The buyer is paying for control, not just compute
A Malaysian bank deciding where to place a regulated workload is not making the same purchase as a developer spinning up a test server in a global cloud region. The bank is buying latency to Malaysian users, comfort with local procurement and audit expectations, a line of accountability in Malaysian law, a familiar telco name in the contract chain, power and cooling inside a certified facility, cyber monitoring that can be explained to a board, and enough flexibility to move between private infrastructure, public cloud and managed services without rewriting the whole estate. TMONE DC is interesting because that is the product it can sell when the customer is not simply looking for the cheapest raw compute.
The value proposition is not that a national provider can outscale Amazon, Microsoft or Google. It cannot. The value proposition is that a Malaysian enterprise or public agency may still prefer to anchor critical systems in a locally owned, locally operated infrastructure environment when the workload is sensitive, latency matters, audit trails matter, and the buyer wants one domestic provider to wrap facilities, fibre, cloud, cyber and managed service into a single operating model. TM One's own data-centre page frames the basic offer as co-location, managed hosting, backup and disaster recovery, and data-centre infrastructure management in Tier III facilities with high availability and 24x7 support (https://www.tmone.com.my/data-centre-services/). TM Global's data-centre page adds the harder infrastructure claims: six strategic data centres including BFDC, KJDC, IPDC, KVDC, SJDC and HKDC; Malaysia's fibre backbone; 33 subsea cable systems; carrier-neutral interconnection; DC-to-DC links; multiple international standards; and up to nine layers of physical security (https://tmglobal.com.my/products-and-solutions/data-centre-solutions/data-centre).
Those claims give TMONE DC its operating surface. It is not only a building full of racks. It is the data-centre layer of a national connectivity group that can sell local cloud, managed security, enterprise network services, disaster recovery, cross-border interconnect and, increasingly, AI compute. Telekom Malaysia's 2025 annual-report summary put group revenue at RM11.9 billion and described more than 750,000 km of domestic fibre cables and more than 400,000 km across 39 submarine cable systems (https://tm.com.my/iar2025/). The B2B business review said TM One's 2025 revenue stood at RM2,782.8 million, with digital solutions including IT services, data centre and cloud recording quarter-on-quarter growth, partly from increased colocation demand from the banking sector and higher contributions from a global digital travel agency (https://tm.com.my/iar2025/template/pdf/Business_Review.pdf). That is the economic clue: the customer base is not only buying storage. It is buying mission-critical infrastructure where contract duration, compliance and service wrapping matter.
Malaysia's cloud market is changing quickly enough that this local premium has to be earned, not assumed. AWS has opened the Asia Pacific (Malaysia) Region and said it plans to invest RM29.2 billion in Malaysia through 2038 (https://aws.amazon.com/blogs/aws/now-open-aws-asia-pacific-malaysia-region/). Microsoft announced a US$2.2 billion Malaysia cloud and AI investment in 2024 and later described further Malaysian cloud-region expansion (https://news.microsoft.com/apac/2024/05/02/microsoft-announces-us2-2-billion-investment-to-fuel-malaysias-cloud-and-ai-transformation/). Google announced a US$2 billion Malaysia data centre and cloud region investment, with projected economic impact and job creation by 2030 (https://www.googlecloudpresscorner.com/2024-05-30-Advancing-Malaysia-Together-Google-Announces-US-2-Billion-Investment-in-Malaysia%2C-Including-First-Google-Data-Center-and-Google-Cloud-Region). These platforms bring developer ecosystems, global capital and cloud-native tooling that a domestic telco cannot simply copy. TMONE DC's opportunity is narrower but still valuable: be the Malaysian control point for workloads that need sovereign handling, low-latency local adjacency, network depth and managed operating assurance.
The core judgment is therefore conditional. TMONE DC is strongest where a customer values a Malaysian control plane more than hyperscale breadth. It is weaker where the customer only wants commodity public cloud or global AI tooling. Its economics improve if power-secure, certified, connected capacity becomes scarce in Malaysia. They deteriorate if hyperscalers make local cloud cheap enough, compliant enough and easy enough that the domestic premium loses force. The next two years will test which of those forces is stronger.
A national-cloud premium built from law, latency and procurement comfort
Data sovereignty is often discussed as if it were a slogan. For buyers, it is usually a bundle of operational anxieties. Where is the data stored? Which court, regulator or foreign policy could touch it? Who answers during an incident? Can the procurement committee explain why the workload is not sitting only inside a foreign hyperscaler? Can a bank, ministry, hospital, utility or state-linked enterprise defend the architecture to an auditor? TMONE DC's premium exists because those questions have real procurement weight in Malaysia.
