Interview with Ezequiel McGovern.
The introduction of Digital Tribulations, a series of intellectual interviews on the developments of digital sovereignty in Latin America, can be read here. 
I met Ezequiel at the first meeting of IT workers in Argentina, held in Buenos Aires in March 2026. Esteban—a friend from the Argentine IT workers’ union, whose interview will be published next—pointed him out to me as someone especially worth speaking to. Ezequiel was wearing an ARSAT T-shirt, which, as I soon learned, referred to the Empresa Argentina de Soluciones Satelitales Sociedad Anónima, Argentina’s state-owned telecommunications company.
Founded in 2006, ARSAT was created at a moment when Argentina risked losing control over key orbital positions assigned to it for satellite communications. From that starting point, it grew into a central actor in the country’s technological infrastructure: developing and operating geostationary satellites, expanding the federal fiber-optic network, supporting digital television, and building domestic data-center capacity. As such, it has become one of the clearest institutional expressions of Argentina’s attempts to build a measure of technological and digital sovereignty.
Ezequiel is an engineer who has been working on these issues since long before “digital sovereignty” became a fashionable term in policy circles. My impression was that he has seen many of these debates come and go, and that he has a particularly clear sense of what should—and should not—be done if Argentina is to achieve technological autonomy under its specific conditions. 
The interview took place online. In it, we discuss the history and role of ARSAT; the relationship between public infrastructure, digital sovereignty, and national development; the material constraints behind data centers and AI infrastructure; the strategic importance of satellites, fiber optics, and cloud systems; finally, Latin America’s dependence on foreign hardware and software and the possibilities and obstacles for regional technological cooperation.
***
What do you do, and how did you get involved with the issue of digital sovereignty?
I’ve been working in this field for 15 years. My specialty has always been industrial and high-availability IT. From there, I worked for Yacimientos Petrolíferos Fiscales (YPF), Argentina’s main energy company, for many years, overseeing its automated lubricant warehouse. I also worked for Telefónica, always on large-scale projects where there were problems to solve or particularly complex systems that had to be integrated.
That’s how I started working with data centers, initially from within the public sector. I first worked for the City of Buenos Aires, then for the National Office of Information Technologies (ONTI). As I saw the challenges we were facing, I began advocating for the idea of having a national-level data center with world-class capabilities. We couldn’t continue relying on many small data centers that lacked proper budgets and qualified personnel and eventually ended up underfunded and plagued by problems.
I brought that vision to ARSAT during a meeting where the board invited me to explain the project: what a new data center would be useful for, what business opportunities it could enter, and what possibilities it offered. I gave a presentation of roughly half an hour to the board. They were interested in the range of potential businesses that could emerge, at a moment when cloud computing was just beginning. Amazon had only recently started investing heavily in the cloud, but the idea of building something similar was already taking shape.
 
After that meeting, the company’s president at the time, Pablo Tognetti, said to me: “Well, when are you going to start working with us to make this project happen?” So, I began with a part-time contract for six months, and then I joined full-time. We built the data center. I was responsible for the infrastructure design: everything related to servers, connectivity, and all the procedures—together with the networking team—to become an internet service provider, obtain IP address allocations from the Latin American and Caribbean Internet Addresses Registry (LACNIC), create the autonomous system, start announcing our addresses on the internet, and, in general, develop the entire project needed to create our own cloud platform, which is what we ultimately achieved.
When was ARSAT founded, and what does it do?
ARSAT was founded on May 20, 2006, in response to a need Argentina had at the time. There was a geostationary satellite operated by Nahuelsat, whose use had been outsourced and privatized. The company was required to launch another satellite to preserve Argentina’s orbital slot, but it failed to fulfill the contract and announced that it was leaving the country.
To avoid losing those orbital positions, the national government submitted a proposal to the ITU to continue the process through a national company. That was how ARSAT was created. From there came the ARSAT-1 and ARSAT-2 projects, the geostationary satellites that are currently in orbit. That was the original reason for its creation. But there was also recognition of the need for communications infrastructure in a country as large as Argentina—which I believe is the eighth-largest country in the world by area—where connectivity is extremely challenging and had been severely underinvested. To give you an idea, at that time there were fewer than 8,000 kilometers of fiber-optic cable in the entire country, counting all providers combined. Private companies were unwilling to make the necessary investments to deploy high-quality fiber-optic infrastructure across the territory.
