Services, a company that has its own streaming service.
There’s a clear risk there. Mercado Libre’s response is that it’s not interested in getting into the data business or the server business, because it’s a business of a scale and expense that’s too large. From there comes the idea that there aren’t just two poles, but an intermediate layer that takes big tech’s technologies, adapts them, and thereby achieves a competitive advantage in a particular market.
It’s not just Mercado Libre in Latin America, which started with an eBay-like model for selling used goods, then created Mercado Pago with a security argument, releasing payment only once the buyer confirms receipt and that the product works, and later shifted toward an Amazon-like model selling new products from small producers. In India there’s Shopee, in China there are equivalent versions: in every region of the Global South, platforms emerge that, with local adaptations, achieve a competitive advantage, but at no point do they try to replace the big tech firms — instead they accept a subordinate position. Any loss of value they suffer from using that infrastructure gets passed downward.
That’s how Mercado Libre ends up with a serious problem around the volume of commissions it charges sellers, in logistics and advertising, replicating Amazon’s model. What we argue is that the main culprit in this dynamic is clearly Amazon, but that it would be naive to think there’s a path to development through these intermediate companies, because they aren’t independent agents with respect to that power, but rather the lever through which Amazon Web Services operates over Argentina.
In our fieldwork on technology adoption among small firms, we also found a local business model: small companies selling algorithms and cloud services with their own very small servers for equally small clients, because contracting AWS was too costly. Mercado Libre is something similar, on a larger scale: it adapted Amazon’s model to Latin America, has a first-mover advantage and controls the market, but is in an accepted position of subordination, with no clear way out of it.
What can be done to avoid the reproduction of these extractive models and reduce dependency form the global actors? I am thinking, for instance, about Pix, the Brazilian public payment platform.
When you reach the conclusion that these companies aren’t the path to development because they’re part of this global infrastructure, two clear positions tend to emerge from progressivism. One says that they’re simply part of the problem. And they are, but that doesn’t mean it makes no difference whether they exist or not. It wouldn’t be the same for Latin America if Mercado Libre existed or if everything were directly Amazon, because that implies there are certain capacities that, put at the service of a different political project, are already in place. You don’t have to go negotiate directly with Amazon in the United States.
What’s missing is that planning logic that defines what particular role we’re going to give these companies. They’re not going to save us, of course not, but having them based in Argentina, with some local connection, is an advantage. Several Latin American countries have public companies linked to this field. Argentina has ARSAT, which operates satellites, has a long history of producing ground data for agricultural applications, and has its own servers, in addition to having laid thousands of kilometers of fiber optic cable across the country’s interior.
We have those capacities. What’s missing is understanding the global picture, and there Cecilia’s book is very useful because there isn’t much material explaining how everything ends up filtering up to these big companies. What needs to be thought through is how to gradually cut those channels, without being able to demand that a single company face off against Amazon on its own and go under. A radical proposal needs to be built, but step by step. Nor can we demand that the public stop using international platforms, because they’re part of daily life. We need to find ways to progressively cut that dependence, developing alternatives.
In that sense, the Pix case in Brazil is exactly what needs to be pursued. There’s a clear point of potential outreach: few people are deeply familiar with Pix beyond having used it on vacation and noticing it works well, but the story behind it, a public development that denied space to international companies, is genuinely interesting. In Argentina we have Cuenta DNI, run by Banco Provincia. I remember that a couple of years ago, in a course we designed with Cecilia and that she still teaches at San Martín, Digital Capitalism and Underdevelopment, one of the students worked at Cuenta DNI, and we often ended up discussing what they did with the data. The answer was that they had huge amounts of data on businesses and people, but they just sat there.
It’s not enough to say data is important: it has to actually be put to use, and you need to understand for what. Starting from public companies like these, you can begin building public infrastructure, cables, data centers, and open up a space for accumulation that doesn’t depend on the big tech firms.
