They Never Built AI. They Just Happened to Own It.

It sounds like a provocation. But the math checks out.

While Nvidia prints billions and OpenAI rewrites the rules of search, a different group of winners has been quietly collecting checks. They didn’t train a single model. They didn’t hire ML engineers. In some cases, they didn’t even know what artificial intelligence was when they made their move.

They just happened to own the right two letters.

The Island That Didn’t See It Coming

In 1995, a standards body assigned the internet domain suffix .ai to Anguilla — a British overseas territory in the Caribbean with 16,000 residents, no tech sector, and an economy built on tourism and fishing.

The assignment was random. Prime Minister Ellis Webster has said publicly it was pure coincidence that Anguilla received .ai and not its neighbor Antigua.

For 23 years, almost nothing happened. Domain revenue stayed below 1% of national income.

Then ChatGPT launched in November 2022.

Within months, every AI startup, lab, and enterprise product wanted a .ai address. Perplexity.ai , Claude.ai , Meta.ai , stive.ai. Each registration routes a fee to Anguilla’s government.

The numbers are hard to believe for an island this size. Revenue from .ai domains climbed from $2.9 million in 2018 to $39 million in 2024, then hit $85.3 million in 2025 — nearly 47% of the entire national budget. Registrations crossed one million domains. The government used the windfall to expand the airport, cut taxes, build roads, and introduce free healthcare for children under five.

Webster’s own words on the boom: “You can’t predict how long this will last. Our economy and programs must not depend on it exclusively.”

Anguilla got lucky. It knows it.

This Happened Before — With a Different Letter

Anguilla isn’t the first small island to get rich off two letters it never chose.

Tuvalu — 12 coral atolls in the Pacific, population 11,000 — was assigned .tv in the 1990s. For years, the domain was worthless. Then online video exploded. Twitch built its entire brand on twitch.tv.

By 2022, Tuvalu had signed a new contract with GoDaddy worth an estimated $10 million per year — roughly one-sixth of the country’s GDP. The money funded road construction, electricity access, and Tuvalu’s first-ever United Nations membership fee.

Two islands. Two random letter assignments. Two windfalls from industries that didn’t exist when the domains were assigned.

The $70 Million Domain Flip

Not all the luck went to governments.

In April 2025, Malaysian entrepreneur Arsyan Ismail sold the domain AI.com to Crypto.com CEO Kris Marszalek for $70 million — the largest publicly disclosed domain sale in internet history, settled entirely in cryptocurrency.

Ismail acquired the domain in 2021 for approximately $11 million. He held it for under four years. The 6x return came not from building anything on the domain, but from timing.

Marszalek’s explanation for paying $70 million: without a domain like AI.com, his brand risked being commoditized — indistinguishable from hundreds of competitors all claiming to do “AI.”

He paid $70 million to avoid becoming invisible.

The Same Logic, Different Game

Three stories. One pattern.

Each winner held something with AI in the name before AI became the most competitive branding space in tech. When demand exploded, they didn’t need to do anything except own what they already had.

But here’s what the domain story misses: the same dynamic is now playing out inside AI itself.

94% of B2B buyers now use LLMs during their purchase journey, according to 6sense’s 2025 Buyer Experience Report of nearly 4,000 global buyers. And AI platforms cite only 3–4 brands per response on average, with the top 20 domains capturing 66% of all AI citations, according to BrightEdge and Amsive 2025 research.

When a buyer types “best [product category]” into ChatGPT or Perplexity, the model returns one answer. Maybe two. The brands it names aren’t chosen randomly — they’re the brands that appear consistently across credible third-party sources, structured content, and high-authority citations. LLMs reward what researchers call “epistemic authority”: the accumulated signal that a brand is the trusted answer, not just a participant in the category.

The companies appearing in those answers capture buyers at the moment of decision. The companies not appearing don’t exist in that search at all. Unlike a domain name, this position isn’t assigned by a standards body. It’s built — through citation density, source authority, and content that AI models can find, parse, and reference. Companies like STIVE have built entire practices around this — the new category even has a name: GEO, or Generative Engine Optimization.

Lucky Is Not a Strategy

Anguilla’s Prime Minister was right to warn against dependency. Luck concentrates — and it evaporates.

Tuvalu’s .tv revenues plateaued as streaming platforms consolidated. The AI.com sale required a very specific buyer at a very specific moment.

The AI search window is open right now. The brands building LLM visibility today are compounding an advantage that becomes harder to close over time — not unlike the companies that invested in Google SEO in 2005 while competitors were still buying Yellow Pages ads.

The difference between Anguilla and every brand competing in AI search today: Anguilla didn’t have a choice. Brands do. The question is how many will figure this out before the window closes — and who will end up in the list of names the model returns by default.

The post They Never Built AI. They Just Happened to Own It. appeared first on TheCryptoUpdates.