Health & AI systems
My bet: AI quietly breaks how the Global South earns dollars.
For thirty years the deal was simple: sell cheap labour, earn hard dollars. Now a model does the work, and the dollars still leave.
That trade is what keeps India, the Philippines, Bangladesh, Vietnam, Kenya and Egypt standing.
Now a model does the work.
And here is the part nobody is pricing in.
Even when AI does not take the job, the dollars still leave.
Every time a model runs, someone pays a foreign AI company in dollars.
I call it the token-dollar.
I run GSD.
We run on-the-ground teams for companies entering Africa and Asia.
So I watch these cost structures up close, and this is the shift I think gets underpriced.
So I put nine economies on one chart, scored two ways.
Left to right: does AI make your work cheaper, or make it disappear?
Cheaper is survivable. Disappear is not.
Bottom to top: can you build your own AI, or are you stuck renting it?
The bottom-right corner is the danger zone.
Work vanishing, no way to fight back.
The Philippines, Bangladesh, Kenya, Nigeria, and Ghana are all sitting there.
The Philippines is the clearest case.
Outsourcing is about 8% of its GDP, mostly call centres and back-office work.
The exact jobs a model does first.
Only India and Vietnam sit in the safe corner.
One reason: they are big enough to build their own AI instead of importing it.
The way out is not to stop using AI.
It is to own the layer next to it.
Pool compute across countries.
Run open models locally.
Climb to the work that a model cannot do alone.
Negotiate as a buyer, because the Global South is a top market for these AI companies.
Where would you move a country on this chart, and why?