The big problem, visible in the contradictions of digital development policy in Brazil or Chile, is that national states are crossed by competing interests, and these companies, not just Amazon but also their local partners, exert a lot of lobbying pressure. Progressivism is missing the ability to “sell” a general project that mobilizes people. If you ask the general public, even progressive legislators, whether the digital economy matters, many just think of their phone and not much else. In reality, behind it there’s an entire economic model and a whole possible development scheme, and that’s still fairly far from common awareness. I’ve heard well-intentioned people say that if the government wanted to, it should “create an Argentine Pix”, when in fact the Brazilian Pix itself could simply be used, as is already happening in Colombia.
Pix is the first piece of a much bigger puzzle: where that data is stored, how it travels from phones or banks toward public infrastructure, toward national companies’ satellites. From there you move toward the idea of the digital stack, and I think different Latin American countries already have capacities for this, with public companies and large state banks that have invested heavily in various areas. What needs to start happening is, for instance, thinking about how Latin America could combine the Brazilian Pix with Argentine satellites. That way the goal stops being so unreachable, so utopian, and becomes something you can progressively work toward.
I’m very interested in the question of a Latin American path. It seems to me there’s more room for experimentation than in Europe, where we talk about regulating but keep using only USA software. When I talked to people from different Latin American countries, they often tell me more collaboration is needed, but that the political conditions for it don’t exist. What can be done in the coming years to move forward?
The first task is getting the far right out of power. In Argentina we’re at a very telling moment: Javier Milei’s discourse, and that of his delegate for artificial intelligence, was simply to install Amazon data centers in Patagonia because it’s cold there. That’s the end of the discussion. Before anything else, certain political consensus needs to be built, not only within progressivism but more broadly.
At least think of it instrumentally: there are ways to collect more taxes if companies are local, to recover sovereignty, to recover privacy for users regarding what data is collected, how it’s used, and where it ends up. It’s a less revolutionary, less controversial agenda, similar to the path that, with its own problems, Europe followed with data protection and the GDPR.
You raised an interesting point: that there seems to be more room to do these things in Latin America than in Europe. That’s where it helps that the region has intermediate actors that Europe doesn’t have. In Europe, the fight is directly against Amazon, with no local player you can tell, through proper industrial policy, to grow at that competitor’s expense. Here we have that space.
We also need to develop the debate over where to locate data centers and what role ARSAT can play in connectivity across such a vast territory as Argentina’s, something applicable to the rest of Latin America: Amazon isn’t going to lay a cable to a city of a hundred thousand people in the middle of nowhere. There are many points of contact between today’s very modern, timely disputes over the digital economy and debates that are already fairly well established in the region about certain things the state has to do.
The problem is that doing that isn’t enough: it needs to scale up. Along with some colleagues, we’re trying to put together a series of talks to spread these issues. I think there needs to be a first stage where this doesn’t stay confined to specialist groups with political activism, but actually starts entering the public agenda. Argentina is in particularly bad economic shape for people to sit down and discuss this in depth, but we do have regional organizations, like Mercosur, that could serve as a space for it. The problem is that over the last twelve years, when Argentina had a relatively progressive government, Brazil didn’t, which complicates coordination. Progressivism in Mexico is relatively recent, and Mexico is too closely tied to the United States, which makes it hard to bring into this discussion, though if it were more aligned it would be a very significant player. Colombia, meanwhile, is joining in.
I don’t see it as such a hopeless space, but there’s a lot of work to do to get this onto the public agenda. If you look at the Argentine Congress, the legislative bills on artificial intelligence, even if they won’t pass, show the state of the discussion: some are very much on the right, along the lines of “don’t regulate anything,” close to Amazon’s or Peter Thiel’s discourse, while others at least attempt to advance privacy protections. We need to identify which of these issues best opens the door to public opinion.
When Australia regulated social media for minors, many said it was impossible, that it would never happen. Now all of Europe is taking steps in that direction and it’s gradually being accepted. At least let’s start there, and from there move toward general data privacy. There’s a lot of work to do before, or as a path toward, building that regional alliance capable of pushing change, because the interests on the other side are very strong and there’s a lot of money involved.
