Ask a founder in 2026 why they're launching in six markets at once and the answer is usually some version of: because we can. AI translates the product overnight. Support runs through agents in forty languages. Market research that used to take a consultant a month arrives before lunch. The mechanical cost of going international has collapsed.
That part is true. The conclusion founders draw from it — that international expansion got easy — is not. The mechanical work was never what killed expansions. Companies with perfect localisation die in market after market, and companies with barely-translated products win, because someone on the ground got three judgment calls right. AI has made the cheap parts of expansion cheaper. The expensive parts haven't moved.
The cheap parts got cheaper
The genuine change is worth naming. Localisation, first-line support, regulatory summaries, competitive research, entry-plan drafts — work that used to require weeks and headcount now takes days and a good prompt. A ten-person company can credibly look global. That leverage did not exist when I was doing this at Uber.
But lower entry cost has a second-order effect: more companies attempt expansion, earlier, with less preparation. When launching a market costs almost nothing, the discipline to not launch becomes the scarce resource.
The expensive parts didn't move
Across 45 countries at Uber, the decisions that determined whether a market worked were never mechanical. Which market next — chosen on regulatory fit, competitive gap, and distribution leverage, not headline size. Which regulator to sit down with before launch, and what to concede. Which local partner actually controls distribution. Who runs the country — the single highest-leverage decision, and the one founders consistently under-invest in.
None of that is automatable, because none of it is information. It's judgment under local ambiguity. AI will hand you a convincing forty-page market entry plan for free — which is precisely the problem. Confidence is now cheap. Knowing which page of the plan is wrong still costs what it always did.
The line that moves
The operating model I've written about before — maximum local autonomy on genuinely local decisions, maximum standardisation on everything else — doesn't get replaced by AI. The line moves. More can now sit in the centre: support, tooling, reporting, most of localisation. Local teams get smaller. But that makes the judgment of each person on the ground more valuable, not less. The new shape is fewer people per market, and a much higher bar for who they are.
So AI didn't lower the bar for international expansion. It raised the penalty for expanding without judgment — because now everyone's deck looks ready. The map got cheaper. The territory didn't.