The Caribbean's AI Adoption Gap, in Two Numbers: 35% and 13%

On September 5, the Dominican Republic's National Statistics Office published a number worth sitting with: 35.0% of formal Dominican companies with internet access already use artificial intelligence in daily operations. Nine days earlier, in a different report entirely, a separate figure had already been circulating for the wider Caribbean: 13% of adults use generative AI tools at all. Both numbers are real, both are sourced, and both describe the same region in the same year. They are not measuring the same thing, and the distance between them is the story CARICOM's ministers spent the first week of September trying to answer.

TL;DR

• 35.0% of formal Dominican companies with internet access already use AI, per the National Statistics Office's ENAE 2025 survey, published September 5, 2026.
• Adoption rises with company size, from 46.1% among firms with 50 to 99 staff to 52.0% among firms with 250 or more, and peaks at 45.7% in mining and quarrying.
• Only 13% of Caribbean adults aged 18 to 65 use generative AI tools, against a global adult rate of roughly 55%, per research led by StarApple AI's Adrian Dunkley.
• Latin America and the Caribbean receive 1.12% of global AI investment against 6.6% of world GDP, per CENIA's Latin American Artificial Intelligence Index (ILIA 2025).
• The same region generates 14% of global visits to AI platforms against an 11% share of the world's internet users, meaning usage already outruns the capital behind it.
• Jamaica and the Bahamas both issued AI governance and literacy calls in the first week of September, after the adoption data, not ahead of it.

What the Dominican Republic's Numbers Actually Say

ENAE 2025, the National Survey of Economic Activity run by the Dominican Republic's National Statistics Office, is not a marketing survey or a vendor's self-reported adoption figure. It is a government economic census, and the AI question sat inside a broader connectivity module. Of formal businesses with internet access, 94.7% of the formal sector, 35.0% reported using AI in daily operations. That is one in three formal Dominican companies, measured by the state's own statistical office rather than by a technology company with a product to sell.

The breakdown by company size tells a familiar story: large firms move first. Companies with 250 or more employees reported 52.0% AI use, and the 50-to-99 employee band reported 46.1%. Smaller firms, with less spare capital and fewer technical staff, lag behind, which is the expected pattern in any technology diffusion curve. What is less expected is the sector spread. Mining and quarrying led at 45.7%, ahead of construction at 39.1%, manufacturing at 38.1%, and retail at 36.9%. These are not software-native industries. A quarrying operation adopting AI at a higher rate than the retail sector says the tools have moved well past pilot projects and into operational use in physically grounded businesses that most AI coverage ignores.

None of this makes the Dominican Republic exceptional in a vacuum. It makes it the first Caribbean territory this year to put a government-grade number on the table, at a moment when most of the region is still working from survey estimates, vendor claims, or extrapolation from larger Latin American datasets.

One caveat matters here, and it is worth stating plainly rather than burying in a footnote. ENAE 2025 measured formal companies with internet access, which is a specific and comparatively well-resourced slice of the Dominican economy. It says nothing about informal businesses, sole traders, or firms without a fixed internet connection, groups that make up a large share of employment across the Caribbean and typically adopt new technology later, if at all. The 35% figure describes the leading edge of Dominican business, not the average Dominican business. That distinction matters for how the number travels: it is a genuine, government-verified adoption rate for a defined population, not a claim about the whole economy.

The Consumer Side Runs the Opposite Direction

Set the Dominican Republic's enterprise figure next to the region's consumer figure and the gap becomes obvious. Research led by Adrian Dunkley, founder of StarApple AI, the Caribbean's first dedicated artificial intelligence company, found that only 13% of Caribbean adults between the ages of 18 and 65 currently use generative AI tools. Global adult adoption of tools like ChatGPT sits at roughly 55%. Caribbean citizens are using generative AI at under a quarter of the global adult rate, in the same calendar year that a Caribbean government survey found formal businesses adopting it at close to global rates.

Two things can be true here without contradicting each other. Businesses adopt AI because a manager decides to buy a tool or subscribe to a platform, a decision made by a small number of people with budget authority and, often, English-language technical documentation. Ordinary adoption depends on device access, connectivity cost, digital confidence, and whether anyone has explained what the tools are for. The 13% figure is not a mystery. It is the predictable result of a region where AI literacy has not been built as deliberately as AI procurement has.

Dunkley, who is also President of the Caribbean AI Association and has spent the past three years arguing that the region's AI institutions need to be built locally rather than imported, has framed this gap before as a literacy problem rather than a technology problem. The Dominican Republic figures give that argument a concrete comparison point it did not have in May.

