How Governments Around the World Are Adopting AI, and Where Mexico Stands

Seven in ten civil servants already use AI in their daily work. The operational question for 2026 is no longer whether governments are adopting artificial intelligence, but who is converting that adoption into real public value, and what pieces Mexico has on the board.

In January 2026, the OECD published a study mapping 200 real-world cases of governments using artificial intelligence across 11 functions, from justice and public services to auditing and anti-corruption. The central finding: 57% of cases automate or personalize services, and 45% improve decision-making and forecasting.

That same month, the Public Sector AI Adoption Index 2026 revealed a data point that reframes the entire conversation. More than 70% of civil servants across 10 surveyed countries already use AI, yet only 18% believe their governments are leveraging that technology effectively. The gap is not in the tool; it is in the enabling architecture.

That gap between individual use and institutional value capture is precisely the window where the next decade of public competitiveness is decided. And Mexico is entering with concrete pieces on the table.

Who Is Leading: Three Models Worth Watching

Singapore leads global adoption with a score of 61 out of 100 on civil servant empowerment. Eighty-five percent of its public employees feel confident using AI, 73% have clarity on what they can and cannot use it for, and 58% know exactly who to consult when a question arises. The key is central coordination under the Smart Nation Agenda and the National AI Strategy, where GovTech provides shared platforms, approved tools, and practical guides for the entire government apparatus.

Estonia demonstrates the compounding effect of mature public digitization. Its virtual assistant Bürokratt connects a network of official chatbots that allows citizens to ask questions and complete administrative procedures through supervised AI workflows. Full public-service digitization in Estonia has saved more than 13 million work hours and raised electoral participation through online platforms.

The United Kingdom, alongside the United States and Brazil, falls into the "uneven adopters" category: visible progress, but with gaps in infrastructure and operational guidance. It is the mirror that shows what happens when adoption speed outpaces institutional architecture.

What Mexico Is Doing: Coatlicue, ATDT, and the Bet on Talent

Since October 2024, the National Development Plan 2025-2030 has included an explicit commitment to optimize public services and improve policy design with analytical tools, including AI. President Claudia Sheinbaum has placed science and technology as a cabinet priority.

Concrete pieces are already operational. The Digital Transformation and Telecommunications Agency (ATDT) was created to issue regulations and standards for the digital transformation of the state. The Ministry of Science, Humanities, Technology and Innovation, led by Dr. Rosaura Ruiz Gutiérrez, coordinates scientific policy.

In November 2025, the government launched the Public AI Training Center as part of the "Mexico, Country of Innovation" project, with the goal of becoming the largest public AI school in Latin America. Weeks later, the Coatlicue project was unveiled: a public supercomputer designed to be one of the most powerful in the region, and the most ambitious investment the Mexican state has made in AI infrastructure to date.

The Implementation Roadmap: Three Levers Already on the Table

Comparative experience suggests that value capture depends on three simultaneous levers. The first is a coordinating agency with technical and political authority, equivalent to Singapore's GovTech. The ATDT fulfills that role in Mexico, and its capacity to coordinate across government agencies will be decisive.

The second is sovereign computing infrastructure. Coatlicue positions Mexico with its own strategic assets for training models and processing sensitive data without relying exclusively on private hyperscalers.

The third is talent and operational guidance. The OECD's January 2026 report identified skills gaps as the primary bottleneck. The Public AI Training Center addresses that piece directly.

Risks and Mitigation: The Institutional Learning Curve

Comparative models reveal three typical risks. The first is fragmentation: each agency adopting different tools without common standards. The second is lack of clarity on which data can be shared, an area where Singapore excels with 79% clarity among its civil servants. The third is uneven adoption speed across levels of government, which can produce elite-level adoption with no improvement in citizen-facing services.

Mitigation, as demonstrated in successful cases, depends on central standards, shared platforms, mass training, and public metrics. The pieces Mexico already has in operation point in that direction.

Synthesis and 2026-2030 Outlook

The global picture shows that the differentiating factor is not access to technology; it is the speed and consistency with which a state converts individual adoption into institutional value capture. Mexico has relevant pieces in play: a coordinating agency, computing infrastructure under development, a public AI school, and an explicit presidential mandate. The 2026-2030 window is the phase where these pieces come together, and where impact on concrete public services will be measured.

To track the institutional progress of public AI in Mexico and the international benchmarks shaping the frontier, subscribe to the Scientika weekly analysis.

Sources

• OECD -- Building an AI-ready public workforce (January 2026)

• OECD -- Governing with Artificial Intelligence (200 cases)

• National Development Plan 2025-2030, Government of Mexico

• Digital Transformation and Telecommunications Agency (ATDT)

• Coatlicue / Public AI Training Center announcement, Office of the President

• Public Sector AI Adoption Index 2026, Public First

• Information Technology and Innovation Foundation (ITIF)

• Center for Data Innovation

• Wilson Center -- Mexico's Bet on Artificial Intelligence

• Baker Institute -- Mexico's Artificial Intelligence Future

• Public Sector Network -- Estonia AI Implementation Case Study

• Bulletin of Advanced Spanish -- AI and Beyond: Sheinbaum's Tech Ambitions

Frequently Asked Questions

Where does Mexico rank in global government AI adoption?

Mexico is not yet ranked among the top performers in global indices. Singapore leads with a score of 61 out of 100 on civil servant AI empowerment, while Mexico is in an earlier implementation phase, having only recently launched its coordinating agency (ATDT), the Coatlicue supercomputer project, and a national public AI training center.

What is the Coatlicue project?

Coatlicue is a Mexican government supercomputer initiative unveiled in late 2025. It is designed to be one of the most powerful public computing systems in Latin America, giving Mexico sovereign capacity to train AI models and process sensitive government data without full dependence on private cloud hyperscalers.

What is the ATDT and what role does it play in Mexico's AI strategy?

The ATDT (Digital Transformation and Telecommunications Agency) is the Mexican government body responsible for issuing regulations and standards for the digital transformation of the state. It functions as Mexico's central coordinating authority for public-sector AI adoption, analogous to Singapore's GovTech.