On 28 September, Digital Minister Gobind Singh Deo said his ministry has submitted Budget 2027 proposals on AI affordability, access and adoption for micro, small and medium enterprises. He did not say what is in them; that sits with the Ministry of Finance until the Budget is tabled on 9 October.
We think preparing now beats waiting, not because we know what will be announced, but because recent incentives show who tends to benefit.
Who Is Asking for What
Industry groups have lined up behind the ministry. The Federation of Malaysian Manufacturers wants a RM1.5 billion smart-manufacturing package for 2027–2030, with RM250 million for AI. PwC Malaysia has proposed extending the current AI-training tax deduction to all businesses and widening it to AI adoption spending. ACCCIM wants a higher ceiling on the digitalisation grant.
Meanwhile, MIDA reported that data centres and cloud took RM95.8 billion of the RM218.5 billion approved in the first half of 2026. Capacity is arriving; the question is how smaller firms will use it.
Broad Adoption, Shallow Use
An AWS-commissioned study by Strand Partners, published in August, found 38% of Malaysian businesses now consistently use at least one AI tool, up from 27% a year earlier. But 67% of those adopters use AI only for basic tasks, and 19% have a strategy for scaling it. Among manufacturers, 44% cited technical or data barriers. For a CFO, what counts is AI running inside a finance process, and most firms are not there yet.
What Budget 2026 Tells Us
Budget 2026 is the best guide to what may come:
| Measure | What it offered | Key conditions |
|---|---|---|
| AI and cybersecurity training deduction | 50% further tax deduction for MSMEs | Training endorsed under MyMahir NAICI; tax advisers report it applies to HRD Corp contributors for 2026–2027 |
| Accelerated capital allowance | Faster write-off of ICT equipment, software and custom-software consultation fees | Spend from 11 October 2025 to 31 December 2026; disposal within two years may trigger a clawback |
| Geran Digital PMKS MADANI | 50% matching grant, up to RM5,000 | Approved solution panel; confirm the current window with the administering agency |
| DFI financing | “Nearly RM1 billion” via BPMB, SME Bank and EXIM | Application details are not consistently published; check with each institution |
These measures were largely tax-based, time-limited and gated by certification, panels or levy status. Tax-based incentives also help only firms that pay tax, so loss-making or break-even SMEs gain little from them. With Kenanga projecting a 2027 deficit of around 3.5% of GDP, Budget 2027 will likely follow a similar pattern, and such incentives are harder to use for firms still deciding what they want AI to do.
One deadline already exists unless Budget 2027 extends it: the allowance for ICT and software spending closes on 31 December 2026, about twelve weeks after tabling. Whether AI spending qualifies depends on its structure (a subscription may be treated as revenue expenditure, not capital), so check with your tax adviser.
We found no published figures on how much of the 2026 AI incentives was claimed. The AI Governance Bill, due by the first quarter of 2027, is expected to place obligations on businesses that deploy AI; its final form is not public.
Our Perspective: The Bottleneck Is Usually the Data
From a BI and data-platform standpoint, the constraint that tends to decide an AI project's pace is not access to models. It is whether the financial data underneath is usable: a consistent chart of accounts, subledgers that reconcile, and an accounting system that exports more than spreadsheets.
A practical readiness test: could you give an AI analytics tool what it needs within a week? That means read access to your accounting system or a clean export, two to three years of history, a mapping from your chart of accounts to management categories, and a named person in finance to validate the first outputs. If yes, you can move quickly on whatever is announced. If not, fix it regardless of incentives.
Governance matters too: the same study found 30% of AI adopters had no defined accountability for AI. Naming an owner now costs little and prepares you for both incentive paperwork and the coming law.
Before 9 October
- Confirm eligibility: MSME status and HRD Corp contributor status.
- Map planned ICT and software spending against 31 December 2026, and confirm its tax treatment with your adviser.
- Pick one process and measure it today: days to close, days to produce a cash position, hours on receivables follow-up.
- Get an itemised quote separating licence, implementation and training.
- Run the data-readiness test above, and confirm your e-invoicing position (LHDN now exempts turnover below RM3 million).
- Ask vendors where data is stored and who can see it, and assign an internal owner.
- Set a decision rule before the speech, for example: “If the AI deduction is widened to adoption spending, we sign within four weeks.”
Should You Act Now?
Experiment. Start a bounded pilot on one finance or operations process, with the eligibility work done in parallel. Monitor the AI Governance Bill and any new grant mechanics before committing to larger programmes.
Conclusion
Recent incentives have been short-lived, tax-based and conditional. The preparation that makes a firm ready to claim them — clean data, a scoped use case, a named owner — is also what the governance law is likely to ask for. That work pays off whatever is announced on 9 October.
Working with Mandrill Tech
If your finance team still waits on spreadsheets to answer basic questions about cash, receivables and margins, Lestar AI CFO Assistant is a scoped place to start. It connects to your accounting system, maps your chart of accounts with finance validation, and gives management self-service views of growth, margins and liquidity, with sensitive names masked before analysis. Onboarding can take as little as two weeks, and packages start from RM2,500 a month, so the cost is known before you decide. Talk to Mandrill Tech.
