Jamaica’s announcement of an exploratory MOU for “AI infrastructure” has sparked immediate debate, and rightly so. Although positioning the Caribbean region for the AI economy is an urgent priority, headlines alone don’t build resilient digital ecosystems. Our latest analysis unpacks the gap between political rhetoric and small-island realities.

 

At the Jamaica Labour Party’s annual conference in Montego Bay on Sunday, November 15, 2026, Prime Minister Dr Andrew Holness announced that the Government had approved a Memorandum of Understanding (MOU) with a group of investors to explore creating artificial intelligence infrastructure in Jamaica (Source:  Jamaica Observer). Framed around a familiar and urgent refrain—that Jamaica was left behind during previous industrial and digital waves and cannot afford to miss the AI revolution—the announcement was pitched as a springboard toward a new economy and future jobs for young Jamaicans.

Within 24 hours, Opposition Spokesperson on Science, Technology and Digital Transformation Christopher Brown pushed back sharply, demanding that the MOU be published and tabled in Parliament, and questioning how such an agreement could move forward ahead of a settled National AI Policy and data-centre governance framework (Source:  Jamaica Observer). By the post-Cabinet press briefing on November 19, Information Minister Senator Dr Dana Morris Dixon sought to cool the temperature, describing the pact as an early, exploratory MOU with no finalised deal or committed investor, while insisting Cabinet had set guardrails against straining the electricity grid or water systems (Source:  The Gleaner).

For the Caribbean ICT ecosystem, this back-and-forth is more than routine political sparring. It exposes a widening gap between headline-grabbing “AI infrastructure” announcements and the hard engineering, policy, and economic realities of Small Island Developing States.

 

The vagueness behind the announcement

What actually is “AI infrastructure” in the context of Sunday’s announcement? Right now, the public has far more rhetoric than clarity, and thus a lack of transparency whilst still wishing to assert that the Government is being progressive and proactive.

First, the Prime Minister was very vague about the MOU and has not clarified whether it contemplates hyperscale AI data centres, colocation facilities, sovereign/private compute clusters, or broader digital/connectivity investments. Additionally, no information was shared about the prospective investors, the exact terms of exploration, and any potential state incentives, land allocations, or other commitments made. Finally, although the later post-Cabinet remarks stressed that an exploratory MOU is not a binding investment deal, announcing an approval on a party platform creates expectations of imminent momentum – even when no investor is locked in, and no concrete project plan exists

In tech policy, MOUs are often routine conversation-starters. Treating an early exploratory document as a grand strategic milestone without publishing its text invites scepticism, rumour, and unnecessary polarisation.

 

Where the Opposition gets it right

Christopher Brown’s demand for transparency and policy sequencing hits on several real governance vulnerabilities that ought to be carefully considered.

First, signing or approving agreements touching high-impact digital infrastructure before finalising a clear National AI Policy and legal/regulatory framework for data centres risks putting governance on the back foot. Rules on data governance, environmental compliance, land use, and permitting should anchor negotiations, not trail behind them.

Further, the development of a National AI Policy has been a high-profile Government initiative, emphasising the need for a coherent framework to guide the use of one of the most transformational technologies of our generation. Further, it may not bode well if this group of investors become excited about potential opportunities in the current environment only for the Government to have to claw them back when the policy is established.

Second, and for such a transformational technology, it is concerning that there has been no parliamentary scrutiny and public accountability. Further, tabling or releasing the MOU lets lawmakers, industry practitioners, and citizens test whether the state’s exploratory posture protects public interests rather than fostering information asymmetry and unmanaged expectations.

Finally, asking questions such as who the investors are, what power and water demands might arise, and who pays for any grid or infrastructure upgrades goes straight to the heart of public accountability. Although assurances were given post-Cabinet that the Cabinet intends to set boundaries around the impact on water and the electric grid, those principles need concrete technical and contractual mechanisms rather than broad assurances alone.

Moreover, these questions are confounding larger economies that are considered better positioned to support large-scale AI deployments.  It could thus be argued that, knowing our priorities, imperatives and constraints, we in the Caribbean ought to identify the segments in which we would welcome investment—to achieve our own objectives and goals—instead of seemingly giving prospective investors carte blanche to do as they please.

 

The broad challenges for Jamaica and the wider Caribbean

If AI compute or data-centre infrastructure is eventually built in Jamaica and/or other neighbouring Caribbean countries, it will land in a region with very different constraints from North America or Europe:

  • Energy cost, reliability, and grid capacity. High-density AI servers require massive, continuous power. Across much of the region, electricity tariffs are already high, grids face reliability and reserve constraints, and adding heavy continuous loads, or requiring dedicated generation, raises serious questions about tariff stability and system resilience.
  • Water stress and cooling trade-offs. Water-intensive cooling remains a sensitive issue in island jurisdictions where communities already experience supply disruptions or seasonal shortages that are getting longer year on year. Air-cooled or closed-loop systems can help, but every facility must be evaluated against local water balances along with environmental and climate pressures.
  • Climate exposure and physical resilience. Hurricanes, flooding, storm surge, heatwaves, and coastal vulnerability are everyday operational realities in the Caribbean region. Any critical compute facility must be engineered for severe weather, power fluctuations and other challenges and be able to recover without creating liabilities for surrounding communities.
  • Subsea connectivity and latency economics. An often overlooked point is that Caribbean digital ambitions depend heavily on submarine cable diversity, landing station resilience, and transit pricing. Without strong redundancy and competitive bandwidth economics, noting that we currently have a monopoly in submarine connectivity in most countries, local compute can struggle to serve regional or international markets cost-effectively.
  • Data sovereignty, regulation, and institutional capacity. Finally, noting that a potential AI data centre hosted in the region could host sensitive government, health, financial, or enterprise data, could result in questions being raised around jurisdiction, cybersecurity, data protection enforcement, and regulatory readiness. Jamaica, for example, has a comprehensive Data Protection Act that was brought into force in 2020, but nearly six years later, the Government is struggling to fully implement it. Concerted work is also being done to strengthen the country’s cybersecurity and cybercrime posture and capabilities, but its successful and timely realisation will depend to an appreciable degree on having access to the requisite funds and expertise.

 

The job-creation myth: Compute does not equal mass employment

One of the most persistent myths in digital development that the Prime Minister seems to be perpetuating is that building data centres or AI infrastructure automatically generates large numbers of local jobs. In reality, modern data centres and AI compute facilities are capital-intensive, not labour-intensive. Once construction wraps up, day-to-day operations typically require a relatively small, specialised technical team, typically covering electrical/mechanical systems, cooling, physical and cyber security, and network/server maintenance. They do not function like traditional BPO contact centres or factories that employ thousands of workers on a single site.

Where real economic and employment value can come from is adjacent rather than automatic, such as:

  • Local engineering, construction, and maintenance contracts, if structured so local firms can participate rather than being crowded out.
  • Software, AI application development, and domain-specific solutions built by Caribbean engineers and startups solving regional problems in tourism, logistics, fintech, agriculture, and public services.
  • Targeted workforce pipelines that are connecting technical training, university AI programmes, and apprenticeships to real industry demand.

Further, it must be highlighted that these job opportunities, especially the high-value job opportunities, are likely to require specialist skills. Hence, if our countries are serious, there should be active work in ensuring that we possess the requisite skills to actually take advantage of emerging AI opportunities.

 

In summary, if Jamaica and the wider Caribbean want AI infrastructure to translate into broad-based prosperity, compute cannot be treated as a silver bullet. It has to sit inside a deliberate strategy, starting with clearer facts on the table.

 

 

Image credit:  kues1 (Magnific)