Government AI procurement in Singapore and Malaysia is the process by which public agencies buy, test and approve AI systems from vendors under national e-procurement rules, GeBIZ in Singapore and MyDIGITAL-coordinated tenders in Malaysia, while also satisfying separate AI governance requirements such as Singapore’s AI Verify testing framework and Malaysia’s incoming AI Governance Bill.
Why this matters now
Governments are not short on AI ambition. They are short on AI maturity. McKinsey’s July 15, 2026 article on public sector AI found the sector’s AI Quotient score at just 26 out of 100, the lowest of any sector it measured, against a global cross-sector average of 35.
That gap is exactly where procurement rules and AI governance requirements collide.
Singapore and Malaysia have both moved fast this year to close it, but they have done so through separate tracks: procurement reform on one side, AI governance legislation on the other. Vendors and procurement officers who treat these as one process are the ones getting stuck.
Singapore’s procurement machinery today
Singapore runs AI procurement through GeBIZ, the central e-procurement portal managed by GovTech and the Ministry of Finance. GeBIZ is widely estimated at roughly SGD 30 billion in annual procurement across IT, construction, professional services, defence and healthcare, though this figure comes from e-procurement intermediaries rather than a confirmed government-published total; the underlying award-level data is published by the Ministry of Finance via data.gov.sg. In 2026, GovTech added a self-serve layer on top of that: a products catalogue at products.tech.gov.sg, launched to address the common complaint from public officers of “not knowing where to start” when picking tools.
Governance runs on a parallel track. AI Verify, built by IMDA and now open-sourced on GitHub, gives agencies and vendors a shared way to test AI systems against international principles before those systems reach GeBIZ. It is not a procurement gate by law, but it is increasingly the reference point agencies ask vendors to speak to.
Procurement volume and governance maturity are advancing on separate clocks, and vendors have to track both.
Malaysia’s parallel build-out
Malaysia is assembling the same two tracks from a newer starting point. MyDIGITAL Corporation is running a procurement process to select a delivery partner for the National AI Action Plan 2026 to 2030, while a separate RFP is underway for the development of an AI Governance Bill, intended to make the legislation technically sound and practical for both public and private sector adoption.
Budget 2026 backs this with money: an RM53 million allocation for the Malaysia Digital Acceleration Grant to catalyse adoption of technologies including AI, and an RM18.1 million allocation for the National AI Office. In practice, teams building for Malaysian public agencies find that the governance framework a vendor will eventually be measured against is still being written even as procurement tenders for AI-adjacent work go out.
Where governance and procurement rules actually meet
This is the section both audiences skip and both later regret skipping. Deloitte’s 2026 government trends research found that governments are increasingly simplifying procurement rules first, then applying digital tools afterward, a reversal of the older pattern of digitizing already-complex processes. That sequencing matters for AI specifically, because a poorly scoped tender can lock in a system that later fails a governance test that didn’t exist when the contract was signed.
Simplify the process before you automate it, or you automate the wrong process faster.
Academic research backs this up from the vendor-oversight side. A 2025 study of U.S. city procurement found that AI systems acquired by cities often bypass conventional procurement channels entirely, which weakens oversight before the tool is ever deployed. Singapore and Malaysia are trying to avoid that exact failure mode by keeping AI-specific testing and disclosure requirements attached to the procurement record rather than treated as a side process.
What vendors get wrong
Three patterns recur. First, vendors bid on the procurement timeline and treat governance testing as a formality to complete afterward, which backfires when agencies ask for AI Verify-style evidence during evaluation, not after award. Second, vendors assume a checklist equals compliance. Research on AI procurement checklists warns that procedures built to protect marginalized groups from AI harms risk becoming performative rather than substantive in practice, the same risk applies to a vendor’s internal checklist if it isn’t backed by real testing evidence.
Third, vendors underestimate how much procurement and governance maturity varies even within one country’s own agencies. Interviews with procurement managers found generally low AI maturity across the agencies studied, with benefits still concentrated in operational efficiency rather than deeper transformation, which means the agency you’re bidding into may be less AI-ready than its national policy documents suggest.
A practical compliance sequence
| Option | Key Strength | Best Used When |
|---|---|---|
| GeBIZ standard tender (Singapore) | Fully open, transparent bidding aligned to WTO GPA rules | The requirement is a defined IT/AI system with clear specifications |
| MyDIGITAL delivery-partner RFP (Malaysia) | Direct alignment to national AI strategy priorities and funding | The engagement involves shaping policy or national-level AI infrastructure |
| AI Verify testing pathway (Singapore) | Produces an auditable governance report vendors can reuse across bids | The AI system makes decisions that affect citizens and needs demonstrable trust evidence |
Teams building for this market typically find the sequence works best in this order: scope the procurement requirement narrowly, run governance testing before submission rather than after, and keep the testing evidence attached to the tender record so it survives agency handovers.
FAQ
How does government AI procurement work in Singapore? Agencies publish AI-related tenders through GeBIZ, Singapore’s central e-procurement portal. Vendors bid through the same process as any IT contract, though agencies increasingly expect AI Verify governance evidence as part of the technical proposal.
What is AI Verify and is it mandatory? AI Verify is an open-source AI governance testing framework built by IMDA. It is voluntary by law, but is increasingly referenced in Singapore public sector procurement and financial sector guidance as evidence of responsible AI practice.
Is Malaysia’s AI Governance Bill already in force? No. As of mid-2026, MyDIGITAL Corporation is still running the RFP process to develop the Bill. Vendors should expect the governance framework to firm up over the next 12 to 24 months.
Do GeBIZ and MyDIGITAL require the same AI compliance documents? No. They are separate national systems with separate procedures. A vendor working across both markets needs distinct compliance packages, though the underlying AI testing evidence can often be reused.
What is the biggest procurement mistake vendors make with government AI deals? Treating governance testing as a post-award formality. Agencies in both markets increasingly want testing evidence during evaluation, not after the contract is signed.
Conclusion
Three things matter most here. Procurement volume and AI governance maturity are moving on separate tracks in both Singapore and Malaysia. Simplifying the underlying process before automating it consistently outperforms digitizing complexity as-is. And governance evidence needs to sit inside the procurement record, not beside it, or it disappears exactly when an agency needs it most. Where does your organisation’s AI procurement checklist actually get tested, before the bid, or after the award?
About the Author: Shivi
