Half a Life Between Two Futures

AI generated image capturing the essence of the author’s work and reflecting his dual Brazilian American heritage. Drawn by ChatGPT in the style of Aluisio Carvão (1920-2001, Brazilian contemporary concrete artist), who happens to be the author’s great-uncle.

This year, I celebrate a milestone: half of my life in the United States. I was born and raised in Brazil, educated as an engineer in Rio de Janeiro, and started my professional career in a country still searching for its democratic footing. When I came to the United States, it was supposed to be temporary, as part of a development assignment in a global technology firm. For a few years, I moved back and forth, shuttling between Rio, New York, São Paulo, and Miami, translating not just language but business practices and cultural nuance. One year led to another. My family and I created roots. Eventually, we stayed.

I chose the United States as a country of social mobility and promise, where education and entrepreneurship could lift people into the middle class and beyond. I found opportunity here, not without effort, but with the reasonable expectation that work would be rewarded and dreams could be pursued. I became a U.S. citizen. Yet, the Brazil I left and the United States I embraced have changed dramatically. And now, as artificial intelligence emerges as a force of global transformation, I reflect on the trajectories of my two homelands and the divergent futures that lie ahead.

Two Democracies in Flux

Brazil exited its military dictatorship and embraced democracy just as I was entering adulthood. The optimism surrounding the 1988 Constitution gave way to decades of experimentation: economic booms, financial collapses, cycles of left and right populism, and a persistent inability to escape the label of “country of the future.” Political debates hardened into dogma, and democratic institutions faced rigorous testing.

Meanwhile, the United States, long seen as the model liberal democracy, promoted globalization while undergoing a slow process of deindustrialization. A wave of techno-capitalism brought extraordinary innovation but also deep inequality. The middle class, once the anchor of prosperity, began to erode. Wealth concentrated. Social mobility stalled. Populism took root, and democracy began to be questioned.

Today, both countries face similar ailments in different forms: political polarization, economic uncertainty and a loss of shared narrative about what progress looks like. And now, layered atop these existing tensions, comes artificial intelligence, a general-purpose technology that can redefine labor, power and the very notion of human agency.

The New Force

Technology has always driven transformation. The spinning jenny, the steam engine, electricity and computers each shifted the nature of work, the shape of markets, and the terms of power. But AI feels different. It operates faster, reaches deeper and has abilities, from predictive analytics, generative capabilities and reasoning systems, closer to replicating human cognition. Its diffusion happens not over decades, but in a few years. This timeline compresses our ability to adapt, straining institutional, social and regulatory frameworks in both the Global North and South.

The United States enters this moment with significant advantages: an unparalleled research ecosystem, deep capital markets and an entrepreneurial culture that rewards risk. It is locked in a race with China for global technological leadership, one that spans silicon, software and the standards that will govern how AI is used. But this race is not just about infrastructure. It’s about values, geopolitics and who gets to shape the future.

Brazil, by contrast, has a different kind of opportunity. While it lacks the foundational AI infrastructure of the United States or China, it sits atop critical inputs in the emerging AI supply chain, from rare-earth minerals to clean energy, and retains a creative, entrepreneurial population. Brazil must now decide: should it align itself with one of the global AI titans, or carve out an independent path?

The AI Supply Chain: Two Triads

AI generated image capturing the essence of the author’s work and reflecting his dual Brazilian American heritage. Drawn by ChatGPT in the style of Aluisio Carvão (1920-2001, Brazilian contemporary concrete artist), who happens to be the author’s great-uncle.

To understand Brazil’s opportunity, it helps to look at the structure of the AI supply chain through two conceptual triads. The first is the traditional AI triad: computing, algorithms and data. These three elements fuel the development of modern large language models.

The second reflects the physical infrastructure needed to scale AI systems: land, labor and energy. This triad captures the demands of hyperscale data centers for sustainable computing capacity.

Brazil performs modestly in the first triad. As of mid-2025, the country has nine high-performance computing clusters on the TOP500 list, representing just 1.8% of the global number of systems from the list, with only 122 Petaflops of peak installed capacity. A Petaflop is a way to measure how fast a computer can do math, especially with very big or very small numbers (called “floating-point operations”). One petaflop means a computer can do one quadrillion (that’s 1,000,000,000,000,000) math calculations every single second! Petrobras, the national oil company, accounts for six of the Brazilian clusters; academia, represented by the Laboratório Nacional de Computação Científica’s Santos Dumont supercomputer, with its 20 petaflops, lags. The United States, by contrast, exceeds 10,600 Petaflops, with the national labs and cloud providers leading the way.

The second pillar of the AI triad, algorithms, is fundamentally a product of human talent. In the field of artificial intelligence, this means individuals trained at the intersection of mathematics, computer science and increasingly, domain-specific expertise. These are the minds that design the architectures, fine-tune the models, and push the boundaries of what machines can learn. Unfortunately, Brazil has historically struggled to retain such talent. Our research institutions are underfunded, salaries are not competitive, and many of our best-trained Ph.D.s seek better-funded opportunities abroad.

This brings us to the third element of the AI triad: data. Here, Brazil can have a strategic advantage. With large, diverse populations, rich linguistic and cultural complexity, and extensive sector-specific data, in areas like healthcare, agriculture and climate, Brazil has the raw material needed to build AI systems that are both globally relevant and locally attuned. But realizing this potential will require intentional policies that enable data access while protecting privacy, ownership and equity.