That concern led the government to act, and ARSAT took responsibility. The Federal Fiber Optic Network was created, which today consists of 35,000 kilometers of fiber operating throughout the country. Alongside the Federal Fiber Optic Network, it became clear that we also needed a data center, which was exactly what I had been advocating for. That’s why they called me: to take advantage of those services and build an internationally competitive data center inside Argentina, with the data remaining in Argentina.
So the process began with nationwide telecommunications and the satellite segment. Then came the Federal Fiber Optic Network and the data center. Alongside that, there was also the rollout of Digital Terrestrial Television, which was tied to telecommunications because spectrum had to be freed up: frequencies used for analog television had to be reassigned to mobile networks and other technologies. That’s how the project to transition all television broadcasting to digital began. ARSAT also took on that task, and digital terrestrial television coverage eventually reached almost 85% of the country, with the remainder covered through satellites.
I think it’s great that a public company plays such an important role in developing technology. But it also seems that the government has changed significantly. President Milei has recently declared that Argentina will become an artificial intelligence hub. What would be the role of digital sovereignty, or of an actor like ARSAT, in relation to this shift?
The key issue that governments always forget when they make these kinds of announcements is that these are extremely large investments. These are highly complex systems, and they also require enormous amounts of energy to operate. Today, all of those conditions—combined with ongoing wars and disruptions to global supply chains—are wreaking havoc on the data center market.
In fact, I was reading that many of the data centers planned for this year in the United States will not be built because the necessary transformers cannot be manufactured. Electrical infrastructure doesn’t expand overnight; it’s not like software. Building infrastructure is extremely complex. And given the current geopolitical situation and supply-chain disruptions, for example, copper is hard to obtain, there is a shortage of specialized steel for transformers, and trade routes are being disrupted.
In that context, having your own data center and technological sovereignty means having something smaller and better suited to the country’s needs, rather than depending on investments that we know are unlikely to arrive—especially in the current international environment. That’s why we always envisioned relatively small data centers of around two megawatts, nothing extravagant. If there were a need for artificial intelligence or intensive computing, then we could consider building a larger facility. But the investment required is enormous. To give you an idea, a single AI server costs millions of dollars and consumes around 30 kilowatts. To build something reasonably capable, you need to think in terms of 35 megawatts. That means well over one billion dollars in investment between equipment and construction.
Those are figures that are far beyond the current model of the Argentine economy. And globally, I don’t think that investment is going to come to Argentina under these circumstances. If even in the United States the number of data centers planned for construction this year has already been cut in half, all of this is going to be delayed. That is also encouraging companies to make better use of the infrastructure they already have. Google and other companies have managed to run models four to six times larger using the same amount of memory. So they no longer need to expand as rapidly as they originally planned. That affects the entire production chain. On top of that, we now face external constraints that nobody anticipated. No one expected a war with Iran this year. And that directly affects one of the most important variables in the economy: the cost of energy.
In that regard, Chile developed a system to assess the environmental impact of data centers. What do you think about that? Can it help determine where data centers should be built?
We have always envisioned locating data centers within industrial hubs that already have environmental conditions clearly defined. Places where water management has already been addressed, where any water used can be recycled, where electricity supply is guaranteed, and where the facilities are located away from urban centers.
In fact, one of our earliest plans was to build a facility in the Bariloche technology park. We had a World Bank loan of 200 million dollars approved for that purpose. But after a change of government, the loan was redirected elsewhere and was not used for the data center. Even so, the project remains active. We are still in discussions with the World Bank, and when the political context changes, the idea is to move forward with construction. But we have always approached it this way: locating data centers in places already designed to host businesses or industries with high energy and water consumption.
Fortunately, in Argentina we do not have a shortage of space. We have many possible locations for building data centers, unlike Chile, which faces much greater geographic constraints as well as limitations in energy availability.
So what Milei said about building data centers in Patagonia would be more difficult to implement?
 
Yes, at the moment it is much more difficult. We also have a connectivity problem in the south. We would need more submarine cables landing there, as well as a connection through the Pacific. We had planned a trans-Pacific fiber-optic link together with Chile, but the current government halted the project. That would have given us direct internet access, through dark fiber, to Asian markets via Australia and the broader Asia-Pacific region. It would have allowed services to be provided to those markets from southern Argentina. Today, by contrast, all traffic has to pass through Miami, which adds nearly 300 milliseconds of latency. And latency is another factor that companies examine very carefully.
