Anguilla has no data centers, no chip fabrication plants, and no AI research lab of note. It has 16,000 people, a coastline, and two letters. In 2025 those two letters generated an estimated $93 million for a territory whose technology minister, Jose Vanterpool, told the BBC that .ai domain sales were expected to fund nearly half of government revenue that year. The 2026 budget closed in surplus, driven largely by that same non-tax revenue stream. No island in this region has ever converted a coincidence of alphabetical assignment into fiscal policy quite like this.
Here is the coincidence. When the International Organization for Standardization handed out two-letter country codes in the 1970s, Anguilla received AI. Nobody in that room was thinking about artificial intelligence. They were building a directory. Fifty years later, a global funding boom made ".ai" the default domain suffix for any company that wants to signal what it does, and Anguilla found itself sitting on the internet's most in-demand piece of branding real estate. More than a million .ai domains were registered by the start of 2026, with roughly 2,000 new registrations a day in January alone, and close to 28 percent of newly founded tech startups now choose a .ai address over a .com. Anguilla did not build that demand. It owned the supply.
I want to be precise about why this story matters to the rest of the Caribbean, because the easy reading of it is wrong. The easy reading says Anguilla got lucky and the rest of us didn't. The harder and more useful reading is that Anguilla is the only clear counterexample to a regional pattern that Anthropic's own usage data confirmed again this year: the global AI economy is not spreading its gains evenly, and the Caribbean, with one small exception, is mostly watching from outside it.
What Anthropic's Own Data Says About Who Is Winning
Anthropic publishes an Economic Index built from real Claude usage patterns rather than survey responses, and its March 2026 report contains a number that should worry every Caribbean finance ministry more than it apparently has. The top 20 countries by per-capita Claude usage now account for 48 percent of all per-capita usage worldwide, up from 45 percent in the previous reporting period. That is not a rounding error. It is concentration moving in the wrong direction. Inside the United States, adoption is converging: the top 10 states' share of usage dropped from 40 percent to 38 percent over the same window, and Anthropic's researchers now project full state-level convergence within five to nine years. Internationally, the opposite is happening. The countries that started ahead are pulling further ahead.
A companion analysis from Anthropic's economic research team found that a 1 percent increase in a country's GDP per working-age capita correlates with roughly a 0.7 percent increase in Claude usage per capita. That is a strong relationship, and it explains a pattern the report lays out by name: Indonesia using Claude at 0.36 times the rate its population would predict, India at 0.27 times, Nigeria at 0.2 times, Bolivia at 0.48 times. Most Caribbean nations do not appear in these country-level breakdowns at all, which is its own signal. You do not get named in a global usage ranking when your usage is too small to register.
None of this is a knock on Caribbean people or Caribbean ambition. It is a description of what happens when AI adoption tracks existing income levels almost mechanically, and small, import-dependent Caribbean economies sit well below the income threshold where the correlation Anthropic describes starts producing heavy usage. The gap is not a mystery. It is arithmetic.
Anguilla Broke the Pattern Without Adopting More AI
What makes Anguilla worth studying is that it did not close this gap by getting more of its 16,000 residents to use Claude or ChatGPT more heavily. Its own domestic AI usage almost certainly still sits far below the countries topping Anthropic's per-capita rankings. Anguilla closed a different gap entirely. It found an asset that the global AI industry needed badly enough to pay for, and it priced that asset well.
Revenue from .ai registrations climbed from about $2.9 million in 2018 to $39 million in 2024, according to registry figures reported by Semafor, before nearly doubling again to an estimated $93 million in 2025 as reported by Domaintechnik. That growth curve tracks the AI funding boom almost exactly, because the demand was never really about domain names. It was about branding pressure from a market where a company without ".ai" in its name looks like it is explaining itself instead of stating it. The most striking recent data point: a single domain, bot.ai, sold in February for $1.2 million, the highest publicly known .ai domain sale on record.
Anguilla's government used that windfall for airport expansion, road construction, tax relief, and expanded health services, according to reporting on the 2026 budget. That is a small government doing something a lot of larger Caribbean governments have struggled to do with AI-adjacent revenue of any kind: converting a technology cycle into concrete public infrastructure within a single budget cycle, without a decade of committee meetings first.
The Hardware Race Shows Where the Next Windfalls Will Come From
Anguilla's registry windfall is a domain-name story, but it points at something bigger happening one layer down in the AI stack. Reuters reported in mid-June that Qualcomm is in advanced talks to acquire AI chip startup Tenstorrent for between $8 billion and $10 billion, a move aimed squarely at breaking into the AI accelerator hardware market that Nvidia and AMD currently dominate. Tenstorrent, founded in 2016 and led by veteran chip architect Jim Keller, builds its processors on the open RISC-V architecture rather than proprietary designs, which is part of why Qualcomm wants it: RISC-V lowers the cost of building custom AI silicon for whoever controls the design.
I am not suggesting any Caribbean territory is about to build a competitive chip company. That would be a bad prediction and I am not making it. What the Qualcomm-Tenstorrent talks illustrate is where the real capital in this industry is actually flowing: toward the physical and architectural layer of AI, the chips, the power, the data centers, the subsea cables that connect them, not only toward the software layer most Caribbean AI conversations stay fixated on. Anguilla's domain windfall is a software-adjacent asset. The next round of windfalls, for whichever small economies position for them, will come from the physical layer: land, power capacity, connectivity, and regulatory frameworks the AI buildout genuinely needs and cannot manufacture out of nowhere.
The Dominican Republic has already started moving on this. Reporting from February described agreements positioning the country as a regional technology hub, including infrastructure commitments involving Nvidia and Google alongside a subsea cable and connectivity project. That is a Caribbean government doing the asset-ownership math instead of the adoption-rate math. It will not show up in Anthropic's per-capita usage rankings for years, if ever, because the value it captures is infrastructure revenue and strategic positioning, not chatbot usage.
What This Means If You Run a Caribbean Government or Business
The instinct in most Caribbean AI strategy documents I have reviewed over the past decade is to treat the gap Anthropic's data describes as a training problem. Teach more people to use AI tools, run more workshops, publish more adoption guides. I have built plenty of that training myself and I still believe in it. But training alone will not move a country's position in a global usage ranking driven primarily by GDP per capita, because training does not change GDP per capita quickly enough to matter.
Anguilla's example points at a different question every Caribbean government and firm should be asking right now: what do we own, or what could we own, that the global AI buildout structurally needs? For some territories that is renewable power capacity that a data center operator would pay a premium for. For others it is subsea cable landing rights, or a data sovereignty and privacy framework attractive enough to host regional AI workloads that cannot legally sit on US servers, or specialized technical talent trained deeply enough in a narrow domain that global AI labs come looking for it rather than the other way around. None of these are guaranteed windfalls. All of them are more within reach for a small economy than trying to out-adopt countries with ten times the GDP per capita.
At StarApple AI, the company I founded in 2016 as the Caribbean's first, this is the shift we have been pushing regional partners toward for the past two years: stop measuring success purely by how many people in a territory use a foreign AI assistant, and start measuring what assets a territory actually holds that the AI industry needs. The Caribbean AI Association has been making a version of this argument at the policy level, and groups building out national AI capacity, from the Jamaica Artificial Intelligence Association to the Caribbean AI network's Guyana and Barbados chapters to Saint Lucia's and Trinidad and Tobago's national AI bodies, are the ones with the standing to turn asset-ownership thinking into actual government policy rather than another conference panel.
The Gap Is Real. So Is the Playbook for Closing It.
I want to end where the data actually points rather than where it is comfortable to end. The Anthropic numbers are not encouraging for the Caribbean this year. Global concentration went up, not down. Most of this region is invisible in the country-level rankings that matter to investors and multilateral lenders deciding where AI-linked capital goes next. That is the honest starting position.
But Anguilla's $93 million did not come from a Caribbean nation catching up on AI adoption metrics. It came from a 16,000-person territory recognizing, faster than almost anyone else in the world, that it already owned something valuable and pricing it accordingly. Every Caribbean territory has some version of that asset sitting unpriced right now, whether it is power, connectivity, regulatory positioning, or talent built for a specific niche the rest of the world hasn't filled yet. The countries that go looking for their own version of Anguilla's two letters, rather than waiting for the next Anthropic report to explain why they are behind, are the ones that will show up differently in next year's data. Insurers modeling Caribbean climate and economic risk at caribbeanairisk.com, and the growing set of national AI bodies tracking this from Georgetown to Bridgetown to Castries, are watching the same numbers I am. The gap will not close itself. Somebody has to find the asset and price it.
Frequently Asked Questions
How much revenue does Anguilla's .ai domain generate?
Anguilla's .ai country-code domain generated an estimated $93 million in 2025, up from about $39 million in 2024 and roughly $2.9 million in 2018, according to registry data reported by Semafor and Domaintechnik. Anguilla's technology minister, Jose Vanterpool, told the BBC the domain was expected to produce nearly half of the government's total revenue in 2025. The 2026 government budget recorded a surplus driven largely by that non-tax revenue.
Why does the AI industry want .ai domains?
The .ai suffix reads as a direct label for an artificial intelligence company, which makes it valuable branding real estate during an AI funding boom. More than a million .ai domains were registered by early 2026, with roughly 2,000 new registrations a day in January, and about 28 percent of newly founded tech startups now register a .ai domain rather than a .com. Anguilla holds the registry rights to the suffix purely because it was assigned the ISO country code AI decades before anyone anticipated what those two letters would come to mean.
What does the Anthropic Economic Index say about global AI adoption in 2026?
Anthropic's March 2026 Economic Index report found that the top 20 countries by per-capita Claude usage now account for 48 percent of all per-capita usage globally, up from 45 percent in the previous reporting period. That is concentration increasing, not narrowing. A separate Anthropic analysis found that a 1 percent increase in a country's GDP per working-age capita is associated with roughly a 0.7 percent increase in Claude usage per capita, and identified countries including Nigeria, India, and Indonesia using AI at a fraction of the rate their population size would predict.
Is the Caribbean part of the global AI adoption gap?
Largely yes. Most Caribbean nations are small economies without the working-age GDP per capita levels that correlate most strongly with heavy AI assistant usage in Anthropic's data, and most Caribbean territories are absent from published top-adoption country rankings entirely. The exceptions prove the pattern rather than break it. Anguilla's outsized position in the AI economy comes from owning valuable registry infrastructure, not from broad domestic AI adoption, which shows there is more than one way for a small Caribbean territory to capture value from this technology cycle.
What is the Qualcomm-Tenstorrent deal and why does it matter here?
Qualcomm is reportedly in talks, first reported by Reuters in mid-June 2026, to acquire AI chip startup Tenstorrent for between $8 billion and $10 billion, aiming to compete with Nvidia and AMD in AI accelerator hardware. Tenstorrent, founded in 2016 and led by veteran chip architect Jim Keller, builds AI processors on the open RISC-V architecture. The deal illustrates how much capital is chasing the physical infrastructure layer of AI, the chips and data centers, not just AI software. The Caribbean has no stake in that infrastructure layer today, and that absence is a bigger long-term problem than the region's software usage numbers.
What should Caribbean governments do differently based on this?
Stop treating AI strategy as a training and adoption question only, and start treating it as an infrastructure and ownership question. Anguilla did not out-adopt anyone. It owned a scarce digital asset the AI industry needed and priced it correctly. Caribbean governments and firms should audit what digital, regulatory, or physical assets they hold that the AI buildout actually needs, whether that is renewable power capacity, subsea cable landing rights, data sovereignty frameworks, or specialized local talent, and negotiate from that position rather than waiting to be a market someone else's AI product eventually reaches.
What is StarApple AI's role in closing this gap?
StarApple AI, founded in 2016 as the Caribbean's first AI company, works on building regional AI capability that does not depend entirely on foreign platforms, training local talent, and pushing Caribbean governments and firms toward the same asset-ownership thinking that made Anguilla's .ai windfall possible. The gap the Anthropic data describes will not close through more conference panels. It closes when Caribbean entities own something the global AI buildout needs.