One area where we’re going to face immediate problems in the political discussion is data centers. There’s an established narrative that presents them as the “railroad of the twenty-first century,” according to which whoever doesn’t have data centers is left out of progress. But having data centers doesn’t mean being part of progress, because progress lies with whoever controls the algorithms, and that’s generally done remotely. People imagine a data center as an office full of people working, when in reality it’s a building that could have its lights off, with nothing but processors running. That’s where I see an urgent need for the region: to spread awareness, to explain these mechanisms, because you don’t need to be a programmer to understand the economic scheme behind them.
Moreover, within the context of a very fierce geopolitical struggle between the United States and China. Even among people who are somewhat aware of these debates, there’s sometimes a naive view of China’s role, as if the problem were only U.S. big tech and one could do business with Chinese big tech instead, following a non-aligned country logic. That shows a failure to grasp that the problem is structurally the same. China managed to develop its own tech ecosystem domestically, and it’s only now starting to expand abroad, but that’s not a different way out for us: we need to look for a genuinely different alternative.
If at some point Europe manages to steer its own ideas, that could serve as a reference, even if we say it somewhat mockingly, the “Brussels moment,” when Europe, after years of discussion, sits down and decides to move forward with regulation. For countries further behind on regulation, even if it doesn’t solve everything, it helps show that if it was done there and nothing terrible happened, social media didn’t disappear and people didn’t lose the ability to order food delivery.
For instance, we haven’t mentioned it yet, but there’s the case of Spain’s “rider law” for platform workers. In 2021 Spain passed a law recognizing platform workers as employees. Here, Milei’s latest labor reform, passed two months ago and currently suspended by the courts, goes in the opposite direction: it defines them as independent contractors, in a commercial relationship. When you point out that they’re actually workers, the usual response is that this would make deliveries disappear. Just look at Spain to see that isn’t the case.
Some basic awareness that these discussions are already happening elsewhere helps people understand that it’s not some outlandish idea to advocate for protecting people’s data, especially minors’, or to raise the tax issue: how these companies book their profits elsewhere to avoid paying taxes, when their entire wealth is based on data and yet they pay almost nothing. There are many possible entry points, and a discourse needs to be built that brings them all together, so that each country can then decide which of these issues resonates most with its public.
In Brazil, what happened with Twitter, with Elon Musk, and the clash with Trump, brought the issue to the forefront. Earlier, the wiretapping of Dilma Rousseff’s phone had already made data privacy a priority. We need to see what that triggering mechanism could be here. If we don’t manage to get this into public conversation, that’s going to remain the main problem, because after that, the public policy discussion, how far we can actually advance concrete initiatives, becomes easier. In Argentina today there’s nothing: no AI plan, no public infrastructure planning, no process of reflection on this. There’s even a push to privatize public companies. We’re so far behind that any progress would be radically different from what we have now. These two years are going to be mostly about resistance.
Speaking of institutions and wages: I’m a fan of a French philosopher, Bernard Stiegler, a Marxist who died a few years ago and who began studying platforms through the philosophy of technology, and he concludes that the wage as an institution no longer works, and he talked about finding other forms beyond wages, such as universal basic income. What do you think of that?
I can’t speak to his specific proposal, but it seems to me we’re at a moment where, after two centuries of capitalism, we had managed to move from piecework forms of payment toward a wage determined by the hour, so that the economic risks of the firm stayed with the firm and not with the worker. Today we’re in a moment of reflux, where firms are once again imposing piecework forms. In the case of platforms this is very clear: waiting time isn’t paid, and neither is connection time.
I think the first thing is to fight that battle on those very terms. I don’t think any of the new technologies necessarily comes bundled with piecework pay or the absence of a wage. I think it’s simply an abuse by firms, which take advantage of a labor market that doesn’t work well and an economy in crisis, and push forward given the weakness of labor unions. I’d still put at the center the need for a wage that pays for the worker’s available time, and for all the risks, health, accidents, economic, to fall back on the employer.