But the biggest issue with the investment figures being discussed—the 25 billion dollars often mentioned—is that, globally, that level of investment is no longer expected to go into AI servers. OpenAI, for example, failed to follow through on its most recent memory-chip purchases from Samsung and SK Hynix. So the growth projections they had previously announced have already weakened. They had claimed they would consume nearly 60% of global memory-chip production for three years. This month, however, they bought nothing. As a result, those projections are now considered unreliable, and people are waiting to see how purchasing trends evolve. Because if companies continue buying memory, that means they are still buying servers and still building new data centers. If those purchases stop, it suggests that the projected number of data centers will not materialize as expected.
And given the infrastructure limitations we have in southern Argentina, the challenge looks even greater. To give you an idea, today ARSAT is the only company with fiber-optic infrastructure along Route 40. Along Route 3, we are building what would become the Atlantic coast segment. The Andean region already has fiber, but a data center without connectivity is useless, especially for this type of service.
As for energy for data centers in the south, there are several projects that use gas turbines capable of generating enormous amounts of power. Mining companies are already using this approach. So energy could be available through on-site turbines powered by flare gas. In fact, that is much more environmentally friendly than simply burning the fuel, because if that gas is not used for electricity generation, it still has to be burned or vented into the atmosphere. But these are very remote and inhospitable locations. Where you place the facility matters a great deal. There is also very limited water availability, which means those data centers could not rely on water cooling. They would have to use direct-expansion cooling systems, which completely changes the design.
Despite the cold climate, you cannot rely on the weather itself, because it is also an arid environment with a great deal of dust. And you cannot have a data center exposed to dust. You can install multiple layers of filtration, but then those filters have to be maintained, and that introduces additional costs that work against you.
When we talk about governance of the technology stack, satellites seem to be something that receives relatively little attention, even though they are extremely important. What are the advantages of having your own satellites rather than relying entirely on private companies?
The most important advantage, in a country like Argentina that struggles with balance-of-payments constraints and foreign currency shortages, is that with two national satellites you can provide communications across the entire country while paying in local currency. Argentina spends roughly between 60 and 80 million dollars a year on satellite communications. Of that amount, around 60 million is absorbed by ARSAT. Without ARSAT, those dollars would flow abroad. So from a balance-of-payments perspective, it is enormously beneficial. ARSAT is a profitable company. You can pay for its services in pesos and continue receiving them; otherwise, you need hard currency available to pay foreign providers.
That is the economic side. From the perspective of sovereignty, you are also developing an industry that improves the quality of production at every level: electronics, engineering, metalworking, propulsion systems, systems integration. It allows you to create a network of SMEs that INVAP can contract, and those companies begin to gain international recognition for other satellite projects because they have developed expertise and high-quality processes.
In the satellite industry, quality has to be guaranteed from beginning to end. It is not like other products. There are insurance requirements, verification procedures, and extremely demanding standards. All of that helps create new industries and higher levels of quality. And in a country as large as Argentina, satellites are essential. There are many remote areas—the foothills, the Andes, the northern regions with ravines and valleys—where installing fiber-optic infrastructure is extraordinarily expensive or simply impossible. Sometimes you can build it, but then a flood destroys it and cuts the entire line. So you cannot rely solely on fiber. You absolutely need satellite capacity.
Satellites are also critical for strategic purposes: border control, communications for the Armed Forces, the fishing fleet, and Antarctica. All of these functions require satellites if you want to exercise sovereignty. Moreover, satellites also shape symbolic and territorial sovereignty. With ARSAT-1 and ARSAT-2, we provide coverage over the Malvinas/Falkland Islands and Antarctica, territories over which we claim sovereignty. Providing communications through those satellites is another way of reaffirming the country’s rights, by saying: we made the investment, we are providing coverage, and we are guaranteeing services throughout Argentine territory and for all its inhabitants. And the costs are competitive as well. It is not as though we are losing money to achieve that. The company is profitable.
Of course, we do not have the most advanced communications technology today, because these satellites have been in orbit for more than ten years. When they were launched, having five megabits of internet bandwidth was considered very good. The challenge is that geostationary satellites last between 15 and 18 years, while technology evolves very quickly. We will need to consider how to improve services, perhaps by launching ARSAT-SG, which is designed to help close the digital divide and provide high-quality coverage across the Andes, the foothills, and northern Argentina. That would allow us to offer bandwidths of around 50 megabits per second.
