Illustration of a luminous research network spanning South America

About

The backbone organization for equitable AI in Latin America.

The premise

AI is being built by a handful of firms, in a handful of countries, optimized for a handful of markets. Its benefits and its harms will not land evenly. The roughly six billion people who live outside the frontier-AI countries deserve a hand in shaping how this transition goes, not just a place to absorb what it leaves behind.

Meridian exists to close that gap.

↑ Back to top

The central question

How can the benefits of AI be broadly shared rather than concentrated among a small number of companies, countries, and individuals?

Between people

Wages, work, bargaining power

Between firms

Who captures the gains from data, compute, and capital

Between countries

A new great divergence between AI producers and AI consumers

Of power itself

Who controls the models, the infrastructure, the decisions

Everything we do traces back to some piece of this question.

↑ Back to top

What we are

Meridian AI Institute is a non-profit that acts as a backbone organization coordinating and giving visibility to the researchers, institutions, and innovators shaping AI across Latin America, so that its benefits are shared broadly, not captured by a handful of companies and countries.

We work from a position few institutions share: an office in San Francisco, close to the center of where AI innovation happens today, and a base in São Paulo, the core hub of Latin American tech — a head start built into our address.

↑ Back to top

How we work

We work through three gravitational forces: Map, Convene, Bridge.

Map. AI's impact on those left outside the rooms where it's being decided is already being studied and written about by economists, environmental researchers, journalists, and technologists, but no institution is mapping that work into a shared picture in Latin America. Meridian's first job is to be that map: identifying institutions, talent, and amplifying their work through an open ledger available to everyone.

Convene. Once the map exists, the next force is convening. By putting everyone in the same room at an annual in-person AI Forum, in Brazil first and then across Latin America, Meridian connects work done in isolation, credits and contextualizes it rather than duplicating it, and builds the conditions for aligned efforts and practical initiatives, like policy advising, to emerge.

Bridge. The third force carries that work outward. Meridian takes what the map shows to the people who act on it: policymakers, investors, and the AI community in Silicon Valley, where we represent those who choose to be represented in the forums that shape the AI revolution. We connect people and projects across the public-private divide, so that the benefits of AI are shared among all.

The Meridian AI Institute ultimately curates, amplifies, and convenes the people and institutions documenting how AI is reshaping lives, livelihoods, and ecosystems across the parts of the world with the least say in how it's built.

↑ Back to top

Why Brazil, why now

Brazil is our laboratory — an active legislative process, serious thinking around a national AI strategy, and an ecosystem of different players advancing the field, yet no institution built specifically around coordination and the central question of AI and Global South equity.

Latin America is our core geography — untapped potential, a digitally native and talented population, and all the raw components of AI: minerals, energy, and data.

The Global South is our ambition. We don't claim to speak for it. But once the Latin American ecosystems are aligned, we can fulfill our ambition to promote technology and knowledge transfer, while coordinating voices to have a say in the development and deployment of this secular technological disruption.

The window to shape how this gets built is open, not permanent. The institutions, norms, and funding flows for the AI era are being designed right now. We intend for emerging economies to be in the room when they are.

↑ Back to top