Billions in foreign investment. Thousands of jobs. A seat at the table of the global AI economy. That’s the pitch, and it’s not a bad one for a country that’s spent decades trying to shake its dependence on oil and gas. All Trinidad and Tobago has to do is find the electricity — and the water — to power it.

In July 2026, the government signed memoranda of understanding with two US firms, Ernst & Young and the relatively obscure Hummingbird AI Holdings, opening the door to what’s being billed as the Caribbean’s first major AI data centre push. Between the two projects, we’re looking at at least 450 megawatts of load capacity, with some reporting suggesting the combined footprint could eventually reach 800 MW if expansion options get exercised. EY’s plan centres on a 300-megawatt facility built through its Energy to Intelligence platform. Hummingbird is starting smaller — a 150 MW campus targeted for early 2028, with room to grow to 500 MW down the line. A third agreement covers recommissioning the Point Lisas iron and steel plant, tied to vanadium production, a metal used in aerospace and defence manufacturing that officials claim could eventually meet up to half of US demand if it all comes together.

Government officials are calling this the diversification story the country has been waiting for — deals that could bring in more than US$5 billion in investment and create over 5,000 jobs. It’s a good headline number. But look past the announcement and the questions start piling up.

Water T&T doesn’t have to spare

Data centres are enormously thirsty, and Trinidad is not a country with water to spare. A lot of households already live on rationed supply — many get tap water as infrequently as once a week and rely on storage tanks to get through the rest. Adding a hyperscale cooling operation on top of that isn’t a footnote. It’s the central problem.

Public Utilities Minister Barry Padarath has tried to get ahead of this, floating Debe as a proposed special economic zone and describing a hybrid plan built around man-made ponds in the short term, backed up eventually by desalination plants, with promises that residential and commercial WASA customers won’t be shortchanged. That’s fine on paper. Whether it holds up against an actual 300-megawatt facility running around the clock is something worth scrutinizing before ground is broken, not after.

A grid that wasn’t built for this

Electricity is the other half of the problem, and honestly the tighter one. The country’s generation fleet — roughly 2,289 MW combined across PowerGen, Trinidad Generation Unlimited, and Trinity Power — runs almost entirely on natural gas. Adding several hundred megawatts of constant, industrial-scale demand isn’t a rounding error against that. It’s a real chunk of the whole system. So who pays for the new generation, transmission, and battery capacity this would require? And how does that cost get split between the developers and everyone else’s electricity bill?

There’s an irony here too. Marketing this as the foundation of a “modern digital economy” while deepening the country’s reliance on fossil generation is a contradiction the government hasn’t really addressed. Trinidad’s heat and humidity don’t help either — the climate limits how much operators can rely on free outside air for cooling, which likely pushes facilities toward more energy- and water-intensive mechanical systems.

The public isn’t taking this quietly

This isn’t just a boardroom conversation. A Change.org petition calling for the projects to be paused pending real consultation picked up several thousand signatures within its first day and later crossed 13,000. Bioacoustician Syam Nath, who started it, has pushed back on the word “hyperscale” itself, arguing facilities of this size carry water, electricity, and infrastructure demands the public hasn’t been given straight answers about. Social activist Dr. Wayne Kublalsingh put it more bluntly, telling the Associated Press the government is presenting something that looks like development without actually being development.

Environmental consultants have raised a quieter set of concerns too — land use, e-waste from short hardware lifespans, and climate resilience. A facility meant to be critical digital infrastructure still has to survive Trinidad’s flood risk and heavy rainfall, and that needs to be built into the design from the start, not bolted on afterward.

What it would take to get this right

None of this makes data centres wrong for Trinidad and Tobago. The economics of AI infrastructure are real, and a small, stable country with subsea cable connectivity has genuine appeal to developers looking beyond an oversaturated US market. The MoUs, as the government keeps pointing out, are non-binding — a framework for due diligence, not a signed construction contract.

But frameworks tend to harden into facts on the ground faster than public consultation can keep up with. If this is going to be the diversification story Trinidad and Tobago needs, rather than the cautionary tale critics are already predicting, a few things need to happen before shovels go in the ground, not after: an independent environmental and resource-impact assessment made public in full, binding commitments (not aspirational language) on who pays for new generation and water infrastructure, and genuine consultation with the communities near Debe and Point Lisas — not a town hall staged after the decisions are already made.

There’s a real opportunity here for the government. There’s also real collateral at stake for everyone else, in the form of water pressure and electricity bills. Getting the balance right is still possible. It just depends on whether transparency or enthusiasm ends up driving what happens next.