Latin America Intelligent Age

Other countries have shown how public- private collaboration can drive AI infrastructure development. Stargate UAE, a partnership between the United Arab Emirates and OpenAI that includes establishing a 5GW AI campus, was the first project under the OpenAI for Countries initiative that aims to help build AI infrastructure and capabilities.15 OpenAI also announced Stargate Argentina in October 2025, its first partnership in Latin America, which includes plans for a data centre powered by renewable energy and floats an opportunity to partner with the Argentinian government to drive AI adoption across the country.16 Computing power is one core element of AI infrastructure, another is connectivity – although it is important to acknowledge that compact and efficient AI models, such as Meta’s Llama and other large language models (LLMs), are frequently optimized for ’edge computing’, enabling autonomous operation even without continuous internet access. Connectivity has been improving but is still a barrier in the region. The International Telecommunication Union estimates that 15% to 17% of Latin American households still lacked fixed broadband availability in 2024.17 Moreover, a persistent urban-rural connectivity divide threatens to widen inequality and affect the promise of AI as a democratizing force. World Bank research identified a fixed internet gap of over 30 percentage points between rural and urban households (42% and 74%, respectively).18 It is worth noting that expanding connectivity without growing digital skills might not solve the issue – a challenge discussed later in this white paper. Chile’s planned Humboldt cable to Australia and Peru’s National Broadband Plan are extending connectivity to remote mountain and Amazon regions. Humboldt is a collaboration between the Chilean government and Google that aims to improve regional bandwidth and latency.19 These types of public-private collaboration could help close Latin America’s connectivity gaps more quickly. Curate diverse, high-quality datasets Access to open data is improving in Latin America, but coverage and standards differ across the region. Brazil’s national portal provides access to more than 12,000 datasets across finance, health, agriculture and education,20 while Mexico’s federal and state portals publish statistics that have improved public safety.21 These initiatives can provide raw material for local developers and start-ups, although many datasets still lack consistency and metadata, limiting reusability. Some governments have created interagency councils to encourage interoperability and are experimenting with sector-specific solutions. For instance, financial regulators in Brazil and Mexico are piloting open finance frameworks, enabling banks to share data with customer consent. However, countries still need to develop clear standards for metadata, stronger privacy-preserving mechanisms and robust institutions to steward data beyond pilots and over the long term. At a company level, our survey results suggest that organizations also have an opportunity to improve their data capabilities. The lack of integration of data sources has inhibited them from obtaining deep insights across datasets. Furthermore, organizations generating no impact from AI are more than three times as likely to have low data maturity, compared to organizations that are generating impact from AI use. Develop responsible AI models The region is both implementing home-grown AI models and adapting existing ones. Mexico has announced its own sovereign LLM with the goal of strengthening national AI capabilities. CENIA’s regional initiative, Latam-GPT, is a Spanish and Portuguese open-source LLM built in partnership with over 30 institutions. It has been designed to capture the nuances of Latin America’s cultures and linguistics, which could be omitted by models from outside the region.22 Developing initiatives like this can take an extensive amount of time and resources. Commitment, communication and collaboration across the public and private sectors are essential to ensure home-grown models produce useful tools and positive outcomes. With the resource-intensive nature of creating home- grown models, Latin America, like much of the world, can benefit from adapting existing, open-source AI models to the region’s reality and challenges. Harness channels of AI investment Attracting and retaining investment requires the right enabling environment. However, AI investments in the region are still below what might be expected. A recent study commissioned by the United Nation’s Economic Commission for Latin America and the Caribbean found that Latin America accounts for only 1.6% of global AI investment, while representing nearly 6.3% of global GDP .23 Investment flows into AI development in Latin America through venture funds, development financing and hyperscalers. Key tools for attracting investment include clear policies, predictable regulation for AI and digital services, targeted incentives, reduced bureaucracy, strong contract enforcement and streamlined procedures for starting and operating a business. Latin America in the Intelligent Age: A New Path for Growth 12