Mexico is expanding AI training
Public training is already operating at meaningful scale.
In January 2026, 10,000 students began activities through Mexico's Public Artificial Intelligence Training Center. In May, the Digital Transformation and Telecommunications Agency (ATDT) opened a second national call with 30,000 places for free five-month programs covering artificial intelligence, cybersecurity, cloud, data analytics, and networks and infrastructure.
ATDT's official agenda also records 5,771 students certified in emerging technologies and AI during 2026. That figure should not be confused with available places, registrations, or participants who started a program. Each metric represents a different stage of the training pipeline.
The expansion matters because it increases the pool of people exposed to relevant technologies and skills. For an industrial company, however, the important question begins after the course:
Where can that capability actually be used inside the operation?
Demand is expanding far beyond AI specialists
PwC found that Mexican job postings requiring artificial intelligence skills increased from approximately 33,000 in 2024 to 68,000 in 2025.
The more useful finding is the composition of that growth. PwC distinguishes between:
- AI user roles, requiring AI literacy and applied capabilities;
- AI developer roles, requiring advanced technical skills to build, train, or deploy systems.
Approximately 32,300 additional positions in 2025 came from growth in AI user roles, compared with about 2,200 additional developer roles.
That distinction matters in manufacturing. A plant does not need every employee to become a data scientist. Engineering may need to evaluate AI-generated outputs. Quality may need documented validation criteria. Procurement may use AI-assisted research without delegating supplier decisions. Maintenance teams may need access to structured histories, while technical specialists design the integrations that support those workflows.
The required skill level follows the work.
Industrial employers already face a broader talent constraint
Talent shortages are not distributed evenly across Mexico's economy.
ManpowerGroup's 2026 Talent Shortage Survey, conducted with more than 1,000 Mexican employers, found that 67% reported difficulty filling positions. The figure rises to 71% in manufacturing and 81% in automotive. AI model development and application are among the positions identified as difficult to fill.
This does not mean every unfilled manufacturing job is an artificial intelligence position. It does show that manufacturers are competing for talent in a labor market where technological, operational, and human capabilities are changing simultaneously.
A strategy based only on hiring specialists or sending teams to generic AI courses is unlikely to resolve that problem. The company first needs to define which capability it is trying to build.
Learning to use AI is not the same as organizational capability
Microsoft's 2026 Mexico findings make that distinction particularly visible.
Among Mexican AI users surveyed, 67% said they can now perform work they could not do a year earlier. Yet only 28% reported clear leadership alignment around translating those capabilities into new ways of working.
Individuals can advance faster than their organizations.
An employee may learn to use a generative model while still working with:
- fragmented information;
- contradictory documents;
- undefined processes;
- unclear ownership;
- security restrictions;
- no validation criteria;
- disconnected systems;
- or no meaningful measure of success.
Under those conditions, an individual skill exists. Organizational capability does not.
Using AI is also different from integrating it
Mexico AI Pulse, from Endeavor, provides another view of this gap.
Within a sample of 163 Mexican organizations, 55% reported using AI. Only 31% reported broad integration across key processes, while 6% reported capturing significant economic value from it.
These percentages are not representative of every company in Mexico. They describe the organizations included in that study.
Their value is analytical: they illustrate an important distinction.
Using a tool, integrating it into a process, and building a repeatable organizational capability are not the same thing.
Talent planning should begin with the workflow
An industrial AI talent strategy can begin with the work rather than the course catalog.
Before deciding what someone needs to learn, a company can ask:
1. Which workflow are we trying to improve?
Define the task, frequency, inputs, output, and expected result.
2. What information does the workflow require?
Identify the data, documents, histories, specifications, and sources required to execute it correctly.
3. What role should AI play?
AI may assist, classify, analyze, generate, recommend, or automate. These functions require different levels of autonomy and oversight.
4. Who uses, reviews, and approves the output?
The person using a system does not necessarily have to approve its output. Usage, validation, technical support, and accountability may belong to different roles.
5. How will the company know whether it works?
Time saved is not enough. Accuracy, quality, errors, rework, escalations, security, and operational usefulness may matter as well.
6. Which skill is actually missing?
Only then can the company determine whether the gap requires AI literacy, process expertise, data analysis, technical integration, security, governance, or advanced development capability.
From training to industrial capability
The transfer can be described as a chain:
Training → Role → Workflow → Information → Tool → Validation → Accountability → Measurement → Organizational capability
A trained person represents potential.
A process where that person can apply the capability with adequate information, explicit responsibilities, and validation criteria represents an implementation.
