In recent weeks, Mexican media and government officials have mentioned the alleged development of a supercomputer named Coatlicue, presented as a national project meant to drive artificial intelligence, scientific research, and technological innovation. But beyond the announcements — what do we really know? What could we expect if it were true? And what risks would it entail, given Mexico’s recent history with large-scale public projects?
1️⃣ The facts: what was said and how it compares
According to official press releases and media coverage, Coatlicue would be a supercomputer built under Mexico’s technological sovereignty policy, with an estimated investment of 6 billion pesos. Its projected performance would reach 314 petaflops, which would — at least on paper — make it the most powerful machine in Latin America and among the world’s top ten supercomputers. That would put it above Leonardo, the European pre-exascale supercomputer designed for scientific simulations, climate modeling, and artificial intelligence, which reaches about 250 petaflops (costing 240 million euros in 2022).
The government stated that Coatlicue would serve for scientific simulations, artificial intelligence, climate analysis, natural language processing, and predictive modeling in health, energy, and education. In other words, it would be a general-purpose computing infrastructure prioritized for public and academic use.
If it were to materialize, Coatlicue would surpass Mexico’s current supercomputer Miztli, located at UNAM, which reaches around 1.3 petaflops, and even Santos Dumont (Brazil), Latin America’s current leader with roughly 15 petaflops.
For global perspective, the most powerful computer in the world — Frontier, at Oak Ridge National Laboratory (U.S.) — exceeds 1.1 exaflops (1,100 petaflops). That means Coatlicue, even at its best, would have roughly a quarter of Frontier’s capacity — enough to position Mexico competitively and mark a significant leap for the country and the region.
However, no verifiable data has been provided about its location, hardware supplier, or related contracts. Everything so far points to a political announcement rather than a tangible project under construction.
2️⃣ First What if: What if Coatlicue were real and fulfilled its promise?
Let’s assume Coatlicue truly exists — and works. In that case, the project could mark a turning point for Mexican science. Not just for the infrastructure, but for what it would mean in terms of collaboration, talent, and technological independence.
A supercomputer of that magnitude would allow universities and research centers to run AI models that are currently unfeasible due to hardware limitations. It could, for example:
- Train language models in Spanish and indigenous languages without relying on foreign infrastructure.
- Perform climate simulations with greater precision, supporting environmental policies and disaster prevention.
- Accelerate medical research through rapid genomic and pharmacological analysis.
- Support AI applications in public education, adaptive learning, and large-scale data analytics.
Beyond its computing power, Coatlicue could become a symbol of national pride, proving that Mexico can develop world-class scientific infrastructure. On an industrial level, it could attract innovation projects that today migrate to countries like Brazil, Chile, or even Spain.
If managed with transparency and long-term vision, it could also reduce Mexico’s digital divide by providing access to state universities and startups that currently lack advanced computing capabilities.
In short, Coatlicue could be the start of a quiet revolution: an infrastructure designed not only to calculate, but to inspire.
3️⃣ Second What if: the darker scenario
Now let’s look at the other side of the coin. What if Coatlicue existed but became yet another grand promise lost to opacity, overcosts, or abandonment?
The recent track record of Mexico’s so-called “Fourth Transformation” megaprojects is not encouraging. The Maya Train, the Dos Bocas Refinery, and the Felipe Ángeles Airport have all faced criticism for budget overruns, lack of planning, and underwhelming outcomes. The recurring pattern: a lack of transparency and the politicization of technology.
Within that context, Coatlicue could become just another emblematic project — scientific on paper, political in practice.
There’s also a more concerning risk: the misuse of technological infrastructure for surveillance or censorship. Mexico has a documented history of government espionage, such as the Pegasus case, where digital tools were used to spy on journalists, activists, and opposition figures. A supercomputer capable of processing vast amounts of data could — in the worst-case scenario — be used to power control systems or manipulate information.
Another issue is bureaucratic inefficiency. If Coatlicue were managed by agencies with limited technical expertise or without clear access policies, it could end up underutilized or monopolized by politically favored institutions.
There are also financial red flags. When compared with similar projects worldwide, the numbers don’t quite add up. For instance, Alps, the Swiss supercomputer inaugurated in September 2024, cost 100 million Swiss francs (around 2.27 billion pesos) and reaches 435 petaflops — roughly 30% more performance than Coatlicue’s projection, at less than half the cost (6 billion pesos).
The likely result: a powerful but inefficient machine — heavy on symbolism, light on substance.
4️⃣ Conclusions: between hope and skepticism
Thinking about Coatlicue sparks mixed feelings. On one hand, it’s inspiring to imagine a Mexican supercomputer serving science — training AI models in Spanish, running climate simulations, or tackling complex public health problems. It would be a statement of ambition and technological sovereignty.
On the other hand, experience tells us that Mexico’s grand public projects often stumble on poor planning, corruption, and political protagonism. Science doesn’t advance through speeches — it thrives on continuity, transparency, and collaboration.
Ultimately, Coatlicue remains more of a powerful idea than a tangible reality. And like all powerful ideas, it can be used to build — or to control.
If this supercomputer ever comes to life, let’s hope it does so to enlighten knowledge, not to obscure it.
References:
Presidencia de la República. (2025, November 26). Presidenta Claudia Sheinbaum presenta “Coatlicue”, la supercomputadora del pueblo de México y la más poderosa de América Latina. Gobierno de México. https://www.gob.mx/presidencia/prensa/presidenta-claudia-sheinbaum-presenta-coatlicue-la-supercomputadora-del-pueblo-de-mexico-y-la-mas-poderosa-de-america-latina
TOP500. (2025). HPCG Benchmark — November 2025 list. https://top500.org/lists/hpcg/2025/11/
The Next Platform. (2024, September 23). A GPU upgrade for “Leonardo” supercomputer but not a budget upgrade. https://www.nextplatform.com/2024/09/23/a-gpu-upgrade-for-leonardo-supercomputer-but-not-a-budget-upgrade/
Swissinfo.ch. (2024, October 3). How Switzerland’s Alps supercomputer aims to advance AI. https://www.swissinfo.ch/eng/science/how-switzerlands-alps-supercomputer-aims-to-advance-ai/87659724
ETH Zurich. (2024, September 14). New research infrastructure: “Alps” supercomputer inaugurated. ETH News. https://www.ethz.ch/en/news-and-events/eth-news/news/2024/09/press-release-new-research-infrastructure-alps-supercomputer-inaugurated.html
Tags: ai, coatlicue, computerscience, mexico
Top comments (0)