- AWS and Strand Partners have confirmed Malaysia’s headline AI adoption figure of 3.4 million businesses is calculated against SSM’s total registered businesses, not active ones.
- SSM’s own past disclosures suggest fewer than a quarter of registered businesses are actually trading, a gap that changes what “3.4 million” really means.
The headline number in AWS’s newest Malaysia AI adoption research is 3.4 million businesses. That is how many companies, AWS says, now use AI in some form, based on a 38% adoption rate this year versus 27% last year, drawn from a survey of 1,000 business leaders conducted by Strand Partners.
AWS calls it a pace of “almost two new AI adopters every minute.” What the research does not say, at least not out loud, is what population that 38% is measured against.
Tech Wire Asia put the question to AWS directly after the figure was presented at a media briefing in Kuala Lumpur this week. The answer, confirmed in writing, changes how the number should be read.
The Malaysia AI adoption math, explained
Buried in the report’s appendix, not in the body of the research, is the base used to turn a survey percentage into a headline count. Footnote two states the 3.4 million figure comes from applying the 38% adoption rate to “approximately nine million registered businesses in Malaysia,” sourced to the Companies Commission of Malaysia’s (SSM) published registration statistics.
Asked directly whether that nine million reflected total registered entities or businesses still operating, AWS confirmed it was the former. “No dormancy adjustment was applied,” the company said in a written response, adding that SSM’s cited statistical series “reports registered totals only and does not include an active/dormant breakdown.”
Tech Wire Asia pulled the same SSM page AWS cited. As of December 31, 2025, SSM’s own table shows 9,697,702 registered businesses, alongside a separate 1,650,847 registered companies and 45,196 limited liability partnerships. The table has no active or dormant column, exactly as AWS described. By the end of July 2026, another 200,726 businesses had been added, taking the running total to roughly 9.9 million, with no indication anywhere on the page of how many of those are still trading.
SSM’s own numbers tell a different story
Ahmad Sabki Yusof, SSM’s chairman, disclosed a breakdown the regulator does not publish in its regular statistics, telling Bernama in August 2024 that as of July 31 that year, SSM had registered 9.32 million businesses and incorporated 1.56 million companies. Of those, only 1.66 million businesses and 677,351 companies remained active.
Put plainly, fewer than a quarter of everything registered with SSM was actually trading. The rest sits dormant, deregistered in practice but not on paper, or simply never activated after registration.
Applying that same ratio, roughly 18% active, to the 9.7 million registered businesses SSM now reports, and Malaysia’s active business population looks closer to 1.7 million, not 9.7 million. Apply 38% adoption to that smaller number instead, and the resulting figure lands somewhere around 650,000 to 700,000 businesses, not 3.4 million.
This is an estimate, not a confirmed figure. It uses SSM’s most current registration total against an active rate the regulator disclosed back in 2024, which may itself have shifted. SSM has not published an updated active-inactive split since then, and Tech Wire Asia is reaching out to the regulator for current data. This article will be updated if and when SSM responds.
The report is telling the real gap well beyond the adoption numbers
AWS said the same registered-business base has been used “on a consistent basis across report editions to allow tracking over time,” meaning last year’s 27% adoption figure was very likely measured against a similarly inflated denominator. The year-on-year direction, adoption rising, is probably real. Its scale, in absolute business counts, is harder to defend.
Strip out the disputed top line, though, and the report underneath makes a more careful argument than the press release lets on. Even among businesses AWS counts as adopters, 67% remain at what the report calls the “basic” stage, using publicly available chatbots or ready-made tools rather than anything built into how they operate.
That is down from 73% the year before, so real progress is happening. It is happening slowly. Only 20% have reached an “intermediate” stage of integrating AI across functions, and just 13%, up from 10%, have reached what AWS calls the advanced or transformative tier. Just 19% have a formal roadmap to scale AI beyond pilots, and only 16% report having a comprehensive AI strategy at all.
The gaps compound from there. Only 21% of businesses have a formal AI governance framework. Fewer than a third, 30%, say accountability for AI decisions sits with a clearly defined person or team, and only 18% have a documented process for escalating AI-related risks. More than half, 53%, name workforce skills as the barrier stopping them from moving further, and just 34% rate their own workforce’s digital skills as good or excellent.
Next-generation AI makes the gap sharper still. Only 37% of businesses say they have even heard of agentic AI, and of that smaller group, just 6% have fully deployed it. Barely 18% of all businesses surveyed say they feel ready to adopt agentic or other next-generation AI tools when they arrive.
AWS’s own recommendations point to the same conclusion. The presentation delivered alongside the report set out four priorities for Malaysia, covering the software ecosystem, workforce skills, governance and public sector leadership. None of the four is about getting more businesses to start using AI. They are about what happens after adoption, when a business has to turn a chatbot habit into something that actually changes how it operates.
The report separately defines small and medium enterprises as businesses with fewer than 250 employees, departing from SME Corp Malaysia’s own threshold of under 200 employees or RM50 million in annual turnover. AWS says the wider definition was chosen for comparability across the Asia Pacific and Japan region, not to match Malaysia’s own classifications, a reminder that much of this research was built for regional consistency rather than Malaysian precision.
That is the quieter argument sitting inside AWS’s own research: Malaysia’s AI story has already moved past counting adopters. The number that gets the headline, though, is still the adopter count, built on a business population AWS cannot confirm is even trading. The maturity gap, the governance shortfall and the unreadiness for what comes next are the parts of this report worth paying attention to.
Whether the true adoption figure is 3.4 million or a fraction of that will matter far less to Malaysian businesses over the next year than whether they have a strategy, a governance framework and the skills to use AI for something more than routine tasks.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post AWS confirms the real story behind Malaysia’s AI adoption numbers appeared first on TechWire Asia.