Malaysia Limits Non-AI Data Centres; Grid and Water Supply Remain Sufficient, Says PM
Malaysia has restricted approvals for new data centres that are not tied to artificial intelligence for nearly two years to ease pressure on electricity and water systems. Applications linked to genuine high-technology and AI work will continue to get priority review.
The Prime Minister noted a clear rise in energy and water usage as data centres come online, but current projections point to sufficient national supply. The goal: protect domestic consumers while keeping room for high-value digital investment.
What This Means for Government Decision-Makers
- Triage new proposals: Prioritise AI-related facilities that add technological value; pause or reject non-AI data centre applications per current policy direction.
- Embed utility impact checks: Require clear electricity and water demand profiles, efficiency plans, and peak-load mitigation in every submission.
- Align approvals with public interest: Ensure data centre growth does not trigger tariff pressure or local supply disruptions.
- Strengthen transparency: Mandate ongoing reporting on energy use and water consumption for approved sites.
Grid Strategy: Regional Links and East-to-West Supply
The government will lean on regional and domestic interconnections to secure long-term capacity. This includes the ASEAN Power Grid and plans to channel supply from Sarawak to Peninsular Malaysia, with potential flows to Singapore.
For context, see the ASEAN Power Grid overview via the ASEAN Centre for Energy. ASEAN Power Grid
AI Governance Bill: Ethics First, Risk-Based Oversight
The upcoming AI Governance Bill will emphasise ethics, citizens' rights, and risk management. This contrasts with approaches elsewhere that lean heavily on legal formalities without centring moral considerations.
- Risk and classification framework to calibrate requirements to the level of risk.
- AI harms and incident reporting to enable systematic monitoring and response.
- Coverage across the full AI lifecycle: development, training, deployment, monitoring, and risk management.
- Clear accountability for both developers and deployers of AI systems.
- Existing laws continue to apply and be enforceable.
The bill is in early drafting at the Digital Ministry and will go through stakeholder engagement, including review by a special parliamentary committee, before finalisation.
Copyright and IP: Enforcement Will Matter
The Prime Minister underscored the need for strong enforcement by the Intellectual Property Corporation of Malaysia, including around materials used to train AI systems and outputs generated by them. Protection is already available under the Copyright Act 1987 and should be applied effectively.
Agency and legal teams can track updates here: Intellectual Property Corporation of Malaysia (MyIPO)
Action Checklist for Ministries, Agencies, and GLCs
- Review all pending data centre proposals: verify AI relevance, utility impacts, and compliance with the current restriction on non-AI facilities.
- Coordinate early with utility providers on capacity, peak management, and water sourcing before issuing any approvals.
- Stand up an AI risk register and incident reporting process aligned to the forthcoming bill.
- Tighten procurement and grant terms: require disclosure of training data sources, IP warranties, and indemnities for AI solutions.
- Prepare inputs for stakeholder sessions: ethics guardrails, public interest tests, sector-specific risks, and enforcement needs.
- Plan for interconnection opportunities: factor regional power imports and Sarawak-to-Peninsula supply into medium-term energy planning.
If you're contributing to the bill or leading policy implementation, this resource can help accelerate your framework work: AI Learning Path for Policy Makers
Your membership also unlocks:
