US vs Mexico: Neighbours With Very Different AI Risk

Same peak score, opposite deployment conditions

The headline number is deceptively similar: the US and Mexico both score 8.5/10 for their most AI-exposed occupation group, clerical support workers. That shared ceiling could suggest the two countries face comparable near-term AI labour disruption. They do not. The exposure score measures task susceptibility - whether a role's core tasks fall within current AI capability. It says nothing about whether employers have the capital, the technology access, or the economic incentive to actually deploy AI against those tasks. On all three dimensions, the US and Mexico diverge sharply.

US GDP per capita reached $90,026 in 2025 (World Bank Open Data). Mexico's stood at $13,889 in the same year. That 6.5x gap matters because enterprise AI deployment requires upfront investment: software subscriptions, infrastructure, integration work, and the organisational change needed to restructure workflows around AI tools. US employers in finance, insurance, legal services, and healthcare - sectors with large clerical headcounts - have both the capital and competitive pressure to act. Mexican employers operating with far lower margins and in a labour market where clerical workers earn $7,273 per year face a fundamentally different cost-benefit calculation.

Mexico's ILO ILOSTAT data (CC BY 4.0, 2025 data year) covers 59,026,480 workers across all major ISCO-08 occupation groups. The US Bureau of Labor Statistics OEWS May 2025 release covers 143,066,500 workers. Together that is 202.1 million workers whose AI exposure can be compared directly on the same ILO scoring framework.

Occupation-by-occupation: United States

The table below shows all major ISCO-08 occupation groups for the United States using BLS OEWS May 2025 data (2024 data year), accessed via ILO ILOSTAT (CC BY 4.0). Wages are BLS median annual figures in USD.

Occupation-by-occupation: Mexico

The table below shows all major ISCO-08 occupation groups for Mexico using ILO ILOSTAT data (CC BY 4.0, 2025 data year). Wages are OECD estimates converted to USD PPP for cross-country comparability.

Why the same peak score means different timelines

Both countries have clerical support workers at 8.5/10 - the highest score in either dataset. In the US that group covers 16.5 million workers earning $45,433 per year. In Mexico it covers 3.8 million workers earning $7,273 per year in USD PPP terms (OECD estimate). The economic incentive to automate is fundamentally different.

A US employer replacing a clerical worker saves roughly $45,433 in salary plus employer taxes and benefits - potentially $60,000 to $70,000 in total annual employer cost. At enterprise AI workflow tool pricing of $5,000 to $20,000 per year, the return on investment is clear and the payback period is short. This is why US clerical disruption is classified as imminent (1-3 years) in sectors like finance, insurance, and healthcare administration. A Mexican employer faces a very different equation: the $7,273 annual salary in USD PPP is close to or below the annual cost of many enterprise AI tools at market pricing, especially once integration, support, and retraining costs are factored in. The economic case for wholesale replacement is thin in the near term.

Mexico's risk velocity score of 2.2 out of 10 directly captures this deployment gap. The velocity score accounts for economic capacity, technology access, and sector maturity - not just task susceptibility. A score of 2.2 places Mexico firmly in the long-term disruption category: the underlying AI capability exists to automate many Mexican clerical tasks, but the deployment conditions (wages, capital, tooling access, employer incentives) will delay realisation by at least a decade at current trajectory.

The nearshoring dynamic: USMCA and plant operator jobs

There is a striking irony in the US-Mexico AI labour comparison. US companies are actively shifting manufacturing to Mexico under the United States-Mexico-Canada Agreement (USMCA), partly in response to AI-driven supply chain restructuring and partly to reduce exposure to geopolitical risk in Asian manufacturing. This nearshoring trend is adding plant and machine operator jobs in Mexico - ILO ILOSTAT data (2025) shows 6,239,100 plant and machine operators in Mexico, a group scoring 3.0/10 on AI exposure.

Plant and machine operators score 3.0/10 because their work involves physical operation of machinery in variable industrial environments - tasks that are substantially harder for current AI and robotics systems to automate than clerical or data-processing work. The same AI-driven disruption that is pushing US employers to move manufacturing to cheaper labour markets is, perversely, creating more low-AI-risk jobs in Mexico. The workers taking those roles benefit from a temporary buffer: the manufacturing tasks being nearshored are precisely the ones that remain relatively safe from AI disruption for the next decade.

This does not mean Mexican manufacturing workers are permanently insulated. Robotics investment in Mexican maquiladora zones has increased year-on-year since 2020 (International Federation of Robotics, 2024 World Robotics Report). But the relevant comparison is: automation of a $6,977/year plant operator in Mexico (USD PPP) versus a $45,844/year equivalent in the US. The US automation case is compelling now. The Mexican case is compelling only once robotics hardware costs fall further or wages rise substantially.

Mexico's professionals: high exposure, low wages, slow adoption

Mexico's professionals group covers 6,341,500 workers scoring 6.5/10 on AI exposure (ILO ILOSTAT 2025). These are scientists, engineers, healthcare professionals, lawyers, and business analysts - the same occupational categories that score identically in the US. The difference is wages: Mexican professionals earn $10,816 per year in USD PPP terms (OECD estimate), compared to $82,032 for US professionals (BLS OEWS May 2025). That is a 7.6x gap for workers facing the same underlying AI exposure score.

