See all news
Índice Latinoamericano de Inteligencia Artificial (CENIA/CEPAL)
Tech in Mexico & LatAm

Latin America seeks its own GPUs! New index reveals a huge gap

The fourth edition of the Latin American Artificial Intelligence Index, presented by CENIA and ECLAC, reveals a high regional concentration of computing resources. Brazil holds 87.7% of the identified GPUs, while the Latin American average is 8.5 GPUs per million inhabitants, compared with 244.7 in a group of comparison countries. The document proposes exploring forms of regional cooperation.

The availability of computing accelerators determines the ability to train and run advanced artificial intelligence systems. Although more and more Latin American institutions are taking part in research, access to this infrastructure remains unequal and is an obstacle to local technological development.

ILIA 2026, prepared with the participation of CENIA and ECLAC, estimates that Brazil holds about 87.7% of the GPUs and 92% of the identified regional processing capacity. The comparison with five external countries shows a considerable gap in available equipment per inhabitant. The report also notes progress in specialized training and research centers, which creates opportunities if shared resources are strengthened.

Among the alternatives proposed, one stands out: building cooperative schemes for computing infrastructure, data access and talent. It is a proposal for discussion, not a fully implemented regional project. For Mexico and other countries, the central question is how to turn investment and cooperation into sustainable research and adoption capabilities.

At a glance

WhatThe 2026 Latin American Artificial Intelligence Index highlighted the regional concentration of GPUs and recommended exploring shared infrastructure.
WhenThe 2026 edition was presented on October 20; its indicators compare the current capabilities of twenty Latin American and Caribbean countries.
WhereThe study covers twenty countries in Latin America and the Caribbean, with Brazil holding the largest share of identified GPUs.
HowCENIA y CEPAL recopilaron indicadores de infraestructura, investigación, adopción y gobernanza y los compararon con referencias internacionales.
WhyThe computing capacity gap can limit research, model development and technological autonomy; cooperation would make it possible to share costs and resources.

Build your digital presence

Start today: write to us and we’ll reply within 24 hours.