Mexico's Free AI School: What the Government Promised and What Its Data Really Shows ...

TEXT START: Mexico's tuition-free AI training program is still widely described as a "10,000 student" initiative, but the government's own Second Government Report puts the real first-cohort number at roughly 12,000 enrolled and just 3,001 graduated as of September 1, 2026.

The Dissection

This is a credibility audit disguised as a development story. It punctures inflated launch numbers, exposes attrition and construction delays, then rescues the program through comparison: Mexico has a public option, while the United States does not.

The article mostly measures inputs and activity—applications, enrollment, certifications, GPUs, projected chatbot usage, and budget. It provides no evidence that graduates obtained durable jobs, higher wages, economic independence, or control of AI capital. The chatbot is a genuine operational success, but it does not validate the school.

The Core Fallacy

The article treats a current shortage of specialized workers as proof of durable future employment demand. That is a supply-side answer to a demand-side destruction problem.

Under the Discontinuity Thesis, P1 means AI progressively performs cognitive work more cheaply and effectively. Training more humans in AI-adjacent fields does not defeat P3 if employers eventually need fewer humans. A reported 77 percent employer difficulty filling specialized roles may indicate a temporary skills mismatch, but it does not prove that 25,000 newly certified workers can be economically absorbed.

The program also blurs categories. It is called an AI school, but its five tracks include Java, cloud computing, cybersecurity, data analysis, and AI. The 25,000-graduate target is therefore not evidence of 25,000 AI specialists, much less 25,000 people made indispensable by automation.

Hidden Assumptions

• Certification by major companies is treated as equivalent to employability.
• Graduation is treated as equivalent to productive participation.
• The current hiring gap is assumed to persist after AI adoption accelerates.
• The second cohort's enrollment is treated as momentum despite having no reported outcome data.
• Coatlicue's computational capacity is treated as national AI capability, despite construction delays and unresolved operating costs.
• Mexico's creation of a public program is treated as evidence of success because the United States lacks an equivalent.
• Latino-focused training is assumed to protect workers from displacement without showing that graduates become Sovereigns or indispensable Servitors.
• The target of 25,000 certified graduates per year is assumed to reflect market absorption rather than bureaucratic throughput.

The most revealing number is the gap inside the first cohort: roughly 12,000 started, 3,001 graduated, and 1,532 remained active. Roughly 7,467 are neither reported as graduated nor active. That is not merely a “lumpy curve.” It is a large unresolved attrition signal.

Social Function

Classification: transition management, prestige signaling, and partial truth functioning as ideological anesthetic.

The article gives policymakers a usable narrative: the state recognizes AI disruption, trains citizens, builds compute, and is still ahead of the United States. The “largest AI and coding school” claim and Coatlicue's petaflop figures supply national prestige. The US comparison establishes a low bar: Mexico appears successful because another country has done less.

This is not pure propaganda because the article admits delays, cost criticism, incomplete delivery, and weak cohort outcomes. But its final “uneven head start” framing converts missing proof into optimism. It softens the central question: whether these programs preserve human economic necessity after cognitive automation.

The Verdict

Mexico has created an early-stage training pipeline, not a counterexample to the Discontinuity Thesis. The data show serious attrition, delayed infrastructure, and no demonstrated employment outcome for graduates. The program may produce a narrow layer of Servitors, transition intermediaries, and public-sector operators. It does not restore the mass employment-to-wage-to-consumption circuit and does not turn trainees into owners of AI capital.

The article is valuable as an audit of state execution. Its systemic conclusion is weak. An uneven head start is a head start only in the race to build institutional AI capacity—not in the race to preserve mass human productive participation. Under DT logic, this is transition management with a real but narrow niche, while the underlying labor-market corpse remains untouched.