The Malaysian government's MyGovCloud move is useful context. The Ministry of Finance reported in May 2022 that the Public Sector Data Centre had been upgraded into MyGovCloud, combining private cloud at PDSA with public cloud from government-appointed cloud service providers. The same notice named Amazon Web Services Malaysia, Google Cloud Malaysia, Microsoft Malaysia and Telekom Malaysia as the four cloud service provider companies involved in the Cloud Framework Agreement, with local managed service providers also participating (https://www.mof.gov.my/portal/en/news/press-citations/govt-introduces-cloud-computing-service-mygovcloud). TM later described TM One as the sole local cloud service provider for that government cloud context, arguing that in-country data centres, infrastructure and services give government buyers data sovereignty, trust, privacy and security under Malaysian laws (https://www1.tm.com.my/news/tm-one-malaysia-government-data).
That positioning should not be read as a blanket guarantee that every Malaysian workload must sit with TMONE DC. It is better read as a procurement comfort advantage. A ministry or regulated enterprise can use global cloud for many workloads and still prefer a domestic telco-owned environment for certain citizen data, sensitive applications, legacy systems or hybrid stacks. TMONE DC can sell the bridge: local data-centre anchoring, cloud migration, managed services, cybersecurity, networking and disaster-recovery design. TM One's Cloud Services page presents Cloud Alpha as a suite of infrastructure, industry-specific platforms and cloud management tools designed for hybrid and multi-cloud deployments, with advisory, consulting and managed services for cloud migration (https://www.tmone.com.my/cloud-services/). That is exactly the shape of a customer that wants cloud adoption without losing control of every dependency.
Latency is the second part of the premium. Malaysia is a large enough digital market that local user experience matters, and it sits beside Singapore's mature data-centre market without being identical to it. A Kuala Lumpur bank, Penang manufacturer, Johor logistics operator or public-sector platform serving Malaysian citizens does not want critical data paths to wander unnecessarily if a domestic facility, domestic fibre and regional interconnect can keep the path shorter and more predictable. TM Global's data-centre materials emphasize low latency, high-availability infrastructure, redundant network paths, DC-to-DC interconnect, subsea cable systems and carrier-neutral exchange access. Those statements are marketing claims, but they match the actual economics: a data centre that is close to users, close to fibre, close to cloud on-ramps and close to cross-border routes can charge for performance and risk reduction, not just square footage.
Procurement comfort is the third part. TM is not a startup asking a public agency or bank to trust an unknown brand with mission-critical workloads. It is Malaysia's incumbent connectivity group. That can be a burden if buyers view a telco as slow, legacy or less developer-friendly than a hyperscaler. It is also an advantage when the customer wants someone with local field teams, regulatory familiarity, government contract experience, fibre access and escalation channels that senior management already recognizes. In data-centre economics, the name on the contract matters because a failure is not a simple refund problem. A major outage can stop payments, customer service, hospital systems, airport processes, supply-chain platforms or government portals. Buyers pay for a provider they believe can be made accountable.
The premium is therefore not sentimental nationalism. It is a price for local enforceability, latency, auditability and operational bundling. The risk is that a premium becomes too comfortable. Hyperscalers are also localizing infrastructure in Malaysia, and they know how to meet enterprise procurement checklists. If Microsoft, Google and AWS give a Malaysian customer local regions, strong compliance tooling, superior developer experience and aggressive pricing, TMONE DC has to compete on integration and sovereign operating assurance rather than on location alone.
The facilities are becoming an AI-capacity business
The old data-centre story was server rooms, co-location and disaster recovery. The new story is power-dense AI infrastructure. TM's public materials show the shift clearly. In November 2024, TM announced that TM Global would expand the Klang Valley Data Centre in Cyberjaya and the Iskandar Puteri Data Centre in Johor, with the second phases scheduled for commercial operations in 2025 and expected to deliver about 20 MW of combined IT load. The same announcement said the expansions would support hyperscalers, OTT players, cloud providers and next-generation AI providers, and would meet Uptime Institute Tier-III standards and LEED Silver rating expectations (https://tm.com.my/news/tm_global_expands_data_centres_cyberjaya_johor). A July 2025 TM Global update said KVDC Block 2 and IPDC Block 2 had completed structural frameworks, were being developed as carrier-neutral, hyperscale-ready facilities, and were slated for Q4 2025 commercial operations with the same combined 20 MW figure (https://tmglobal.com.my/key-highlights/news-and-article/articles/events/KVDC-IPDC-Expansion).
Twenty megawatts is not a hyperscale empire when compared with the largest AI campuses being discussed globally. It is still economically meaningful in a market where certified, connected, power-secure capacity is scarce. For TMONE DC, the 20 MW expansion matters because it moves the offer from traditional enterprise co-location toward capacity that can host larger cloud, AI and content workloads. A high-value rack is no longer only a place for a bank's server. It may be a dense GPU rack, a cloud node, a storage cluster for data-heavy public services, a security analytics platform, or a content/cache workload where network adjacency matters.
