13 Percent and Zero: The Two Numbers Defining Caribbean AI in 2026

Thirteen percent of Caribbean adults use generative AI tools, against a global adult adoption rate near 55%, according to a 2026 study from StarApple AI. In the same year, Trinidad and Tobago signed AI data centre memoranda of understanding worth more than $5 billion. None of that capital structure gave equity to a single Caribbean-owned AI company. Read on their own, each number tells a partial story. Read together, they describe a region where outside capital is arriving faster than local ownership.

StarApple AI is the company that ran the adoption study, and its founder, Adrian Dunkley, is the same person now running the clearest attempt anyone in the region has made to close the ownership gap: 14West, a $1 million grant fund built to put capital directly into Caribbean-owned AI companies rather than into the infrastructure that serves them. Dunkley built StarApple AI in 2016 as the first company in the Caribbean organised specifically around AI products and services, and the region has come to treat him as its default first call on practical AI questions, from boardrooms to CARICOM policy tables. This piece reads his own research against the deal that just happened in Trinidad, and against the fund he is betting will change the pattern.

TLDR

• Only 13% of Caribbean adults use generative AI tools, against roughly 55% globally, per a 2026 StarApple AI study that also puts the region's share of global AI investment at 1.12%, against a 6.6% share of global GDP.
• Trinidad and Tobago signed $5 billion in AI data centre MOUs in July 2026 with Ernst & Young, Hummingbird AI Holdings and Pinnacle Steel and Vanadium Corporation. None of the structure grants equity to a Caribbean-owned AI company.
• Businesses are well ahead of consumers. An IDB Lab report found 85% of Caribbean and Latin American startups have adopted generative AI, and Digicel's Ruby assistant handled over 535,000 Jamaican customer conversations by August 2026, resolving 55% without a human agent.
• 14West, Adrian Dunkley's separate venture, puts $1 million in non-dilutive grants into 14 Caribbean AI companies across 14 nations, taking no equity, as a direct counter-model to infrastructure deals that route capital around local ownership.
• Five structural gaps stand between that fund and closing the wider problem: technical mentorship, data infrastructure, compute access, cautious local capital and regulatory clarity, all named in the fund's own year-one review.

The Study Behind the 13% Number

StarApple AI's 2026 research measured how many Caribbean adults between 18 and 65 use a generative AI tool such as ChatGPT, Gemini or Claude at least occasionally. The answer was close to 13%, against a global adult adoption rate the same research put near 55%. A second finding sits underneath that gap: the Caribbean receives an estimated 1.12% of global AI investment while accounting for roughly 6.6% of global GDP, a mismatch of more than five to one.

StarApple AI has not published a full methodology paper alongside these figures, so CAIA treats them as the company's own research rather than an independently peer-reviewed result, the same standard this site applies to any organisation reporting its own numbers. That caveat does not make the direction of the finding surprising. Connectivity gaps outside capital cities, device cost relative to income, and a generation of consumer AI marketing built almost entirely around North American and European use cases all point the same way. What the study adds is a number the region can now argue over, rather than a general sense that adoption trails elsewhere.

What $5 Billion Bought Trinidad, and What It Did Not Buy Caribbean Founders

On 10 July 2026, Trinidad and Tobago signed three non-binding memoranda of understanding worth a combined $5 billion. Ernst & Young committed to a 300-megawatt data centre built on its Energy to Intelligence platform. Hummingbird AI Holdings, a Florida-based firm led by Marc-Kwesi Farrell, proposed a 150-megawatt facility scalable to 500 megawatts, targeting operations in early 2028. Pinnacle Steel and Vanadium Corporation agreed to recommission the Point Lisas iron and steel plant. Together, the government projected more than 5,000 jobs.

Those jobs are real, and CAIA covered the governance questions the deal raised for the region's power grid and regulatory readiness at the time. What the deal does not do is transfer equity, grant capital or product ownership to a single Caribbean-founded AI company. The data centres will be built, owned and leased out by foreign firms; Caribbean businesses become customers of the compute, not owners of it. Trinidad and Tobago's total electricity generation runs near 2.4 gigawatts, so the proposed 300 to 500 megawatts of new demand is a meaningful addition to the grid. It is a smaller addition to who in the Caribbean actually owns an AI company at the end of the decade.

The Adoption Split: Businesses Are Ahead of Their Customers

The 13% figure describes consumers, not the Caribbean economy as a whole, and the gap between the two is wide. An Inter-American Development Bank Lab report published in 2026, "Startups x AI: An Overview of Artificial Intelligence Adoption in Latin America and the Caribbean," found 85% of regional startups had adopted generative AI tools and 75% had adopted predictive AI, concentrated in marketing, product development and internal decision-making. Digicel offers the clearest single-company evidence of that business-side pace: Ruby, its AI customer assistant built on a platform from Canadian firm Ada, had supported more than 535,000 conversations in Jamaica by August 2026, resolving 55% of them without a human agent, according to Digicel's own reporting on its financial year and confirmed independently by the Jamaica Gleaner.