It makes me think a bit about the postal system: the advantage is not only that it works and connects everyone, but that it also functions better when it is public. In that sense, what do you think about interpreting infrastructure lack in Latin America as a form of digital dependency?
Yes, absolutely. That is one of the things we try to avoid whenever possible. For example, in choosing the cloud management platform software we use, we selected Apache CloudStack, which belongs to the Apache Foundation. That gives us access to the source code and ensures that no company can suddenly change the licensing model or prevent us from using it.
We have already seen that happen with VMware. VMware, which was widely used for virtualization services, changed ownership, dramatically increased prices worldwide, and forced many European SMEs to migrate to open-source solutions. Excessive cost increases push the market toward concentration, where only the major cloud providers can survive while smaller companies can no longer offer the same services. Fortunately, on the open-source side we have KVM, which provides essentially the same capabilities for virtual machines and cloud services and remains available.
Where Latin America truly remains dependent is in computing hardware. We depend on technologies developed by American, Asian, or European companies. We do not have chip fabrication plants, and we do not manufacture memory. We do have the human talent needed to design these technologies, but building a world-class semiconductor fabrication plant requires billions of dollars in investment and an industrial ecosystem that simply does not exist here. That is why those facilities are in Taiwan, and why the issue is so sensitive.
In fact, on April 18 we will be giving a talk in Avellaneda about the need to think seriously about rebuilding a microchip and microcontroller industry in Latin America. Not with the goal of competing at the cutting edge with three-nanometer technology, but by starting with 90- or 45-nanometer processes. For example, you should not have to wait for a chip for a washing machine to arrive from China. Nor should a military drone depend entirely on foreign suppliers. These are not highly complex technologies, and they could be produced with those manufacturing capabilities. We are talking about much smaller investments—less than one billion dollars—compared to the 60 billion dollars that a cutting-edge fabrication plant can cost.
For regional needs, that would be extremely important. Brazil is already far ahead in this area. Brazil clearly understands that the world is organized around strategic resources and acts accordingly. Unfortunately, Argentina at the moment is, I would say, functioning like an occupation government. It is a government focused on dismantling the country’s technological and developmental capacities. If you look at the sectors they have targeted, they are precisely the ones that allow a country to differentiate itself, grow, and integrate. This government will clearly attack all of those areas. We have always looked to Brazil. Brazil has long had an imperial vision of itself as a global actor and behaves accordingly, even at the level of its elites. In Argentina, when I studied our elites, I found that they have largely been rent-seeking and extractive elites. They see themselves as the owners of Argentina, while the rest of us are merely tenants. And that is how they treat us, both politically and economically.
Brazil is an interesting case. Digital sovereignty is discussed there quite extensively, both among the public and within the state. For example, the development of payment infrastructure such as Pix has transformed people’s lives, especially those of poorer citizens. But at the same time, despite all that discourse, American companies are still contracted for certain services. That can also become a trap, because state-led innovation can end up being captured and put at the service of foreign corporations.
Yes, absolutely. Here in the Province of Buenos Aires we have Cuenta DNI, which aims to become something like an Argentine version of Pix. It is a locally developed system, and that is very important. The Province of Buenos Aires also has its own data center and its own development teams. In many cases, these solutions emerge because of budget constraints. There simply is not enough money to hire Accenture or some international consulting firm. That necessity forces you to develop capabilities locally.
In ARSAT’s case, for example, we developed Cine.Ar Play, which allows people across the country to watch Argentine films from the National Institute of Cinema and Audiovisual Arts (INCAA) free of charge. We also developed local content delivery network solutions. For example, that is the infrastructure used by the Official Gazette to publish government documents. So solutions already exist that address specific needs and that, if they did not exist, would make us dependent on companies elsewhere in the world. And they can be built using open-source technologies—good, effective, and affordable, as we like to say—precisely because we do not have the budget for anything else.
Do you think there is something to learn from a more popular Latin American tradition when thinking about digital sovereignty? 
I think there is. We have a fairly good understanding of where we can add value and which costs would not be justified. That is why we focus on software development and infrastructure solutions based on our own engineering capabilities, without relying on third-party engineering, while trying to make everything as maintainable as possible.
