Smart Cities / AI news for Malaysia
From the archive · Event date 18 September 2025
Malaysia signed four AI city collaborations. Here is what each one was meant to deliver
The agreements covered training, cloud and analytics infrastructure, common city frameworks and climate prediction. Signing documents started the work; it did not prove that any city service had improved.
In brief
- MDEC said three memorandums of understanding and one letter of collaboration were exchanged at Smart City Expo Kuala Lumpur on 18 September 2025.[1][2]
- The four tracks covered public-sector training with Ericsson, a national AI Cities platform with AWS, frameworks and standards with MSCA, and a climate-AI testbed with Dell Technologies in Penang.[1][2]
- The announcements described intended scope but did not publish budgets, delivery deadlines, platform architecture, testbed measures or evidence of improved public services.[1][2]
Four agreements created a layered AI Cities plan; delivery evidence had to follow
Malaysia used Smart City Expo Kuala Lumpur 2025 to announce four separate collaborations under its AI Cities agenda. Digital Minister Gobind Singh Deo witnessed three memorandums of understanding and one letter of collaboration at the Future Citizen Stage on 18 September 2025.[1][2][3]
The announcements were not four versions of the same project. One focused on training council and public-agency staff. Another proposed a national platform for real-time operations and prediction. A third covered frameworks and standards. The fourth proposed a climate-prediction testbed in Penang's UNESCO World Heritage Zone.[1][2]
Together, they described the ingredients of an AI-enabled city programme. But a memorandum records an intention to collaborate, not a completed deployment. The public releases did not provide budgets, delivery milestones, data-governance details or outcome reports, so each initiative still needed a visible route from announcement to service.[1][2]
The four collaborations addressed skills, infrastructure, rules and one testbed
DNB and Ericsson's 21st Century Technologies Education Programme was the people layer. MDEC said the four-hour course would give civil servants in councils and public agencies practical knowledge of AI, 5G and the Internet of Things. The announcement did not state a participant target, completion schedule or assessment method.[1][2]
The DNB-AWS letter of collaboration was the platform layer. It proposed combining cloud infrastructure, machine learning and advanced analytics for real-time urban operations and predictive decisions. A useful public specification would still need to explain participating cities, data sources, access controls, procurement, uptime, model review and how local systems connect to the national platform.[1][2]
The DNB-MDEC-MSCA agreement was the rules layer. It aimed to develop frameworks and standards for mobility, digital citizen services and sustainable urban solutions. The MDEC-Dell agreement supplied a place-based use case: a climate-AI prediction testbed for Penang's heritage zone, intended to support cultural protection and environmental resilience.[1][2]
Human-centred AI needs an operating definition, not only a theme
MDEC framed the collaborations around inclusive, people-centred transformation. That is a useful direction, but it becomes testable only when a programme names the people affected, the service problem, the decision the AI will influence and the remedy available when the system is wrong.[1]
For a traffic system, human-centred could mean shorter and more predictable journeys without shifting congestion into a nearby neighbourhood. For a digital council service, it could mean faster resolution while keeping an accessible non-digital route. For a climate model in a heritage area, it could mean earlier warnings that conservation teams can verify and act on.[1][2]
The standards collaboration is therefore important because separate vendors and councils need common definitions for data quality, security, interoperability, human review and reporting. Without shared requirements, each pilot can produce a dashboard that looks successful while using different measures and leaving residents unable to compare results.[1][2]
The next proof should be a public delivery map for each initiative
The four announcements can be followed with a compact delivery register. For training, it should show target agencies, enrolment, completion and assessed capability. For the platform, it should name participating councils, technical scope, go-live stages and service-level measures. For the standards, it should publish drafts, consultation dates and adoption status.[1][2]
The Penang testbed needs its own scientific and public-service measures. Those could include the climate hazard being predicted, data coverage, forecast horizon, error rate, who receives an alert, what action follows and whether the action reduces damage or improves conservation work. A model score alone would not establish public value.[1]
Publishing that delivery map would also reduce a common smart-city problem: pilots that remain isolated demonstrations. Councils could see what is reusable, suppliers could design against common requirements and residents could judge whether the programme improved a service rather than merely installing technology.[1][2]
Why Malaysia should care
The four collaborations split Malaysia's AI Cities ambition into useful layers: people who can run systems, a platform that can process urban data, shared rules for deployment and one place-based testbed. Malaysians still need public milestones and outcome measures to know whether those layers become better city services.
Local councils
The programme could offer shared skills, infrastructure and standards instead of forcing every council to start from zero.[1]
Practical move: Name one service problem, accountable owner, baseline, data source and public outcome before joining a pilot.
Malaysian residents
AI Cities should improve a recognisable service and preserve a fair way to question or correct automated decisions.[1][2]
Practical move: Look for a public service measure, human-review route and evidence that benefits are not limited to one neighbourhood or user group.
Technology providers
Common standards and a national platform could lower integration friction, but only if requirements and interfaces are published.[1]
Practical move: Build against defined interoperability, security, audit and outcome requirements, and report limitations with performance.
What Malaysians can do now
- Track the four collaborations separately because each has a different owner and success measure.
- Ask for budgets, milestones, participating agencies and public reporting dates.
- Judge a city AI project by service outcomes, reliability and remedy—not by the signing or dashboard.
What we still do not know
The scope was public, while the implementation plan remained incomplete.
- Budget, contract value, delivery deadline and accountable programme owner for each collaboration.
- Which councils and public agencies joined the training, platform and standards work.
- The Penang climate model's hazard, data, accuracy target, operational user and conservation outcome.
Sources
- 1.Malaysia shapes human-centred AI Cities through four landmark collaborations Malaysia Digital Economy Corporation, 18 September 2025
- 2.Malaysia shapes human-centred AI cities through four collaborations Digital News Asia, 18 September 2025
- 3.Malaysia Shapes Human-Centred AI Cities Through Four Landmark Collaborations Malaysiakini announcement, 19 September 2025
