Malaysia's 3.4 Million AI Businesses: What the Number Is Counted Against

This year's AWS report put Malaysian AI adoption at 38% of businesses, up from 27% the year before, and turned that into a headline of 3.4 million businesses using AI in some form. When Tech Wire Asia asked what the 38% had been applied to, AWS answered in writing that the base was roughly nine million registered businesses and that "no dormancy adjustment was applied." The adoption trend is probably real, but the 3.4 million is a statement about registrations, and a reader who takes it as a count of working firms will overestimate the market by a wide margin.

What the percentage was applied to

The survey itself is sound enough in design: Strand Partners questioned 1,000 business leaders for AWS, and the 38% is the share of them who said they use AI in some form. The difficulty arises one step later, when that share is multiplied by a population figure, and the population AWS used was the Companies Commission of Malaysia (SSM) register, which stood at 9,697,702 registered businesses at the end of 2025, with a further 200,726 added by the end of July 2026. That register records registrations only, with no split between active and dormant entities, and a business that was incorporated years ago and trades no longer still counts in it.

The last published split I can find is from August 2024, when the SSM chairman told Bernama that of 9.32 million registered businesses, 1.66 million were active, which is roughly 18%. Tech Wire Asia applied that ratio to the current register and arrived at about 1.7 million active businesses, and 38% of that is somewhere between 650,000 and 700,000 adopters rather than 3.4 million. I would treat this as the publication's own estimate and not as a corrected official figure, since the 18% is two years old and SSM has not released a newer one (the publication says it has asked, and will update if SSM responds). There is also a fair argument that a survey of 1,000 business leaders is, by construction, a survey of people who are still running something, so applying its result to dormant registrations quietly assumes that dormant firms adopt AI at the same rate as live ones.

The findings that matter more than the headline

The same AWS report contains figures that seem to me far more useful to an owner deciding what to do next, and they do not depend on the denominator at all because they describe the businesses that answered. Among adopters, 67% are still at a basic stage (down from 73%), 20% are intermediate, and 13% are advanced or transformative, up from 10% the year before, which fits what I wrote in the earlier piece on why these surveys disagree: the movement that counts is from scattered personal use towards changed processes, and it is slow.

The governance numbers are where I would look first. Only 19% of respondents have a formal roadmap for scaling beyond pilots and 16% a comprehensive AI strategy, while 21% have a formal AI governance framework, 30% say accountability for AI decisions sits with a defined person or team, and 18% have a documented escalation process for AI risks. Set against that, 53% name workforce skills as the main barrier, and only 34% rate their own workforce's digital skills as good or excellent. In short, most businesses that count as adopters have tools in use and no one clearly answerable for them.

A note on who counts as an SME

One more detail worth knowing if you intend to compare yourself with these figures: the report defines SMEs as firms with fewer than 250 employees, whereas SME Corp Malaysia uses fewer than 200 employees or RM50 million in annual turnover. AWS says it chose the wider definition for comparability across Asia Pacific and Japan, which is a reasonable choice for a regional study but means the Malaysian SME numbers in it are not drawn from quite the same population as the one your own sector body uses.

What This Means for You

Adoption in Malaysia is very likely rising, and AWS's own year-on-year movement in maturity points that way, but the absolute number of businesses involved is harder to pin down than the headline suggests, and until SSM publishes a current active count nobody outside it can say precisely. For an owner, the practical response is to stop benchmarking against the headline and benchmark against the maturity and governance figures instead: whether you have a named person accountable for AI decisions, and whether anyone in the business knows what your staff are actually using. Most firms I speak with could answer both in an afternoon, and the answers will tell you more about your position than 3.4 million ever could.