Now, looking ahead, if someone argues that the employment relationship itself is at risk because a large share of the human workforce won’t have jobs and another form of social reproduction needs to be found, that’s a different debate, distinct from the one raised today by the rise of platforms. I’d frame it more as a discussion about how to regulate capitalism going forward. There I have my doubts, in line with the research we do on the replacement of labor in Latin America. It doesn’t seem that capitalism is moving so much toward building an industrial reserve army or a fully redundant population, but rather that it operates on both sides of the counter: it applies automation where it’s convenient for productivity, efficiency, or control, and hires under precarious conditions where that’s more profitable.
So the dispute is going to keep existing. We’re not heading toward a world without jobs, but toward a world with jobs under terrible conditions. So rather than just thinking about how to give people income under those conditions, which I don’t deny is necessary, I think the fight has to be over which technology is applied, how it’s applied, and what role we’re going to give companies in deciding where they place people and where they don’t. Recovering that control. There the role of the union and the shop steward is central, because you need to be present at the workplace itself. The state can do a lot, but without control over production at the point of production, everything becomes much harder.
We’re back to issues that were already being discussed in the 1970s. If schemes of German-style co-determination, or Nordic-style collective bargaining with near-universal coverage, can be pushed forward, there’s much more to gain over a twenty year transition, by being able to say that a given production unit doesn’t need automation, that there’s no real efficiency advantage, and that if it’s automated it’s only to control workers, so it shouldn’t be done.
A serious problem we’re going to face soon is the one we mentioned about programmers and entry-level jobs. Technology is wiping out those positions, and in ten or twenty years, once those who learned with the previous technology and now oversee AI retire, who’s going to know how to use it? If you eliminate all the entry-level jobs today, in twenty years not even the company itself will have real control over production.
We’re at a moment of a lot of experimentation, where firms, in the middle of a global crisis with no demand growth, push forward as much as they can. Then we see the results: it happened with the thousands of layoffs tied to AI over the last six months, and then it turned out that many of those workers were rehired because, without them, the company couldn’t sustain its business development. I wouldn’t trust very short-term trends as projections of what’s going to happen.
I think that in this moment, clearly one of ebbing labor struggle, what needs to be done is strengthen organizations. My optimism rests on concrete cases, for example in the United States, where there’s a revitalization of union struggle in a country with anti-labor legislation dating back to Reagan, with “right to work” laws and similar measures. There you see a model centered on the shop steward, unit by unit, like the small Starbucks locations. I think that’s where we need to start.
The neoliberal globalization scheme, in which geopolitical risk and the fragility of supply chains didn’t matter and production was sent wherever was cheapest, has broken down. So the threat that used to be relocation, companies are now trying to turn into the threat of automation, as a means of control. But there’s also an opportunity there, because it’s no longer a dispute with a Vietnamese worker thousands of kilometers away whom you don’t know, but a direct negotiation with the local company, where the option of simply “leaving” no longer works the same way.
We need to return to that logic. The state can do a lot, I’m not sure through basic income, but by finding ways to sustain a welfare state that gives workers the security to take up that fight, funded by taxing big tech and machines. Even though the technology is new, I don’t think we need to invent great new things: we need to revalue political and power-building strategies that already worked. The geopolitical context of the coming years resembles earlier eras more than it does the 1990s. That extreme free-market neoliberal globalization scheme doesn’t seem to be working anymore, not for Europe, not for the United States, and that opens up space to redistribute a bit more of the gains companies made, through higher taxes and higher wages. It won’t be easy, but I think that’s the direction.
The end of the end of history. Now there’s a lot of talk about a new multipolar order, driven by China’s development, a mixed system, capitalist but with a lot of state control, which seems to work better than other models and could become a reference point for countries in the Global South.
Yesm and you can see that in the narratives. Over the last four or five years, the International Monetary Fund, which used to be completely opposed to industrial policy and state control over production, has published several papers explaining the success of Southeast Asia and China through the active role of the state. And then there’s Trump imposing tariffs. If a company’s entire business model depended on the existence of free markets, and trade wars keep breaking out, the space to discuss which technologies and which sectors to protect opens up again. Rather than thinking about a “new capitalism” centered on basic income, I think there’s still quite a bit to recover by recognizing that this remains a conflict between capital and labor. It’s an enormous undertaking, but I think it resembles that more than anything else.
The problem with the capital-labor conflict is that, with the digital platform, the figure of the employer shifts and hides, and that makes it hard to clearly identify the adversary. It seems that labor fight against labor.
There we need to update our concepts and look for new ways of analyzing this, but if you strip away the technological veil from the digital infrastructure, it’s not actually that hard to understand. When the debate comes up over whether delivery riders are workers or not, companies often argue that the worker brings their own means of production, the bike or the car. But the digital infrastructure that makes that work possible is provided by the company. What’s the difference with going to work in a factory? None. On that point, things are fairly clear.
The same happens in marketplaces, in the dispute between small merchants who get charged very high commissions by Mercado Libre or Amazon. What’s the difference with a supermarket? Large chains, like Walmart, also charge very high fees to place products on the main shelf: it’s the same logic of one company over another. As Cecilia says, a good part of that dispute could be addressed with antitrust arguments. Companies have been smart about keeping prices low to avoid being targeted from that angle, but essentially that’s what it comes down to.
But Antitrust policy in the United States has been corrupted by that consumer-welfare logic. When I was in New York, I noticed the level of debate is overly technical and generally and always about the same topics, because if the company benefits the consumer, everything seems fine, even though prices no longer say much, since the platform loses money in one area to make it up in another. So, I’m more interested in the relationship between platforms and infrastructure, because it forces us to think of them as public utilities.
When people say Amazon is the biggest company in the world, they think of the marketplace, not Amazon Web Services. But Amazon is investing in cables, data centers, and energy, building the infrastructure that states used to build, in order to become independent of them. When you frame it that way, and add that it charges very high commissions or is squeezing out other players, it becomes clear that it’s a monopoly.
Does it make sense to open a competitor alongside it? In Argentina we had the experience of Correo Compras, a public marketplace that never really worked, because neither companies nor consumers were required to use it: the necessary network economy just wasn’t there. Setting up a second Amazon or a second Mercado Libre doesn’t work like that, on its own. To get to nationalization or regulation, you first have to make clear that this is infrastructure involving billions of dollars, and that the state needs to build its own cables and its own data centers, and offer them to the public.
The problem is financing, but that’s where the sovereignty argument comes in: if a country enters into conflict with any power, it could lose internet service. In Russia’s invasion of Ukraine, or in tensions with Iran, it became clear that if you clash with the United States, you could have your GPS shut off or lose access to the SWIFT system. Everything points toward these issues, which used to be hidden, becoming visible. Internet cables are a public service, and they can’t depend on a handful of private companies: whether you’re a liberal or not, that’s a risk. Antel, in Uruguay, has a stake in submarine cables; at the very least we could move toward public submarine cables connecting Mercosur. That’s something that can be done.
But that connects to the problem is investment, because Big Tech has accumulated so much capital that they can lose money for ten years to win a market, while states are always in crisis.
That’s why one of the most important issues is that the digital economy seems like something so complex and novel that it requires inventing new tools, and that if you’re not a software engineer you can’t understand it. But once you break it down, the old discussions come back. Putting dependency theory back at the center makes it easier to connect these issues. In the past the problem was the U.S. banana company that set up its enclave, where the country’s rules didn’t apply, produced, took everything, and left very little behind. What’s the difference with an Amazon data center? None, really: it uses the country’s electricity and water, but everything else, the data, stays outside, and it’s much harder to recover afterward.
We need to find ways to connect today’s problem with elements closer to the historical public discourse we’ve had in Latin America. That’s why I have reservations about terms like “technofeudalism,” which suggest a radically new moment. I think the conceptual tools we already have still work in similar ways, and that the analogies of monopoly, abuse, and dependency remain more accurate. What needs to be built is an answer to the question of what the utopia of the digital world looks like for progressivism. Once that’s clearer, it will be easier to connect public policy debates with union struggle.