High AI exposure at low wages creates a different adoption dynamic. In the US, professional AI tools - coding assistants, document analysis software, legal research AI, diagnostic decision support - are adopted because they produce measurable productivity gains that justify their cost against high professional wages. A US law firm billing $400/hour per attorney has strong incentive to deploy document review AI. A Mexican law firm operating with attorneys earning $10,816/year (USD PPP) has far weaker incentive to invest in the same tools. The AI capability exists and is technically applicable. The economic driver is not yet present at scale.

Mexico's HDI of 0.789 (UNDP Human Development Report 2025, 2023 data year, rank 81) reflects a workforce with growing but still developing access to tertiary education and digital skills. The US HDI of 0.938 (rank 20) reflects a workforce with deep digital literacy across the professional class. As AI tools become cheaper and more accessible - and as Mexican professionals increasingly access global AI platforms - the adoption gap will narrow. But in 2026 it remains wide.

Economy comparison: US vs Mexico side by side

The table below puts both countries' economic indicators side by side using World Bank Open Data (CC BY 4.0) and UNDP Human Development Report 2025 (HDR 2025, 2023 data year, licence CC BY 3.0 IGO).

Mexico's unemployment rate of 2.67% (World Bank 2025) is notably low - lower than most comparable middle-income economies. This reflects a tight labour market driven in part by nearshoring demand, demographic factors, and informal economy absorption of workers who do not appear in official unemployment counts. A low unemployment rate reduces the immediate pressure on employers to automate even where AI could technically replace tasks: when labour is scarce and cheap, the case for capital-intensive automation weakens further. For Mexican workers in the near term, this is protective.

The safest jobs from AI in both countries

At the bottom of the AI exposure spectrum, both countries converge. Elementary occupations score 2.0/10 in both the US and Mexico. But the scale difference is striking. Mexico has 12,130,700 elementary workers (ILO ILOSTAT 2025) - more than 3x the US equivalent of 3,846,500. Elementary occupations represent 20.5% of Mexico's entire workforce versus 2.7% in the US. Mexico's large informal and semi-formal employment base absorbs many workers in cleaning, manual loading, packaging, and basic agricultural support tasks that resist current AI and robotics automation.

Mexico has 8,250,000 craft and trades workers (ILO ILOSTAT 2025) scoring 2.5/10 - a substantial group representing 14.0% of the workforce. Construction, electrical work, plumbing, and mechanical repair are all trades that require physical dexterity in variable environments, making them resistant to current AI and robotics approaches. The USMCA-driven infrastructure investment flowing into northern Mexico has sustained strong demand for these workers. For the foreseeable future, trades workers in Mexico face lower AI disruption risk than almost any other group in either country's dataset.

Shared supply chains, separate AI trajectories

The US and Mexico are deeply integrated economically. The US is Mexico's largest trading partner; Mexico is the second-largest export destination for US goods (US Census Bureau, 2024). Under USMCA, automotive, electronics, aerospace, and agriculture supply chains flow continuously across the border. AI automation in a US factory can directly affect Mexican workers in the same production chain - a US company that automates parts procurement and scheduling in its Ohio headquarters may reduce orders from its Juarez supplier, affecting Mexican plant operators even if no automation tool directly touches them.

This supply-chain linkage is the geopolitical AI risk that the occupation-level scores do not fully capture. A Mexican factory worker scoring 3.0/10 for direct AI exposure faces indirect risk from upstream AI adoption by their US supply chain partners. The ILO ILOSTAT data measures direct task susceptibility - it does not model demand-side shocks propagated through cross-border production networks. For Mexican workers in export-oriented manufacturing sectors, this indirect channel is the more relevant near-term concern than direct AI deployment by their own employers.

The reverse dynamic is also present. AI-driven quality control systems and just-in-time inventory optimisation tools deployed by US manufacturers can make Mexican production partners more efficient and competitive, protecting their positions in US supply chains. AI is not uniformly destructive to Mexican manufacturing employment - it can enhance the competitiveness of nearshoring arrangements that sustain those jobs.

What this means for workers in both countries

For US clerical workers, the situation is urgent. The 16.5 million US workers scoring 8.5/10 are in roles where AI deployment is already underway in 2026. Document processing, customer correspondence, data entry, and scheduling tasks are being automated in financial services, healthcare administration, and legal support sectors. The realistic displacement timeline for the highest-exposure sub-groups is 1 to 3 years in high-investment sectors. Reskilling toward roles that combine AI proficiency with human judgment - project management, client-facing coordination, technical support - is the most viable path for workers in these roles.

For Mexican clerical workers, the picture is more nuanced. The 3.8 million Mexican clerical workers scoring 8.5/10 earn $7,273 per year in USD PPP terms. The economic case for their employers to automate is weak in the near term. But the technology is not standing still, and AI tool costs are falling. Workers in Mexican business-process outsourcing (BPO) and call centre sectors - which serve US and Canadian clients who are simultaneously deploying AI - face a more compressed timeline than the broader Mexican clerical average suggests. BPO clients will not wait for Mexican wage conditions to change before automating if they can achieve the same outcome through AI tools directed at the US-side workflow.

For Mexican plant and machine operators benefiting from nearshoring, the window of relative safety is probably one decade - enough time for meaningful career development but not a permanent protection. The International Federation of Robotics (2024) projects continued cost reduction in industrial robotics at 5-10% per year. At current trajectories, the economics of automating Mexican manufacturing tasks will shift meaningfully by the mid-2030s. Workers entering manufacturing roles in Mexico today should treat their current AI buffer as a medium-term opportunity to accumulate skills and earnings, not a permanent guarantee.