Cuba AI Job Risk 2026: Which Occupations Are Most at Risk?

The most AI-exposed occupations in Cuba

Cuba's occupation data comes from the Oficina Nacional de Estadisticas e Informacion (ONEI) labour statistics 2023-2024. International data access on Cuba is more limited than for most countries in this dataset due to restricted reporting under the US embargo and constrained internet infrastructure. Employment figures are estimates derived from ONEI published aggregates and ILO ILOSTAT Cuba country data. The dataset covers approximately 4.84 million formal sector workers - note that Cuba's formal sector share is unusually high relative to GDP given the state-dominated employment structure. Cuba's data is tracked at ISCO-08 major group level, not the finer sub-major detail available for some other countries.

Clerical support workers score 8.5/10 - the peak in Cuba's economy, covering 8.0% of the formal workforce. State administrative workers concentrated in Havana's ministries, provincial offices, and state enterprises (CUPET oil, ETECSA telecoms, Cubana de Aviacion) represent the core of Cuba's clerical workforce. These roles handle internal document processing, record maintenance, supply chain administration, and government reporting - tasks that AI tools automate effectively when connectivity and computing infrastructure are available. In Cuba's case, the connectivity constraint is the binding limitation, not the task automation capability.

Professionals at 15.5% of the workforce and 6.5/10 exposure represent Cuba's most strategically significant at-risk group, and its second-largest occupation group overall. This group includes Cuba's 80,000+ domestic doctors plus nurses, dentists, allied health workers, engineers, and scientists. AI diagnostic tools (radiology AI, pathology AI, clinical decision support) are deployed globally and have documented performance advantages in specific diagnostic categories. Cuban health workers currently lack the connectivity infrastructure to access most commercial AI diagnostic platforms. However, BioCubaFarma's domestic AI research program and Cuba's participation in WHO digital health initiatives suggest this constraint is temporary rather than permanent. Havana's small but growing software export sector (Desoft, XETID) also falls within this professional tier.

The embargo constraint - why AI deployment lags exposure

Cuba's AI exposure score of 4.24/10 reflects the task composition of its workforce - what those occupations would be exposed to if AI tools were freely deployable. The actual pace of AI adoption in Cuba is substantially lower than this score suggests, and the reason is structural rather than attitudinal - this is exactly what Cuba's 0.5/10 risk velocity score captures.

The US embargo (known in Cuba as "el bloqueo") restricts Cuba's access to most US-origin technology, including cloud computing services, software licences, and semiconductors. Google Cloud, AWS, Microsoft Azure - the three dominant AI cloud platforms - are all US companies with embargo compliance obligations that restrict Cuban institutional access. Cuban organisations cannot legally subscribe to most commercial AI software-as-a-service products. Hardware procurement is constrained by US export controls on advanced chips. Home internet penetration was approximately 1.4 million connections in 2023 (ONEI), covering roughly 12% of the 11.2 million population - one of the lowest rates in the Western Hemisphere.

The practical consequence is a deployment gap: Cuba's professional workforce has the educational base to use AI tools (literacy 99.7%, universal university access, strong mathematics and science curriculum) but lacks the infrastructure to access them commercially. This gap creates a bifurcated risk profile. In the short term (2025-2028), AI disruption in Cuba will be limited to state institutions that procure AI tools through non-US supply chains - primarily Chinese providers (Huawei AI, Baidu AI platform) and domestically developed tools at UCLV (Universidad Central de Las Villas) and ICID (Instituto de Ciencia Animal).

BioCubaFarma's AI research program is the most significant domestic AI development effort. The conglomerate - which encompasses 32 enterprises, 21 research centres, and over 20,000 workers - developed Abdala and Soberana COVID-19 vaccines using computational protein modelling tools. Its ongoing oncology and neurological research programs incorporate AI-assisted drug discovery. For BioCubaFarma's research scientists and bioinformaticians, AI augmentation is already occurring within a domestically controlled technology stack.

The safest jobs from AI in Cuba

Cuba's agricultural and physical economy - representing approximately 30% of the formal workforce - scores below 3.0/10 AI exposure across all groups.

Elementary occupations are Cuba's largest single occupation group at 17.2% of the formal workforce, scoring 2.0/10. Agricultural workers at 10.4% of the formal workforce score 3.0/10. Cuba's agricultural sector encompasses sugar cane (historically the dominant crop, now producing approximately 1.0 million tonnes annually - down from 8 million tonnes at peak), tobacco (Habanos SA produces Cohiba, Montecristo, and Romeo y Julieta for export), citrus, and food production under the UBPC cooperative system. Cuba's land reform history - collectivised farming under the 1959 revolution, then partial decollectivisation to usufruct cooperatives (UBPCs) from 1993 - means Cuba's agricultural workforce operates in a distinct institutional context from any other country in this dataset.

Service and sales workers at 14.8% of the workforce score 3.5/10. Tourism - Cuba's second-largest legal foreign exchange earner after medical exports - employs a substantial share of this group: hotel workers, guides, taxi drivers, and paladares (private restaurants legalised in 1993). The tourism sector was decimated by COVID-19 (1.0 million visitors 2021 vs 4.3 million 2019) and has only partially recovered. Service workers in Cuba's tourism economy face minimal AI disruption risk in the near term given the infrastructure constraints on AI-enabled service automation.

What this means for you

Cuba's 4.24/10 average places it mid-range in the Americas batch - below the US (5.06/10), above Brazil (4.13/10) and Mexico (3.83/10) in exposure terms - but the risk velocity of 0.5/10 is among the lowest in the entire dataset, reflecting the deployment constraints described above. For workers in Cuba today, the AI disruption timeline is far slower than the exposure score alone would suggest.

If you are a professional in Cuba's healthcare system - a doctor, nurse, or allied health worker - your occupation sits within a group scoring 6.5/10 structurally but is currently protected by infrastructure constraints. The relevant planning horizon is not 2026 but 2030-2035, when Cuba's internet penetration and AI tool access will likely have expanded under whatever geopolitical conditions prevail. Cuban medical professionals deployed abroad - the 30,000+ doctors in Venezuela, Angola, Brazil, and other partner countries - face AI disruption at the pace of their host country, not Cuba's domestic pace. A Cuban doctor working in a Brazilian hospital is exposed to the same AI diagnostic tools as their Brazilian colleagues.

Cuba's data does not include a recovery resilience score, unlike most other countries in this dataset - the underlying World Bank inputs that feed that metric are not reliably available for Cuba. What can be said qualitatively: Cuba's universal education system and strong institutional training capacity (polyclinics, medical schools, UCLV, UCI - University of Information Sciences) provide retraining infrastructure that many lower-income countries lack. On the other hand, the private sector - which provides most retraining pathways in market economies - is small. Cuentapropistas (self-employed workers, now legally expanded since 2021 to include private businesses with up to 100 employees) represent Cuba's growing private sector, but they remain a small share of total employment. Workers displaced from state roles have limited market-sector alternatives in the near term.