On 14 September 2026, Malaysia's investment agency MIDA held Data Centre Nexus (DCN) 2026 in Kuala Lumpur. Its message was a change of emphasis. MIDA said the next phase must carry Malaysia "from building capacity to building capability, from attracting investment to creating value." The keynote, written for Investment, Trade and Industry Minister Johari Abdul Ghani, was delivered by Deputy Minister Sim Tze Tzin, according to MIDA's own release. The question for the MyDigital agenda is whether the buildout creates local capability or simply hosts foreign hyperscaler capacity.
The case for worry
The skeptics have a strong argument. MIDA reports RM385.7 billion in data-centre-related investment from 2021 to mid-2026, with AWS, Microsoft, Google, AirTrunk, DayOne and Vantage among the operators present. Large campuses are capital-intensive and need few permanent staff. The expensive equipment, such as chips, servers and often cooling systems, is typically imported. The local costs are real and visible: grid capacity, land and water. MIDA projects data centre electricity demand above 5,000 MW by 2035.
Water is the clearest example. MIDA's own analysis of Johor says the state attracted 50 data centre investments in two years, that its water reserve margin stands at 16.9% against a 25% target for 2029, and that non-revenue water loss is 26.3% (MIDA). Johor has since set a data centre water tariff of RM5.33 per cubic metre (Asia News Network, 27 August 2025). Anyone who worries that residents and local industry could end up subsidising foreign compute is pointing at a real risk, not inventing one.
What the government is actually proposing
MIDA frames four pillars: AI compute and connectivity, resource-efficient infrastructure, local supply chain and skilled jobs, and wider access to digital technology. The sharpest test of the supply-chain pillar is the business-matching session. DCN 2026 connected 14 data centre companies with 51 local vendors. The 2025 event connected 8 companies with 17 vendors. Meetings are not contracts, but the trend is in the right direction, and the number of signed contracts is the metric to track next.
Experts quoted by EdgeProp point to where Malaysian firms can realistically compete: renewable energy development, engineering, procurement and construction, battery storage, energy management, and operations and maintenance. These are credible niches. Hyperscalers cannot import civil works, power engineering or 24/7 facility operations. They do need them, in volume, near each site. MIDA chairman Tengku Zafrul put the standard plainly: success "will not be measured by the size of the investment we attract, but by the value we build around it."
Where policy should stay light
The tempting response to spillover worries is mandates, such as local-content quotas or compulsory technology transfer. Experience elsewhere suggests these raise costs and deter the investment that makes spillovers possible. A better approach uses conditions that are transparent and tied to the real public cost.
- Price scarce inputs honestly. Cost-reflective water and power tariffs, which Johor has started on water, protect households without singling out foreign firms. They also push operators toward efficient cooling, which matters as AI racks move from 5-10 kW to 70-107 kW density, per the EdgeProp report.
- Publish outcomes, not intentions. MIDA should report the vendor-matching conversions, local procurement value and Malaysian hiring figures each year. Without data, "local value" is a slogan.
- Fix the pipeline. Training programmes such as AWS re/Start, which the MIDA chairman cited, only help if graduates land in the engineering and operations roles the supply chain needs.
- Keep the access promise concrete. The pillar on AI and cloud access for SMEs and startups is the most direct route to domestic use of the capacity. Its test is whether small firms get usable, affordably priced compute rather than only a place in a brochure.
Verdict
Malaysia's data centre policy is moving in a sensible direction. It pairs openness to foreign capital with an effort to build supplier depth, rather than closing the door or ignoring the costs. The risks are strongest where physical limits bind. Water margins in Johor are thin, and grid demand is set to climb steeply. The upside is real too: a domestic base in power engineering, construction and operations that can serve the region's next buildout.
The evidence so far shows rising vendor engagement and honest official acknowledgment of resource strain. It does not yet show spillovers. A fair assessment in 2027 would ask three questions. How much of the capital spend was captured by Malaysian firms? Did resource tariffs and grid upgrades keep pace without raising household costs? Did SMEs get compute they could use? If MIDA publishes those numbers, the policy can be judged on results, and adjusted without resorting to heavy-handed mandates.