The McKinsey Global Institute mapped in 2026 what awaits work in fifteen Latin American countries. In these economies, 57% of working hours are potentially automatable with existing technologies, of which 18% by physical robots. Where wealthy economies encounter automation cognitively first, Latin America encounters the machine first in the workshop, in the field, and in the warehouse. The question of cost and benefit-sharing thus arises before the question of the robots’ arrival.
The essentials
- In 15 Latin American countries, 57% of working hours are potentially automatable with existing technologies, of which 18% by physical robots (McKinsey Global Institute, 2026).
- Robotization strikes first at the most physical economies: in Bolivia and Peru, roughly 40% of automatable hours involve physical tasks, concentrating risk on the least-skilled workers.
- Gilbert Cette and Jacques Barthélémy showed that technological change requires simultaneous adaptation of protections, training, and social dialogue: transferring machines without reforming institutions means transferring risk to workers.
- Adoption and its effects depend on several economic, institutional, organizational, and skills-related factors.
- MGI presents two scenarios for adoption by 2030 and three regional profiles of automation based on institutional configurations and economic policy choices.
The physical geography of Latin American labor
Latin America works with its hands. This observation, banal on its face, has precise implications when crossed with the current capacities of robots.
In the most advanced economies—the United States, Western Europe, and Japan—the shift toward service work means automation threatens routine cognitive tasks first: data entry, administrative processing, customer service. Generative AI models arrive in offices. In Latin America, the sequence is different. Agriculture, mining, construction, and manufacturing concentrate a share of work that wealthy economies largely mechanized beginning in the 1980s. The robots arriving today replace pickers, welders, and material handlers.
McKinsey documents this specificity with precision. In the region’s most physically intensive economies—Bolivia, Peru, and Ecuador—physical tasks represent roughly 40% of automatable hours. This geographic concentration of risk touches first the workers whose alternatives are most limited. Moving from a material handler position to a robotic maintenance technician position requires training that nothing guarantees today.
The contrast with Africa is instructive. As a recent article on African technological choices showed, some continents still have time to design their institutions before technology overtakes them. Latin America is already in the adoption window: robots are entering, institutional frameworks are lagging behind.
Lessons from labor economics on these transitions
Economist Gilbert Cette, together with Jacques Barthélémy, built over several decades a thesis on the relationship between technological change and labor institutions. Their work Labor and Technological Change: From the Factory Civilization to the Digital One documents how each wave of automation produces winners and losers based on the institutions that frame adoption, not on machine logic alone.
The thesis is simple in principle, demanding in its implications. A technology that increases productivity can translate, depending on institutional configurations, into wage increases, job losses, or a combination of both. Among the factors influencing the outcome are mechanisms for sharing gains between capital owners and wage earners, training schemes allowing displaced workers to reposition themselves, and social dialogue capable of negotiating transitions rather than enduring them.
This framework applies directly to Latin America. The 18% of hours automatable by robots will not produce the same effects in Mexico, where integration into North American supply chains creates incentives to modernize and train, and in Bolivia, where extractive industry firms often operate without structured branch agreements. The machine is identical. The institution is different. The result will be too.
The competing reading: the competitiveness argument
A different reading exists, advanced notably by economists close to the market-liberal tradition. It deserves to be taken seriously, not to be dismissed, but because it illuminates a real tension.
According to this reading, the main risk for Latin America lies in robotization that is too slow. Companies that do not adopt available technologies lose competitiveness against Asian competitors already robotized. The Mexican automotive sector, for example, faces South Korean and Japanese producers whose factories display robotic densities several times higher. Slowing adoption to preserve jobs in the short term would sacrifice long-term competitiveness, and therefore future jobs.
This tension is documented by data from the International Federation of Robotics, which shows that robotic density—the number of robots per 10,000 industrial employees—remains very low in Latin America compared to Asia or Europe. Mexico, the region’s most industrialized country, remains far behind South Korea or Singapore. The competitiveness argument says: this lag is itself a risk.
The tension between these two readings is not resolvable in principle. It is resolved by data and conditions. Rapid robotization can disrupt employment, as several studies on American regions exposed to Chinese competition in the 2000s documented. According to certain documented industrial trajectories, institutional conditions accompanying technological adoption influence results in productivity and compensation. The debate is therefore not about adoption speed, but about what accompanies it.