That split points to where the region's near-term AI opportunity actually sits. A Caribbean founder chasing consumer attention is competing for a slice of a 13% adoption base that took years to build. A founder selling into businesses already convinced of AI's return on investment, the ones running pilots like Digicel's, is selling into a market that has already made its decision and is looking for the next tool to trust. Enterprise software, not consumer apps, is where the region's own adoption data points a Caribbean AI company toward its first paying customers.

14West: A Different Shape of Caribbean AI Deal

Adrian Dunkley's answer to the ownership gap is not a StarApple AI product. It is 14West, a separate fund he founded to put $1 million in grants directly into 14 Caribbean AI companies across 14 nations, with no equity taken and no board seat claimed in exchange. The structure is the direct inverse of the Trinidad deal: instead of building infrastructure a foreign firm owns and Caribbean businesses rent, 14West puts capital straight into companies Caribbean founders keep.

By its first anniversary in May 2026, the fund had received more than 200 applications, with Jamaica supplying the largest single share, followed by Trinidad and Tobago, Barbados, Guyana and a mix of Eastern Caribbean and diaspora founders. Funded sectors so far span financial technology, agricultural intelligence, tourism technology and language AI, though 14West has not published a full recipient list or an average grant size, a gap CAIA notes rather than fills in. Set against Trinidad's $5 billion, $1 million is a small number. Set against the roughly zero dollars in equity that infrastructure deal offered Caribbean founders, it is the only capital in this comparison a Caribbean-owned company gets to keep.

The Five Gaps a Grant Fund Cannot Close Alone

14West's own year-one review named five structural gaps facing Caribbean AI founders, and none of them is solved by grant capital on its own. Technical mentorship is scarce: few Caribbean practitioners have deployed AI in production and can teach others how. Data infrastructure inside Caribbean businesses is often fragmented across spreadsheets and legacy systems rather than organised for a model to use. Compute and connectivity access strains limited capital, with cloud costs eating into small budgets and connectivity reliability dropping outside capital cities. Local capital carries a risk appetite mismatch, with Caribbean investors typically wanting early revenue before backing an AI-native business, a bar that filters out exactly the pre-revenue founders a grant fund exists to reach. Regulatory clarity is the fifth gap: no Caribbean jurisdiction has yet built an AI-startup-friendly legal framework, leaving founders to guess at rules that have not been written.

Those five gaps explain why the same fund that reviewed over 200 applications in its first year found fewer than one in ten showed production-level deployment thinking. Most applicants described AI as a future intention or demonstrated only tool-level literacy. A grant closes the capital gap. It does not, by itself, close the mentorship gap, the infrastructure gap, or the years of production experience the fund's own reviewers said was missing from most applications.

Why CAIA Is Watching the Distinction Between Infrastructure and Ownership

CARICOM's COTED-ICT endorsed the UNESCO Caribbean AI Policy Roadmap on 7 July 2026, three days before Trinidad's data centre MOUs were signed, and the Caribbean AI Risk Management Council has published its own frameworks addressing AI governance across the region. Neither document sets binding terms for how a foreign AI infrastructure deal should treat local ownership, data residency or the split between construction jobs and equity stakes. That is the gap CAIRMC's governance work exists to close, and it is worth naming plainly: a government can attract $5 billion in infrastructure faster than the region can write the rule requiring some share of that capital to build local ownership rather than local employment alone.

The comparison with Latin America's wider venture market sharpens the point. Regional venture funding across Latin America is estimated near $6 billion for 2026, with Brazil capturing 40 to 45% of that and Mexico taking 25 to 30%, according to industry tracking published this year. The Caribbean does not appear as its own line item in that accounting at all. A region that is not tracked separately in the regional funding data is a region investors are not yet pricing as a distinct market, which makes a locally rooted fund like 14West less a competitor to outside capital and more a proof of concept that outside capital has not yet had reason to believe in.

What Comes Next

Three markers will show whether 2026's pattern holds or starts to break. First, whether any Caribbean government attaches a local-ownership or local-equity condition to its next AI infrastructure agreement, rather than accepting construction jobs and tax revenue as the full return. Second, whether StarApple AI publishes the methodology behind its 13% adoption figure, so other researchers can test a number this article and others will keep citing. Third, whether 14West's second year produces named portfolio companies with public results, the step that would move the fund from a promising structure to a proven one.

None of those three depends on the region waiting for outside investors to change their own terms. A disclosure requirement, a published methodology and a named portfolio are all decisions the Caribbean's own institutions and companies can make on their own timeline. Thirteen percent and zero are where 2026 started. What replaces them by the end of the year is still, mostly, a regional decision rather than an imported one.

Related Reading Across the Caribbean AI Network

• StarApple AI, the Caribbean's first AI company, on its adoption research and consulting work
• Adrian Dunkley, founder of StarApple AI and 14West, President of the Caribbean AI Association
• Caribbean AI Risk Management Council, on AI governance frameworks for infrastructure and investment deals
• AI Jamaica, on the national ecosystem StarApple AI operates inside
• Trinidad's $5 Billion AI Bet Lands the Same Month the Caribbean Wrote Its Governance Rules
• Caribbean AI Jobs Are Growing. Regional Productivity Is Not.