Malaysia Trains Women in STEM,
Then Sends Them to Run Shops and Kitchens.
I keep seeing “Women in AI” roundtables. The instinct is to ask where all the women in tech are — and the honest answer stings, but not the way you’d guess. Malaysia isn’t short of women who can do this. We produce them, then quietly route them somewhere else. And AI is about to make that detour very expensive.
The roundtable that started this
Our National AI Office recently sat on a “Women in AI” executive roundtable — leaders from a bank, an oil major, the government, talking about harnessing AI, empowering talent, a future-ready workforce. Good people, real event, worth doing.
But I run an AI company. I see who actually walks through the door to build with this technology, and who doesn’t. So the roundtable left me with a blunter question than the one it was set up to answer: forget empowerment slogans — what is the realistic expectation for women in AI in Malaysia, given where they already are? I went and pulled the numbers. They tell a stranger story than either the optimists or the cynics expect.
We do not have a supply problem
Here is the part that surprised me. On STEM education, Malaysia beats the global trend. Women are strongly represented among our STEM graduates — in many cohorts, at or above half. Female graduates work at a high rate too. If the problem were “not enough women can do the technical work,” the education pipeline would show it. It doesn’t.
So the lazy version of this argument — that women are thin in AI because they’re weak in tech — is simply wrong, and the data kills it. We are not failing to produce technical women. We produce plenty.
The failure is downstream. Women make up only about 35% of Malaysia’s technology workforce. We train them at parity and then lose most of them somewhere between the graduation stage and the technical job. That is not a pipeline problem. It is a leak.
Where the leak goes: shops and kitchens
Follow the women who do strike out on their own, and the leak becomes vivid. Women own about 20% of Malaysian businesses. Sounds like progress — until you see where those businesses are.
According to the Department of Statistics, 93.6% of women-owned businesses sit in services, and within that, wholesale and retail (45%) plus food and beverage (31%) account for over three-quarters of them. Tech and ICT is a sliver.
Put the two facts together and the sentence writes itself: Malaysia trains women in STEM, then sends them to run shops and kitchens. Not because they can’t code — many were taught to — but because retail and F&B are the paths with the lowest barrier, the most role models, and the least resistance. The technical ambition gets trained in and then routed around.
I want to be careful here, because this is easy to say badly. There is nothing lesser about running a retail or food business — some of the sharpest operators I know do exactly that. The waste isn’t that women run shops. The waste is that a country producing technical women at parity converts almost none of that into technical ownership, in the one decade when technical ownership is where the wealth is being created.
Why AI changes the maths — in both directions
AI makes this leak more urgent and, for the first time, more fixable. Both at once.
More urgent: AI is the largest wealth-creation event of our working lives, and it is forming precisely where women are already thinnest — technical roles and tech founding. A gap left alone during a normal decade stays a gap. A gap left alone during this decade compounds into a chasm, because the people building with AI now are setting the defaults everyone else will rent for years.
More fixable: here is the twist that made me want to write this. Every previous tech wave demanded years of coding before you could build anything — and that barrier is a large part of what leaked women (and plenty of men) out of tech founding. AI is the first wave where that barrier is gone. You can build a real product now by describing what you want in plain language. The skill that matters shifts from “can you write the code” to “do you understand the problem and the customer” — and that is not a skill Malaysian women are short of.
The early signal is already in the data: the fastest-growing category of women-owned business in Malaysia is information and communication, climbing about 31.8% a year. Small base, steep curve. That is what the beginning of a correction looks like.
What actually moves it
Slogans don’t move a leak; defaults do. If I were spending the effort a roundtable represents, I’d point it at three unglamorous things:
Convert, don’t recruit. The women already exist — STEM graduates who leaked into non-technical work, and the 76% running retail and F&B. Reaching a shop owner who already understands a customer and showing her she can now build the tech herself is a shorter path than raising a new cohort from scratch.
Teach building, not awareness. “AI literacy” talks fill rooms and change nothing. A woman who ships one real thing with AI — a working assistant, an automated back office — has crossed a line no seminar moves her across.
Count ownership, not attendance. The metric that matters is not how many women attend AI events. It is how many women own AI-native businesses in three years. If that number isn’t moving, the roundtables are theatre — however well-intentioned.
The bottom line
The expectation for women in AI in Malaysia should not be low. On the raw material — trained technical women — we are ahead of most of the world. We just throw the advantage away at the exit, every year, and have done for a generation.
AI is the first tool sharp enough to close that gap from the other side — if we aim at ownership instead of attendance. That’s not a diversity nicety. It’s the highest-leverage economic bet Malaysia isn’t making.
Published by IMA AI — July 2026. Figures from the Department of Statistics Malaysia and national labour data. We’d rather help build the businesses than attend the panels about them.