More Malaysian businesses use AI. Few have gone deep.
Two years ago, the question for most Malaysian business owners was whether to try AI at all. In 2026 that question has an answer. Most have tried it, and a growing share use it every week. The harder question now is what comes after the first chatbot.
This piece pulls together the most useful public data on AI adoption in Malaysia from the past year: the AWS survey released in August 2026, Xero's survey of small businesses, the national AI action plan and the Budget 2026 incentives. Read together, they describe adoption that is broad and shallow. The opportunity for the next two years sits in the depth.
The headline numbers
Amazon Web Services' Unlocking Malaysia's AI Potential 2026 study, reported in August 2026, found that over 38% of Malaysian businesses now use at least one AI tool on a consistent basis, up from 27% in 2025. AWS estimates that more than 3.4 million businesses now use AI, with one million joining in the past year.
Xero's survey of 1,033 Malaysian micro, small and medium businesses, published in November 2025, puts the figure much higher: 81% say they have implemented AI in some form.
The gap between 38% and 81% comes from the definitions. AWS counts businesses that use a tool consistently. Xero counts any implementation, which includes an owner who drafts emails with ChatGPT twice a month. Both are true. Together they say that most small businesses have touched AI, and well under half have made it part of how they work.
Where the depth runs out
The AWS study also measured what businesses do with AI once they adopt it, and those figures carry the bigger story:
- 67% of adopters focus on basic applications, such as public chatbots and general-purpose assistants.
- 19% have a formal strategy for scaling AI across more than one function.
- 24% have a formal AI governance framework.
- 57% source their AI capabilities mainly through external providers, and 69% say local software providers and AI developers are important to them.
Xero's data points the same way from the owner's side. 82% of respondents say they need more education to deploy AI with confidence, and only 56% say they are familiar with the different ways AI can be used in a business. Among those already using AI, 30% have no policy governing its use. Asked what would help, 61% chose training and education, and 50% chose advisory and consulting support.
That last pair of numbers matters for anyone planning an AI budget. Business owners are telling surveyors that the bottleneck is knowing where AI fits and how to run it without new risk. Access to tools is no longer the constraint.
Financial services is running ahead
One sector breaks the pattern. In the AWS study, 64% of financial services businesses have moved beyond experimentation, 42% have a scaling strategy and 39% have a formal governance framework. Banks and insurers operate under regulators who expect documented risk controls, so governance arrives early by necessity. Other sectors can borrow their approach without waiting for a regulator to require it. We look at design for regulated AI in AI in financial services: designing compliant, trustworthy products.
The policy tailwind: AI Nation 2030
The government's National AI Action Plan 2026-2030, branded AI Nation 2030, sets the direction for the rest of the decade. According to the National AI Office's fact sheet, the plan contains 28 initiatives: 14 sector-focused Impact Engines and 14 Foundation Enablers. The plan document sets a 2030 goal of placing Malaysia among the top 10 in recognised global AI indices, with up to 1.2 percentage points of annual GDP growth attributable to AI.
For business owners, one initiative stands out. Initiative I10, AI for MSMEs, builds on MDEC's Business Digitalisation Initiative to create what the plan calls a one-stop AI enablement ecosystem. The plan names talent shortages and cost as barriers, along with uncertainty about where to start. It aims to give 1.5 million MSMEs access to AI through tools embedded in platforms they already use, backed by training and advisory support. The Ministry of Digital leads it with MDEC.
The plan's own diagnosis matches the survey data. Awareness is high and adoption is limited. The missing piece is knowing where to begin.
Budget 2026 and training funds
Two funding routes make training cheaper for Malaysian employers in 2026.
The Budget 2026 tax deduction. The Star reported that SMEs can claim a 50% tax deduction on AI and cybersecurity training recognised by the MyMahir National AI Council for Industry (NAICI), which TalentCorp and MyDIGITAL lead. The announcement did not set out caps or claim periods, so check eligibility with your tax adviser or TalentCorp before you plan around it.
HRD Corp levy. Employers who pay the HRD Corp levy can use it for AI courses run by registered training providers, with approval obtained before the course starts. The AI programmes I teach as faculty at AiTraining2U are HRDC claimable through AiTraining2U, an HRD Corp registered training provider. AiTraining2U publishes a step-by-step claim guide, and we cover what good corporate training looks like in HRDC corporate training for AI and design.
A four-rung ladder for AI maturity
The numbers above sort Malaysian businesses into four rough stages. Place your company on one of them before you plan the next step.
Rung 1: Individual use
Staff use ChatGPT, Gemini, Copilot or Claude on their own, with no shared practice. Xero's broad definition of "implemented" counts this stage. The value is uneven, and nobody measures it.
Rung 2: A shared tool
The company buys a team licence or puts a chatbot on its website or WhatsApp line. AWS's 67% on basic applications spans this rung and the one below it. The tool exists, and nobody owns its quality.
Rung 3: One embedded workflow
AI does a defined job inside one process, such as triaging support tickets, drafting quotations or reconciling invoices. The workflow has an owner and a baseline to measure against. Someone catches wrong answers before a customer sees them.
Rung 4: Scaled with governance
AI runs across several functions under a written strategy and a governance framework. AWS puts 19% of businesses at the strategy step and 24% at the governance step. Financial services leads the way here.
The jump from rung 2 to rung 3 is the one that pays. It means choosing one workflow and designing it around the people who use it. Then you measure it. The national plan's MSME initiative talks about moving businesses beyond basic digitisation, and the training and advice owners ask for in the Xero survey point at the same step.
What to do in the next quarter
If your business sits on rung 1 or 2, these moves get you to rung 3 without a large budget:
- Pick one workflow with a clear cost. Look for a task that repeats every day and has a known error rate or turnaround time. The best candidates also frustrate the people who do them.
- Record a baseline before you touch it. Measure time per task, error rate or response time now. Without a baseline you cannot tell a success from a nice demo.
- Write a one-page AI policy. Cover which tools staff may use and what data may go into them. Name the person who approves new uses. This closes the gap Xero found in 30% of adopters.
- Train the people who will own the workflow, before you buy anything large. Hands-on training shows your team what the tools can do on your own data.
- Decide build, buy or partner with the workflow and baseline in hand. With 57% of businesses relying on outside providers, choose one that will show you how the system performs in the months after launch.
We go further into steps 2 and 5 in AI pilot to production: scaling user-centred AI.
The next two years belong to depth
Access to AI has spread across Malaysian business in two years. Depth is the next frontier: AI that does one job well, with staff who trust it and a person who measures it. A company that climbs one rung in 2026 will know more about its own operations than a competitor still running a chatbot nobody owns.
If you want help choosing that first workflow and designing it around your team and customers, our consulting services are built for businesses at this stage, or you can contact us. If your team needs the skills first, start with our training programmes.