The three levers that decide the outcome
Access to financing is one possible brake on adoption, among several constraints. Latin American small and medium-sized enterprises, which represent a majority of regional employment, often lack access to the financing needed to acquire robotic equipment. Large exporting companies adopt quickly. Highly automated production units thus coexist with a fabric of small structures that remain labor-intensive by financial constraint. This results in productivity inequality that translates into wage inequality.
Public credit lines conditioned on internal training plans could correct this asymmetry. Access to robotic financing would be linked to a company’s commitment to maintain employment and train workers whose tasks are modified. Analogous mechanisms existed in several European countries during waves of industrial modernization in the 1990s. They accompanied transformation by limiting its destructive effects for workers in place.
Technical training is the second lever, and probably the hardest to activate quickly. Training a robotic maintenance technician takes time. Latin American vocational education systems, with few exceptions like Brazil’s SENAI, are underfunded and poorly aligned with business needs. Technological change increases demand for skills, and regional training systems present gaps. This mismatch constitutes a risk for affected workers.
Branch agreements on benefit-sharing constitute the third lever. When a robotic investment increases a factory’s productivity, this gain is distributed between shareholders and employees based on negotiated rules or, in their absence, based on dynamics that tend to favor capital. This dynamic is institutional and can therefore be modified.
Sectoral agreements that index part of the wage bill to productivity gains allow robotization to produce both margins for companies and real wages for workers.
Current signals and the trajectory toward 2035
Two trajectories lie open, and data from coming years will allow us to begin distinguishing them.
In the first scenario, the most exposed physical sectors benefit from expanded access to public equipment credit and branch agreements generalize in sectors with high automation. The 18% constitute a technical potential that can create value only according to the rhythm and conditions of adoption. Workers whose tasks disappear access training financed partly by gains generated. The effects of robotization on median incomes and trajectories of the least qualified are not uniformly positive and can be unfavorable. This scenario corresponds to institutional configurations that have worked elsewhere, in different contexts but comparable fundamentally.
In the second scenario, equipment credit remains concentrated on large exporting companies. MGI’s more gradual scenario concerns slower technological diffusion, without hypothesis of equipment credit reserved for large exporters. Small structures face financial, technological, organizational, skills, and infrastructure constraints. A growing number of workers displaced by robotized large companies reposition themselves toward less productive informal jobs, lacking access to technical training. The 57% indicate a theoretical technical potential, not a projected wage pressure.
MGI’s more gradual scenario reduces estimated potential economic value; unequal adoption by firm size could accentuate inequalities.
Two signals will allow us to know which direction the region is moving. The first is the evolution of median wages in the most robotized physical sectors, compared to that of sectors little touched by automation. If wages converge, the institution functions. If they diverge, capture is at work. The second signal is the rate of access to robotic equipment credit for companies with fewer than fifty employees in physical sectors: this is the most direct indicator of diffusion outside large companies, therefore of the dual or inclusive dynamic of the transition.
Agriculture is a particularly revealing area of observation. Agricultural robots are already transforming large-scale corn cultivation, as shown by adoption documented in other contexts. In Latin America, where agriculture still occupies an important fraction of the workforce in the most physically intensive economies, this transition can expose many rural workers. Credit, training, and transitional support can influence the social balance sheet, among other economic and institutional factors.
The decision that belongs to institutions
Robotization of Latin America is not a future scenario: it is an ongoing process. The technologies are available. The most capitalized companies are already adopting them. What remains open is the institutional response.
Governments in the region face a window. It is not closed, but it is narrowing as investment decisions accumulate without a framework. Establishing conditional credit lines, financing vocational technical training, creating conditions for branch agreements on benefit-sharing: these choices are known, documented, and technically feasible. They require political will and labor institutions sufficiently strong to enforce what is negotiated.
The regional trajectory will depend jointly on the adoption rate of robots and on the institutional, economic, and training capacities of states and social partners to adapt the effects of this adoption. McKinsey Global Institute data indicate that Latin America has a margin of choice over what robots will produce for its workers. This choice is being made now.
Sources
- McKinsey Global Institute, Agents, robots, and us: How AI reshapes work and skills in Latin America, 2026. Link
- Jacques Barthélémy and Gilbert Cette, Labor and Technological Change: From the Factory Civilization to the Digital One. Catalog reference
- International Federation of Robotics, World Robotics 2026 (no link: IFR annual report, available at ifr.org)
