Science & Tech · AI · 3 hrs ago
Latin America proposes shared infrastructure to build its own AI
Latin American experts proposed sharing AI infrastructure among countries and changing rules for using data in research.
The proposals were presented with the 2026 Latin American Artificial Intelligence Index, developed by the UN Economic Commission for Latin America and the Caribbean and Chile’s National Center for Artificial Intelligence.
The region has far fewer processors for AI than five countries used as international benchmarks: 8.5 graphics processing units per million people, compared with an average of 244.7.
Within Latin America, Brazil holds most of the region’s processing capacity, while Mexico has about 2% of the registered processors.
Mexico ranks eighth in the regional index, with 52 points out of 100, and has 466 AI-related companies and 63 specialized master’s programs.
Shared resources could widen access to costly computing equipment for researchers, universities and smaller businesses, and support AI models suited to regional languages and economic needs.
Clearer rules for research data use are also needed, while experts say interoperable technology could reduce reliance on a single supplier.
Latin America is seeking shared infrastructure and new data-use rules to expand its capacity to develop artificial intelligence.
The proposals were presented by specialists and technology industry representatives at the launch of the 2026 Latin American Artificial Intelligence Index.
The index says the five international reference countries averaged 244.7 graphics processing units per million people, compared with 8.5 in Latin America.
Within the region, Brazil accounts for 87.7% of identified graphics processing units and 92% of processing capacity.
The report says shared resources could widen access to costly infrastructure and support AI models suited to Latin American languages and economic needs.
- Who
- Specialists from the Economic Commission for Latin America and the Caribbean and the Inter-American Development Bank, along with technology industry representatives, put forward the proposals.
- What
- They proposed sharing infrastructure across countries and establishing new rules for using data in research to build Latin America’s AI capacity.
- When
- At the presentation of the 2026 Latin American Artificial Intelligence Index. The article was published on October 21, 2026.
- Where
- Latin America.
- Why
- Technology resources are concentrated among a few companies and countries, limiting regional research and development.
This story does not have two clearly opposing sides.
El Economista MX published its article about Latin America’s proposals for shared AI infrastructure.
Specialists and industry representatives presented the proposals alongside the 2026 Latin American Artificial Intelligence Index.
- International reference countries
- Germany, Spain, Estonia, Portugal and Singapore
- GPUs per million people
- 244.7 in the five reference countries on average; 8.5 in Latin America
- Brazil’s share
- 87.7% of identified regional GPUs; 92% of regional processing capacity
- Mexico’s ILIA 2026 score
- 52 out of 100; eighth place regionally
- Mexico’s AI workforce and industry indicators
- 466 AI-related companies and 63 specialized master’s programs
What people are saying
While it is true that not everyone has large independent infrastructure centers, I think we can create a regional system to democratize access to this technology.
It is super important, as I mentioned, to have interoperability. The idea is precisely to avoid excessive dependence on a single provider, which ultimately limits optimization, limits our ability to optimize costs and access technology, and limits the opportunity we have in the region to develop talent.