While Brazil does boast a vibrant open-source and developer community, it lacks the scale, compensation structures, and compute access to meaningfully compete in the frontier model space. A frontier model is a highly advanced, large-scale AI model that pushes the boundaries of AI and is trained on extensive datasets with billions or even trillions of parameters. That said, not every nation needs to build models to benefit from AI. New techniques, such as parameter-efficient training and the use of open-weight models, lower the barrier to entry and create opportunities to innovate at the application layer.

The second triad (land, energy and labor), however, presents an opportunity. Brazil holds the second-largest rare-earth mineral reserves in the world. Rare-earth minerals are vital for AI because they are used in chips, magnets, sensors and cooling systems that make up AI hardware and data centers. Its vast territory has space for the construction of the large AI data centers that are becoming increasingly in demand and would benefit from a renewable energy mix, including hydropower and solar. The country’s workforce can be more easily retrained to build and operate data centers.

With the right policies, Brazil could position itself as a regional hub for sustainable AI data center infrastructure, providing the physical backbone of future innovation while building value-added supply chains at home. This requires a holistic approach that ensures environmental sustainability, social responsibility and long-term economic value. Brazil cannot afford to replicate past models of extractive growth. It must invest in ethical practices, human capital and innovation capacity to ensure its AI future is built not just on resources, but on resilience and relevance.

Technology that Looks Like Brazil

AI is not a monolith. Its value emerges not only from its creation but from how it is deployed. This means there is room for Brazil to define an AI ecosystem that reflects its cultural, geographic and social context.

Start with data. AI models are only as strong as the data they are trained on. Yet most available datasets are Eurocentric or Anglophone, failing to reflect the nuances of Latin American cultures, languages and institutions. Creating local, representative datasets, from legal systems to healthcare, agriculture to finance, can enable Brazil to build more effective models for domestic use. But this will require thoughtful policies around data ownership, fair use and the rights of creators.

The public sector could benefit dramatically. Bureaucracy remains one of Brazil’s great developmental bottlenecks. AI tools can streamline service delivery, eliminate antiquated structures and reduce corruption by standardizing decision-making processes. In healthcare and education, AI can expand access, personalize delivery and improve outcomes if equity and inclusion are prioritized from the outset.

Culturally, Brazil’s creative economy, from music to memes, literature to design, is a massive untapped asset. AI tools can support content creation, translation and global distribution if creators retain control and fair compensation. A new generation of AI entrepreneurs could emerge from Campina Grande, Cuiabá, or Curitiba as easily as from São Paulo, building on the country’s tradition of improvisation and resilience.

Risk and Responsibility

AI generated image capturing the essence of the author’s work and reflecting his dual Brazilian American heritage. Drawn by ChatGPT in the style of Aluisio Carvão (1920-2001, Brazilian contemporary concrete artist), who happens to be the author’s great-uncle.

The benefits of this technological revolution are clear, but so are the risks. Navigating them requires government action and the initiative of Brazil’s private sector, civil society and entrepreneurial community. Much of the early momentum must come from outside the state. Brazil’s public institutions, while essential, are often slow to act, weighed down by bureaucracy and budgetary constraints. By contrast, entrepreneurs and businesses are often more agile, better positioned to experiment, and capable of building tailored AI solutions that address real-world problems.

Brazil’s entrepreneurial ecosystem has already demonstrated resilience and creativity, thriving despite high regulatory burdens and volatile economic cycles. This same energy must now be harnessed to shape Brazil’s AI future. Startups can build context-aware applications using open-weight models. Traditional industries, from agribusiness to logistics, can adopt AI incrementally to improve productivity, reduce waste and open new revenue streams. Large firms have an opportunity, and perhaps even an obligation, to invest in local talent pipelines, share infrastructure and participate in pre-competitive innovation networks that can expand access to resources.

None of this means abandoning the role of the state, which still plays a critical part in setting guardrails and enabling innovation. But the responsibility to act is distributed. AI governance in Brazil must be a joint project, blending entrepreneurial dynamism with public interest goals. Accelerators, incubators, venture capital funds and university-industry partnerships must play a larger role in building this innovation architecture.

Labor displacement remains a real concern, but so does the risk of inaction. As AI systems automate not just blue-collar but also white-collar work, we will need new forms of work, new markets and new types of value creation. This is where Brazilian ingenuity can shine, by repurposing traditional jobs, developing innovative applications or exporting culturally specific AI tools across Latin America and the world.

The greatest risk is not that AI will replace us, but that we will fail to adapt fast enough to harness its power. Brazil cannot afford to be only a supplier of raw materials or a passive recipient of foreign technologies.

2050: A Tale of Two Futures

If we act wisely, the best-case scenario by 2050 is inspiring. Brazil builds a vibrant AI ecosystem focused on applications that solve local problems and serve regional markets. Universities evolve, in partnership with industry, to produce talent that stays. The country becomes a destination for green, high-capacity computing infrastructure, powering real-world solutions. We export software, not just soy.

The worst-case scenario? Brazil misses the moment. Innovation happens elsewhere. Data and value extraction continue without fair compensation. Our best minds emigrate. Our infrastructure decays. And we remain a passive consumer of technologies we neither influence nor regulate.

Brazil should not see this as a binary decision between alignment and independence. Like my own experience adapting across two countries, the path forward is to engage globally while building local capacity. Strategic partnerships can support development, but long-term progress depends on investing in talent, infrastructure and innovation at home. Brazil’s role in the AI era will be defined by what it builds, not just by what it imports.

The choice is ours. The window is narrow. But as someone who has spent half a life between two nations, I know that transformation is possible. Brazil has missed waves before. We cannot afford to miss this one.