AMENAZA ROBOTO | interview
"Data centers are not just physical infrastructure"
The government is seeking to attract data centers and position Uruguay as a regional benchmark in artificial intelligence. Cuihong Cai, a digital governance specialist at Fudan University, explains why these investments must be accompanied by policies that strengthen the country's technological capabilities.

By: Miguel Ángel Dobrich.

25 de setiembre, 2026
President Yamandú Orsi this week pitched Uruguay as "a reliable regional hub" for artificial intelligence development, in remarks to heads of state and tech executives gathered in New York. Days earlier, the Secretary of the Presidency, Alejandro Sánchez, had laid out one of the government's targets: landing "four or five" new data centers during this administration.

The push comes as Google builds its Teros Project in Canelones — the company's first hyperscale data center in Uruguay — and as the state telecom company Antel expands its own infrastructure. For the country, the challenge goes beyond attracting investment: it also means anticipating the impact of these facilities and strengthening the State's ability to oversee them.

Two investigations by Amenaza Roboto published in 2026 documented environmental problems and flaws in how these facilities were assessed. The first, The Heat Behind the Cloud, used satellite imagery to identify a heat island around Antel's data center in Pando, and found that the Teros Project's environmental review hadn't accounted for that thermal effect. The second, Google's Neighbors, found that the project's file relied on an inadequate air-quality baseline, showed noise levels above permitted limits, and pushed nine environmental requirements back to the operational stage. In July, four neighboring families filed a legal challenge against the project's authorization.

These findings raise questions about the State's ability to oversee large-scale facilities. The arrival of global tech companies raises another one: what does Uruguay need so that these investments strengthen its digital capabilities instead of deepening its dependence on foreign providers?

To explore these challenges, Amenaza Roboto spoke with Cuihong Cai, professor of International Relations at the Center for American Studies at Fudan University and deputy director of its Center for Innovative Global Governance of Artificial Intelligence. Cai's answers reflect her own views.
What can happen to a small state's critical digital infrastructure when a much larger private hyperscale facility arrives? Is that a governance issue, a sovereignty issue, or both?

It is both. At the governance level, the issue concerns regulatory coordination and policy foresight: governments must evaluate not only individual projects but also their systemic impact on national digital infrastructure. However, at a deeper level it is also a question of digital sovereignty.

National data centers often carry functions that go beyond market logic — they support government services, public data storage, and strategic control over information flows.
If the operational capacity of these institutions is indirectly weakened by the arrival of much larger private infrastructures, especially those operated by global technology firms, the state may gradually lose practical control over parts of its digital ecosystem.

For smaller states with limited technical talent pools, this structural asymmetry becomes even more pronounced.

Your work looks at how developing countries can get locked into cycles of technological dependency. Does hosting Big Tech's infrastructure deepen that dependency, or can it actually help countries break out of it?

The answer is not purely negative or positive — it depends on how the infrastructure is embedded into national development strategies. Hosting hyperscale facilities can bring investment, connectivity, and access to global digital ecosystems. In some cases it may even accelerate technological learning.

However, if the domestic innovation system remains weak, the presence of foreign Big Tech infrastructure may reinforce dependency rather than reduce it.

Data centers are not only physical infrastructure; they are also nodes in global cloud, AI, and data governance architectures.

Without policies that build local capabilities — such as training engineers, supporting domestic cloud services, or ensuring fair access to computing resources — the technological trajectory of a country may become increasingly shaped by external platforms.

When a state both regulates a foreign tech giant and operates competing infrastructure in the same corridor, what does good governance look like? Is that tension manageable?

This type of tension is increasingly common in the digital economy. Good governance requires the state to act in two roles simultaneously: as a regulator and as a strategic actor responsible for safeguarding national digital infrastructure.

To manage this effectively, transparency and coordination across policy domains are essential. Infrastructure planning, competition policy, data governance, and national security considerations should not be evaluated in isolation.

Particularly for smaller countries, the key challenge is to ensure that regulatory capacity keeps pace with the scale and influence of global technology companies.

The tension is manageable, but it requires proactive institutional design and long-term policy planning.

¿Cómo explicaría a una audiencia general por qué un centro de datos operado por el Estado es categóricamente distinto de uno privado y por qué el deterioro de esa capacidad es más difícil de revertir que una pérdida en un balance contable?

A state-run data center is not simply a commercial asset; it is part of a country's digital public infrastructure. It hosts sensitive government services, national databases, and sometimes critical communications systems.

The expertise, institutional knowledge, and technical teams that support such facilities are also strategic assets.

If those capacities erode — for example through talent migration or long-term dependence on external platforms — rebuilding them can take many years. In contrast, a private data center is primarily organized around market efficiency and shareholder value.

Losing market share can often be compensated through new investment or restructuring. But rebuilding sovereign digital infrastructure involves rebuilding institutional capacity, human capital, and technological autonomy, which is far more complex.
Cuihong es autora de Cyber Governance in China: Balancing State Centrism and Collaborative Dynamics (Routledge, 2025), Global Cyber Governance (Fudan University Press, 2024) y Cyber Politics in U.S.-China Relations.
Amenaza Roboto