Argentina has already chosen an AI industrial policy. It chose it through tax incentives for foreign infrastructure, not through a national strategy. At the nadIA Summit, held on 2 October 2026 at the Universidad Torcuato Di Tella, Cronista reported that the central panel concluded Argentina has no national AI strategy and has fallen behind Brazil and Chile, which have already put money behind their plans. The gap matters most because Argentina is courting the largest AI infrastructure project in its history.
The case for the current approach
The government's logic deserves a fair hearing. Argentina has spent decades deterring capital with unstable rules, and a strategy document written in Buenos Aires does not build a 500 MW campus. The Régimen de Incentivo para Grandes Inversiones (RIGI) was created by Title VII of Ley 27.742, sanctioned on 27 June 2024 and published on 8 July 2024. It was regulated by Decree 749/2024, signed on 22 August and published on 23 August 2024. Its premise is that predictable tax, customs and currency treatment attracts the long-horizon capital a country with Argentina's history otherwise cannot. Moving fast on investor certainty, rather than waiting for a perfect plan, is a defensible bet.
What was announced, and what has happened
On 10 October 2025 the Argentine government announced "Stargate Argentina." The Buenos Aires Herald reported that OpenAI had signed a letter of intent with the Argentine-U.S. developer Sur Energy for a Patagonian data center of up to 500 MW, with estimated investment of up to US$25 billion, to be structured under RIGI.
A letter of intent is not an investment. On 28 May 2026 the Buenos Aires Times reported that the Economy Ministry said Stargate had not been formally presented and was not listed among RIGI announcements. A researcher from the RIGI Observatory told the paper that 17 projects had been approved and 23 were under evaluation, and that none of them was Stargate. Sur Energy, OpenAI and the energy partner Genneia did not respond to the paper's requests for updates. I found no official update contradicting that report, so the project's status remains unconfirmed.
The lesson is not that Stargate is a failure. The headline figure did the political work of an AI policy while no binding commitment, local-capability obligation or public plan stood behind it.
What the summit actually asked for
According to Cronista, economist Eduardo Levy Yeyati argued that a plan must be concrete: funded, with data made available and regulatory sandboxes, meaning testing spaces with more flexible rules. Participants also included Daniel Yankelevich of Fundar. This is not a call for heavy regulation. It is a call for the enabling side of policy, which a pro-innovation view should welcome.
The sharpest warning concerned the contract. Levy Yeyati said that if every country competes on cheap energy and tax breaks, the result is a "race to the bottom." He proposed negotiating in blocs and requiring contracts that reserve compute capacity for local use, as Brazil does. His line was: "El contrato con el data center hace toda la diferencia" (the data-center contract makes all the difference). Without such terms, Argentina risks becoming an energy supplier to foreign compute, with the value added captured elsewhere.
Where proportionate policy should land
The summit's critics are right about the diagnosis, but the remedy should be narrow. Four things would improve outcomes without adding a heavy regulatory burden:
- A funded, short strategy. A few measurable commitments on compute access, public datasets and talent, backed by a budget line. The summit's own test was funding, data and experimentation spaces, not documents.
- Sandboxes over statutes. Argentina has no comprehensive national AI law. Rather than rushing a broad EU-style statute, it can let firms and universities test under supervised, flexible rules and write law from evidence.
- Local-compute terms in RIGI deals. RIGI beneficiaries are bound by the terms of their approval. Negotiating reserved capacity, research access or training obligations at that stage costs little and avoids a new regulatory regime.
- Transparency on approvals. The Buenos Aires Times quoted the RIGI Observatory describing opacity around the project's environmental implications. Publishing project status would let investors and citizens distinguish real commitments from announcements.
The regional dimension
Levy Yeyati's point about scale is worth taking seriously. He said that, apart from Brazil and perhaps Mexico, no country in the region has the scale to advance alone. Yankelevich recalled the 1980s Argentine-Brazilian informatics programme as a precedent for cooperation. For a market-oriented Argentina, a joint approach to hyperscaler negotiations is not protectionism. It is a way to raise bargaining power so that incentives buy something in return.
Conclusion
RIGI may well be the right instrument for attracting capital, and nothing here argues for abandoning it. But incentives are a means, not a strategy. If the Patagonian project materialises, the terms of the data-center contract will decide whether Argentina gains compute, skills and firms or only exports electricity. If it does not, the country will have spent a year on a headline instead of building its own AI base. Either way, a lean, funded plan and a negotiating position for the next deal are cheap compared with the investment being courted.