Note

Malaysia's AI Adoption Hit 38% — But 67% of It Is Just a Chatbot

Malaysia's AI adoption rate climbed to 38% this year, up from 27% — roughly a million more businesses picking up at least one AI tool in twelve months, bringing the total to 3.4 million, according to a new AWS-commissioned survey covered by ITbrief Asia. Read as a single headline, it's the kind of number that ends up in a minister's speech: adoption up 11 points, momentum building, Malaysia catching the wave.

Read one line further, and the story changes.

67% of that adoption is a chatbot or an off-the-shelf tool bolted onto an existing workflow. Only 19% of adopters have anything resembling a formal strategy for scaling AI across more than one function. And 57% of businesses using AI at all are doing it entirely through an external provider, consultant, or vendor — meaning the capability doesn't live inside the company, it's rented for the duration of the engagement.

None of that shows up in the 38% figure. It's the difference between a company that has used AI and a company that has built something on it, and right now almost seven in ten adopters are firmly in the first camp.

Why the gap matters more than the number

An adoption rate is a vanity metric until you ask what's actually being adopted. A chatbot answering FAQs and an agentic system reconciling a bank's AML queue are both technically "AI," and both count equally toward that 38%. But one changes how a P&L behaves and the other changes how a support inbox looks.

The maturity breakdown is the more honest picture:

• 67% of adopters are at the basic-tool stage — chatbots, ready-made assistants, point solutions with no integration into core processes.
• 19% have a formal strategy for scaling AI across multiple functions.
• 57% access AI primarily through external providers rather than building internal capability.

If you're a CFO reading a board pack that cites "38% of Malaysian businesses now use AI" as a benchmark, the first question isn't whether your company is above or below that line. It's which of those three buckets your company actually sits in — because the line itself doesn't distinguish between them.

The sector split tells its own story

Adoption isn't even across the economy, and the gap between leaders and laggards is wider than the headline suggests.

Financial services is out front: 53% adoption, with 64% of adopters past the pure-experimentation stage, 42% running a formal scaling strategy, and 39% reporting an actual AI governance framework. Banking has the regulatory pressure, the transaction volume, and the compliance cost base to make the business case obvious — and it shows.

Manufacturing sits at 50% adoption but is materially less mature: 57% of adopters are still in exploration or experimentation, only 15% have a formal strategy, and just 13% say they feel prepared for the next generation of AI capability. The barriers cited are concrete rather than philosophical — 44% point to technical or data issues, 41% to workforce capability gaps, 37% to limited resources. This is a sector that wants to move and is running into infrastructure, not appetite.

Both numbers sit well above the 38% national average — which means somewhere below them, in the country's much larger base of mid-sized and small businesses, adoption is considerably lower than the headline implies. That's not incidental. Micro, small, and medium enterprises are 97% of Malaysian business establishments and close to half the workforce. A national adoption average that's pulled up by two well-resourced, well-regulated sectors is a different story than one broad-based across the economy — and it's the story most coverage of this survey will skip.

The governance gap is the real story underneath the adoption gap

Here's the number that should worry a risk committee more than the maturity gap: only 30% of businesses using AI have defined clear accountability for their AI initiatives. 27% conduct regular monitoring or audits. Just 18% have a documented escalation process for when something goes wrong.

Put differently — for every ten Malaysian businesses now running AI in some form, roughly seven don't have a clear answer to "who owns this if it fails," and eight don't have a documented process for what happens when it does. That gap sits underneath the adoption number regardless of sector, and it's the one a board should be asking about directly, not inferring from a chatbot deployment that looks fine on the surface.

47% of businesses say what they actually want is a practical framework or template to work from — not more encouragement to adopt, but a structure for governing what they've already adopted.

There's also a public-sector angle worth noting: 72% of businesses say greater public-sector AI adoption would encourage their own use, and 74% say government leadership matters for broader acceleration. With Bank Negara's AI discussion paper working through consultation and AI Malaysia Berhad's National AI Action Plan now targeting an AI governance bill by the end of 2026, that expectation is likely to be tested within the next twelve months — which makes this a good year to have your own governance answer ready before a regulator asks for it, not after.

What this means for the next AI conversation in your boardroom

Before the 38% figure gets cited again — in a strategy deck, a competitor benchmark, an investor question — it's worth having answers to three things:

1. Which bucket are we actually in? Basic tool use, formal multi-function strategy, or fully vendor-dependent — the adoption number doesn't tell you, and the honest answer changes what "keeping up" should mean for your budget next year.
2. Who owns this if it goes wrong? With 70% of adopters nationally lacking clear accountability, "we use AI" and "we govern AI" are not the same claim, and a board should know which one it's making.
3. Is our AI spend building capability or renting it? If the majority of your AI use runs through an external vendor with no internal build-up, that's not necessarily wrong — but it's a different financial and strategic commitment than the language of "adoption" usually implies, and it's worth being explicit about which one you've chosen.

This is the same gap The Missing Middle opens with: Malaysia's AI conversation tends to be framed around large, well-resourced enterprises, while the mid-market — the segment where most of the country's actual economic activity sits — is where adoption is thinnest and the strategy question is least resolved. A national adoption rate of 38% is real progress. It's also, on inspection, mostly still the easy part.

Sources: AWS-commissioned Malaysia AI adoption survey via ITbrief Asia; DOSM/SME Corp Malaysia MSME statistics, as cited in The Missing Middle.

← All notes