For example, when we built the data centers, we did not buy a turnkey solution. We became deeply involved in the project itself. We tried to ensure that each component had the lowest possible level of complexity so that maintenance would be easier. I always say that one of the main enemies in engineering is complexity. Sometimes it is unavoidable, but in many cases it leads to larger problems when unexpected behavior appears—which it always does.
Black-box solutions are attractive because they provide something that is already solved. But when a problem arises, you no longer have visibility into what is happening inside. We try to avoid that. Even if it is not always the most efficient approach, we prefer to keep everything separated, understand the system as a whole, reuse components, and know exactly which part is causing which problem. And if there are things that cannot be avoided, then so be it—but at least we want them clearly identified.
We do not think in terms of building the entire technological stack in Latin America or manufacturing everything locally. But we do believe in developing the areas where we can differentiate ourselves, where we can add value, and where foreign companies would charge enormous sums to adapt solutions to our needs. Because otherwise you become dependent on their development roadmap.
A clear example is Oracle databases, which are still used extensively throughout Latin America. We have always advocated moving toward PostgreSQL, which is open source and offers virtually the same performance for normal use cases. But fashions and managerial fears carry a lot of weight: “I don’t want to move away from this because it already works.” The problem is that these technologies are extremely expensive, and they can also leave you vulnerable to situations where a company suddenly says: “We are no longer selling licenses to Argentina.” Then, overnight, you lose access to updates. That happened to Russia, it happened to Cuba, and it is happening to Iran right now. In the long term, the risks are much greater if you depend on those technologies.
I’ve spoken with people across Latin America, and everyone tells me the same thing: yes, we need more cooperation at the Latin American level, but it’s very difficult. What can be done to strengthen cooperation among countries?
The most important thing is creating spaces for people to meet. Those spaces do not exist, deliberately so. There are no regional technology forums, even though there easily could be. Brazil, for example, could host them. Europe has many more of these spaces: user groups, regional conferences, and so on. We have often been invited to conferences organized by the KVM Foundation and to regional technical conferences in Europe. Obviously, we participate mostly online because we are not allowed to travel, but we still contribute in our own way. And everything we develop ends up being used around the world.
I think that is the direction we should pursue. But there are always what I call “agents of chaos,” people who actively undermine opportunities for communication. That happens even within Argentine universities, where research groups working on exactly the same topics often do not even know each other exists. Right now I am very focused on energy issues, and I keep finding universities carrying out similar analyses without realizing that other groups are doing the same work. So I often end up acting as a kind of matchmaker between research teams: “Look, this faculty is working on this topic—go talk to that other group,” and I put researchers in touch with one another. That is when synergies begin to emerge. But we have built a model of isolated islands. And those islands are a major obstacle to scientific and technological development. We need to start breaking them down and creating larger collaborative communities. There is also a very Latin American problem, which is the “me first” mentality: “I do it better, everyone else does it worse.” These ego-driven competitions also work against us. Egos can sometimes be useful for driving projects forward, but when we need to pursue a common cause, they become a major obstacle. And many projects ultimately fail because of that. There is also a cultural element, if you like: we are more expressive, and disagreements can become irreconcilable very quickly. And that happens not only across Latin America but also within individual countries.
And what about Uruguay? What do you think of ANTEL, which seems to have a somewhat more public-oriented approach?
We have had two meetings with ANTEL because we have many things in common. They do not have satellites or satellite capacity, but they do have fiber coverage throughout Uruguay and they operate data centers. Their situation is somewhat mixed, though: they have data centers with their own equipment, but they also purchase capacity from Amazon and other major providers. It is a hybrid model.
That said, Uruguay has a major advantage: it is a very small country. That means the cost of building and maintaining high-quality infrastructure is much lower. Our challenge is that Argentina is a very large country and infrastructure costs are astronomical. Uruguay has a system that works very well. What they lack is professionals, because Uruguay loses many highly skilled workers. A great many people end up moving to Argentina or to other countries due to economic conditions and quality-of-life considerations. If those conditions improved, they could retain that talent and advance much further, because their infrastructure and expansion costs are significantly lower.
Brazil, by contrast, has strong integration between industry and universities and has always maintained an industrial focus. Petrobras is also a giant company that drives much of the country’s scientific computing, data center development, and communications infrastructure. They have a very clear understanding that these capabilities are necessary if they want to continue developing and growing.