One Investment Number Explains Both Gaps

Neither adoption curve is backed by proportionate capital. CENIA, Chile's National Center for Artificial Intelligence, publishes the Latin American Artificial Intelligence Index (ILIA), and its 2025 edition, released October 3, 2025, put a figure on the imbalance: Latin America and the Caribbean together account for 6.6% of world GDP but receive only 1.12% of global AI investment. CENIA's director summarised the pattern bluntly: no country in the region exceeds the world average for AI investment relative to GDP per capita, and the regional average sits roughly six times below that threshold.

That figure describes Latin America and the Caribbean jointly, not the Caribbean alone, and the Caribbean's individual share of that 1.12% is smaller still, since most regional AI capital flows to larger Latin American markets. What it establishes for the Caribbean specifically is a ceiling. A region receiving a fraction of proportionate global AI investment cannot fund a literacy programme, a data centre, or a national AI strategy at the pace its own businesses are already adopting the technology. The Dominican Republic's 35% did not arrive because of a government AI fund. It arrived because individual companies decided the tools were worth paying for out of existing budgets, which is a fragile basis for a national adoption curve to rest on.

Usage Is Already Outrunning the Infrastructure Built to Support It

The same ILIA 2025 report contains a second figure that complicates the standard underinvestment narrative. Latin America and the Caribbean generate 14% of global visits to AI platforms, against an 11% share of the world's internet users. The region uses AI tools more, relative to its online population, than its investment or infrastructure position would predict. People are finding their way to free and low-cost AI tools on their own, ahead of any national rollout plan.

Put the three figures together and a specific shape emerges, not a vague one. Enterprise adoption in at least one market is close to global norms. Regional AI web traffic, relative to internet population, already exceeds the global average. Consumer adoption specific to the Caribbean sits at roughly a quarter of the global adult rate. Investment sits at roughly a sixth of the proportionate global rate. Usage is ahead of investment nearly everywhere it has been measured this year, except in the one place, ordinary household adoption, where it depends most directly on literacy programmes nobody has funded yet.

September's Governance Scramble Followed the Data

The first week of September produced a cluster of ministerial statements that read, in sequence, like a region reacting to numbers it had just seen rather than numbers it had planned around. On September 3, Jamaica's technology minister called publicly for a unified CARICOM approach to AI governance. The next day, the Bahamas' minister responsible for the digital economy urged the country to build AI skills quickly and reform education to match, warning that the country needed to move fast on AI literacy specifically. Two days before that, Barbados' National Organisation of Women launched a free AI training programme for 35 women entrepreneurs, framed explicitly as making AI an affordable alternative for small business owners who could not otherwise access it.

The Barbados programme is small in absolute terms, 35 women in one cohort, but it targets the exact mechanism behind the 13% figure rather than the general idea of it. Consumer AI adoption does not fail evenly across a population. It fails first among people without an employer paying for a subscription, without English-language technical support built for their market, and without anyone in their network who has already tried the tools and can explain what they are for. A women-focused small business programme addresses all three at once, at a scale that a national literacy campaign, still being drafted in most CARICOM capitals, has not yet reached.

Behind all three statements sits the Caribbean Telecommunications Union's Caribbean AI Task Force, whose interim report, "Toward Harmonized AI Policies and Recommendations for the Caribbean," names the underlying condition directly: a connectivity paradox, in which strong digital engagement across the region collides with fragile infrastructure, fragmented regulation, and limited data governance. The report sets out five priority areas: regional AI governance, data sovereignty and digital infrastructure, innovation and industry development, human capacity and AI literacy, and sustained multi-stakeholder engagement. The Dominican Republic's 35% and the region's 13% are two live data points sitting inside exactly the gap that report describes.

None of these statements are wrong to make. The concern is sequencing. A business sector that has already reached 35% adoption, unsupervised, is a business sector already generating the data governance, liability, and accountability questions that the Caribbean AI Risk Management Council has spent the past year documenting across Caribbean AI deployments. Governance statements made after adoption has already reached one in three formal companies are catching up to a fact on the ground, not shaping one still forming.

What This Means for the Region's Next Twelve Months

The practical argument here is not that CARICOM needs another AI strategy document. The region has several already, in various stages of endorsement. The argument is that the money and attention now need to move toward the 13% figure with the same urgency businesses applied to reaching 35%, because that gap will not close on its own. Enterprise adoption is a budget decision made inside a company. Consumer literacy is a public investment decision, and public investment decisions require someone to fund them, which the region's 1.12% share of global AI capital makes structurally difficult without deliberate, coordinated effort.

National AI associations are where that effort is already visible in practice. AI Jamaica has been tracking the same literacy gap that shows up in the 13% figure, and its work sits alongside the Caribbean AI Association's broader push for a regional data sovereignty and governance framework that treats adoption and oversight as a single problem rather than two separate timelines. The Dominican Republic did not wait for that framework to reach 35% enterprise adoption. The rest of the region should not wait for it either, but the citizens left on the wrong side of the 13% figure will need a plan that businesses, moving on their own budgets, were never going to build for them.