When the organization can repeat, measure, govern, and improve that process without depending entirely on one person who “knows the tool,” organizational capability begins to emerge.
That is the point where training starts becoming operational infrastructure.
The skills problem is not unique to Mexico
OECD research has identified skills shortages as a meaningful barrier to AI adoption. In international surveys of employers in manufacturing and finance, around 40% reported that a lack of relevant skills represented an obstacle to adoption.
That percentage is not specific to Mexico. Its relevance is contextual: training, hiring, and developing people able to integrate artificial intelligence into real work is a broader adoption problem across economies.
Mexico combines that challenge with a large manufacturing base, still-early adoption, and growing pressure to compete in higher-value industrial activity.
What these data do not prove
ATDT figures describe places, participants, and certifications in public programs. They do not show how many participants will work in manufacturing or how many AI projects will reach production.
PwC job-posting data measures employer demand for skills. It does not show that every position was filled or that those hires resulted in successful implementation.
ManpowerGroup data measures hiring difficulty by sector. It does not mean that 71% of manufacturers specifically lack AI specialists.
Mexico AI Pulse uses a sample of 163 organizations and should not be extrapolated to the entire Mexican business population.
And the 4.8% AI-adoption figure for Mexican manufacturing refers to manufacturing businesses with more than ten employees in the underlying study. It does not measure the depth or quality of each implementation.
Keeping these populations separate prevents training, labor demand, tool usage, and operational capability from becoming one misleading statistic.
Platonica Perspective
Growth in artificial intelligence talent is a positive signal for Mexico.
But industrial advantage does not emerge automatically from more courses, certifications, or software licenses.
It emerges when an organization can clearly answer:
- where the capability will be used;
- which problem it is meant to solve;
- what information supports it;
- who uses and who validates;
- what risk is acceptable;
- how performance will be measured;
- and who owns improvement of the process.
Training creates potential. A company builds capability when it knows where to apply that knowledge, what information supports it, who is accountable for the result, and how it will prove that the workflow works.
The next stage of adoption is therefore not simply teaching more people how to use artificial intelligence.
It is building organizations that can convert knowledge into repeatable, governable, and measurable work.
Talent is one layer of adoption, not the whole adoption problem; the broader industrial context is developed in artificial intelligence in Mexican manufacturing.
Frequently asked questions
How many people is Mexico training in artificial intelligence?
ATDT reported 10,000 students beginning the first cohort of the Public AI Training Center and opened 30,000 places for its second call. Its official agenda also records 5,771 students certified in emerging technologies and AI during 2026. These are different metrics and should not be added together as though they describe one population.
Is employer demand for AI skills increasing in Mexico?
Yes. PwC reported approximately 68,000 job postings requiring AI skills in 2025, compared with about 33,000 in 2024. Most of the growth came from AI user roles.
Does every company need to hire AI developers?
No. Many organizations need more people capable of using, supervising, and validating AI within their functions than people capable of building models. Technical depth should follow the workflow and its risk.
Does training employees mean a company has adopted AI?
No. Training develops individual skills. Adoption also requires use cases, information, tools, integration, responsibilities, controls, and evidence that AI is being used within real processes.
Why does this matter particularly for manufacturing?
Because 71% of manufacturing employers surveyed by ManpowerGroup report difficulty finding talent while AI adoption in Mexican manufacturing remains early. The challenge combines skill availability with the ability to integrate those skills into real operations.
Sources and documentation
- Mexico Digital Transformation and Telecommunications Agency (ATDT). “México pone en marcha el mayor Centro de Formación en Inteligencia Artificial de Latinoamérica.” January 26, 2026. Official source.
- ATDT. “Centro Público de Formación en IA abre 30 mil lugares para segunda generación.” May 26, 2026. Official source.
- ATDT. Agenda 2030, Artificial Intelligence. 2026 certification data. Official source.
- PwC Mexico. “AI Jobs Barometer 2026, Mexico edition.” June 2026. Source.
- ManpowerGroup Mexico. “Talent Shortage 2026, Mexico.” February 2026. Source.
- Microsoft. “La IA ya transformó al talento mexicano; ahora es el turno de las empresas.” August 31, 2026. Source.
- Endeavor Mexico. “Mexico AI Pulse: de adopción ad-hoc a ejecución organizacional.” August 2026. Source.
- Mexico Ministry of Economy. “La adopción de IA en la manufactura mexicana se asocia con retornos medibles.” June 5, 2026. Official source.
- OECD. “Skill needs and policies in the age of artificial intelligence.” International context on skills and AI adoption. Source.