Craig's Blog · Essay
Latin America enters the second half of 2026 split almost cleanly down the middle. The “pink tide” of the 2010s has receded; a conservative wave has carried Chile, Bolivia, Honduras, Costa Rica, Peru, and now Colombia to the right, while Mexico and Brazil — together more than half the region’s people and output — remain governed by the left. The Pacific coast is turning into an unbroken conservative bloc; the Atlantic giants hold the other line. Between them sits a continent of nearly evenly matched electorates, most recent presidential runoffs decided by a handful of points.
Into this fracture, a different kind of alignment has become thinkable. Not a partisan victory for either side, but a pragmatic one — pragmatic in the older, Christian sense of judging a thing by its fruits for the human person. The AI Middle Way exists precisely to ask whether both camps can be brought to collaborate on a single shared project: activating the lower-middle class (LMC) into the AI economy on terms that serve human dignity rather than concentrate power. The question is not who wins the culture war. It is whether AI can be governed so that consciousness directs intelligence, and not the reverse.
The moral opening arrived on May 25, when Pope Leo XIV released his first encyclical, Magnifica Humanitas — a 42,000-word treatise on safeguarding the human person in the age of artificial intelligence. Signed on the 135th anniversary of Leo XIII’s Rerum Novarum, it deliberately casts AI as a new industrial revolution and revives the full architecture of Catholic social teaching to meet it: the common good, subsidiarity, solidarity, social justice, and the universal destination of goods. Leo argues that technology is neither evil nor neutral — it carries the intentions of those who finance, regulate, and deploy it — and he calls for AI to be “disarmed,” freed from the logic of military, economic, and cognitive domination.
This matters for Latin America more than for almost any region on earth, because Catholic social doctrine is the one moral grammar that both the progressive and conservative blocs still speak. The left hears in Leo a critique of extractive capital and inequality; the right hears a defense of the human person, the family, and labor against a technocratic paradigm. For a continent whose democratic tradition has always braided Athenian deliberation with Christian conscience, the encyclical offers a way to restore that alignment — to make AI governance a matter of shared civic reasoning rather than another front in the polarization wars. That is the pragmatic bridge the Middle Way seeks to walk across.
The constituency for that bridge is enormous and fragile. By the World Bank’s latest regional reckoning, roughly 38 percent of Latin Americans belong to the “vulnerable” class — above poverty but below stability — while the middle class has climbed back toward 43 percent and poverty has fallen near 25 percent. But scale conceals fragility: of the region’s roughly 400 million nominal middle-class people, about half remain one lost contract, one illness, or one climate shock away from sliding backward. This is the true LMC: not the secure bourgeoisie, but the aspirational majority for whom mobility is real yet reversible. Depending on where one draws the line, they are between a third and nearly half of all Latin Americans — and the decisive swing bloc in every close election on the map.
What holds them back in the AI era is data colonialism. Their behavior, language, faces, and transactions are harvested to train models they neither own nor govern; the value flows north to US hyperscalers and, increasingly, east to Chinese platforms. They are treated as raw material for foreign systems rather than as participants in their own digital development. Sovereign capacity — local data centers, local models, local governance — remains thin, so the LMC consumes AI without ever activating economically through it. Breaking that extraction, and converting the LMC from a data source into a market and a citizenry, is the material core of the Middle Way’s proposition.
The two superpowers are already pulling the region toward opposite poles. Chinese AI hyperscalers plan to spend upward of $70 billion in 2026, explicitly expanding data-center construction across Asia, the Middle East, and Latin America, backed by cheap energy and open-weight models. US hyperscalers, by contrast, are projected to spend on the order of $630 billion, dominating the frontier and the compute. The rivalry has a hard geopolitical edge: Washington’s “Shield of the Americas” summit in Florida this March gathered more than a dozen leaders around security and hemispheric influence, and the US has weighed roughly $1.5 billion to counter a Chinese-built port in Peru. Each investment nudges a government toward one camp or the other.
The Middle Way’s wager is that neither pole serves the LMC. American frontier capital tends to extract data and export dependency; Chinese infrastructure tends to entrench surveillance-friendly architectures. A third path — human-centered, sovereignty-respecting, oriented to LMC activation — is the only one that lets a nation take investment from both without being captured by either.
The alignment cannot be declared; it must be built one country at a time, because the incentive is different in each.
Mexico is the natural anchor. President Sheinbaum has made technological sovereignty a stated priority — a national ethical-AI declaration in January, a Digital Transformation Agency, sovereign-model and supercomputer ambitions, and a promised public debate on AI regulation after the World Cup. Her incentive is to convert the $277 billion Plan México investment portfolio into genuine domestic capacity rather than foreign dependency. Human-centered AI that activates the Mexican LMC is the pragmatic completion of her sovereignty agenda.
Colombia, having just elected the hard-right Abelardo de la Espriella over the leftist Iván Cepeda, is the harder, more urgent case. A country this freshly ruptured needs a project that transcends the rupture. Here Mexico’s leadership becomes decisive: if Sheinbaum can demonstrate that sovereign, dignity-centered AI is not a partisan program but a national-development one, she offers Colombia’s incoming government a model it can adopt without ideological cost — a way to serve the same fragile majority that both De la Espriella and Cepeda were competing to represent. Mexican precedent lowers the political price of a Colombian “yes.”
Venezuela is the reconstruction case. With Maduro captured in January and a transitional order now governing under heavy US tutelage, the country faces the question of what fills the vacuum. Ideology has failed twice there — 21st-century socialism and now externally imposed transition. The Middle Way can help “resolve” post-Trump Venezuela not by choosing a side but by offering LMC activation as the substance of reconstruction: rebuilding a shattered economy around the digital inclusion of ordinary Venezuelans rather than around whichever faction holds Miraflores. Peru, with Keiko Fujimori’s narrow win, and Brazil, heading into a knife-edge October election, complete the sweep — each with a fragile LMC majority that no partisan program has managed to secure.
Much of this hinges on the arc of US power. Whatever one’s read of the current administration, the constitutional clock is fixed: a second-term president is a lame duck, the November 2026 midterms loom, and hemispheric dominance built on one leader’s personal pressure tends to soften as that leader’s leverage wanes. The Middle Way should plan for a post-Trump landscape in which Washington’s grip on Latin America loosens — not to celebrate it, but because a less coercive hemisphere creates space for genuinely multilateral, nation-led AI governance rather than alignment by intimidation.
Regionally, the institutions are not yet ready to fill that space. CELAC, Mercosur, and the Andean bodies remain fragmented on AI, and the region still lacks a unified position; most governments have imported European-style principles without building their own implementing capacity. That vacuum is itself the opportunity: a coalition organized around LMC activation and the encyclical’s moral frame can supply the shared language these bodies lack.
Project forward to 2032. If the US (frontier models, compute) and China (deployment, open weights, cheap energy) settle into a complementary rather than purely rivalrous equilibrium — each doing what it does best — the terms of the game change. Latin America would no longer be forced to pick a patron. And the hundreds of millions of newly activated LMC citizens across the Global South — digitally included, economically mobile, governed by their own sovereign systems — would emerge as a transnational market and a political constituency in their own right. A bloc that size, aligned across the old left-right divide by a shared stake in dignified AI, could exert political effect no single government commands today.
That is the pragmatic bridge in full: a Christian-Athenian conscience supplying the ethics, the LMC supplying the constituency, sovereignty supplying the structure, and 2032 supplying the horizon. One nation after another, the Middle Way asks the same question of progressive and conservative alike — not whose intelligence will govern, but whether intelligence will finally be made to serve the human person. In Latin America, for the first time, both sides may be ready to answer yes.