TM's sovereign GPU-as-a-Service announcement makes the AI turn explicit. In December 2024, TM said its GPUaaS offering would be hosted entirely in Malaysia, use NVIDIA GPU technology at Uptime Tier-III certified data centres, provide low-latency and high-performance AI computing, and preserve data sovereignty and security (https://www.tm.com.my/news/GPUaaS_empower_ai). TM Global's GPUaaS page presents the product as sovereign AI infrastructure for Malaysia and ASEAN, with secure TM data centres, high-performance GPU clusters, ASEAN-wide connectivity, workload orchestration, autoscaling, storage and network fabric (https://tmglobal.com.my/products-and-solutions/tm-gpuaas). This is more than a new SKU. It is a test of whether a telco-owned data-centre operator can move up the stack into AI compute consumption without becoming only a landlord for other cloud providers.
AI changes the cost model. Dense GPU workloads require more power per rack, better cooling design, stronger electrical redundancy, higher floor and cabling discipline, and tighter procurement around chips, storage, high-speed networking and spares. They also create a sharper sales question: will customers pay TM for GPU consumption because the infrastructure is local, sovereign and connected, or will they use hyperscaler GPU instances because the software ecosystem is broader? TM can win certain workloads where local data residency, Malaysian public-sector adoption, nearby inference and bundled connectivity matter. It may struggle on frontier model training if customers need huge clusters, cutting-edge AI platforms and global developer tooling that hyperscalers can update faster.
That division matters. TMONE DC should not be judged by whether it becomes the Malaysian AWS. It should be judged by whether it can own a profitable slice of Malaysian AI infrastructure: sovereign inference, regulated-sector AI, public-sector AI pilots, healthcare diagnostics, local-language services, edge-adjacent workloads, enterprise GPU access and cloud-adjacent data processing. TM's GPUaaS page talks about model training, LLM inference, MLOps pipelines, data sovereignty and enterprise SLAs. The economics will depend on utilization. GPUs are expensive. If usage is steady, the gross margin can be attractive. If adoption is episodic, hardware can sit underused while depreciation, power, cooling and support costs continue.
This is why TM's telco base is both a strategic advantage and a discipline problem. Its fibre, data-centre and enterprise relationships can seed demand. Its sales teams can cross-sell GPUaaS to banks, ministries, hospitals, manufacturers and digital platforms. But AI capacity cannot be sold like ordinary connectivity. Buyers will compare frameworks, libraries, model availability, security controls, cost per inference, latency, quota certainty and migration support. TMONE DC's AI opportunity is real, but it will be won in operating detail, not in national branding alone.
The revenue model: racks, power, cross-connects, cloud wrappers and managed risk
The economics of TMONE DC sit on several layers. The first is traditional co-location: rack space, dedicated cages, private suites, power, cooling, cross-connects, remote hands, physical security and facility access. TM Global states that its facilities offer flexible rack space from single units to dedicated cages and private suites, with 42U to 48U rack configurations, standardized cabling and cross-connect options, data-centre infrastructure management, DDoS-protected data-centre internet and Metro Connect for DC-to-DC linkage (https://tmglobal.com.my/products-and-solutions/data-centre-solutions/data-centre). That is the base recurring revenue engine.
The second layer is network monetization. A pure real-estate data-centre operator sells space and power, then monetizes interconnection. A telco-owned data-centre operator can do more. It can sell IP transit, private lines, cloud connect, metro connectivity, cross-border interconnect, wholesale Ethernet, enterprise WAN, SD-WAN and managed security alongside the rack. PeeringDB's public profile for Telekom Malaysia Berhad AS4788 describes TM Network as MPLS enabled and IPv6 ready, with a selective peering policy and private-only contract requirement (https://www.peeringdb.com/net/236). This does not make AS4788 the subject; it is evidence that the parent connectivity surface is real and visible. For a Malaysian enterprise, the value of a TMONE DC rack can include the network path out of the rack.
The third layer is managed infrastructure. TM One's data-centre page lists managed hosting, backup and disaster recovery, and data-centre infrastructure management. Its cloud page describes Cloud Alpha infrastructure, cloud management and hybrid/multi-cloud models. Its cybersecurity risk-management page describes Managed SOC, Managed SIEM, CSIRT, SOAR, managed vulnerability management, digital risk protection, cybersecurity service assurance and threat intelligence advisory, including proactive 24/7 monitoring, threat detection and incident response (https://www.tmone.com.my/cybersecurity/cybersecurity-risk-management-services/). These wrappers are important because managed services can turn low-margin facility capacity into higher-retention customer relationships. A bank may negotiate hard on rack price, but it may pay more for a provider that can combine the facility, network, cyber monitoring, disaster recovery and compliance reporting.
The fourth layer is cloud and AI consumption. Cloud Alpha, GPUaaS, cloud management and private/hybrid cloud can create recurring revenue that is less tied to physical square footage and more tied to usage, performance and service outcomes. TM's annual-report B2B review said recurring revenue remained solid and was supported by longer-term service-based contracts that improve revenue visibility and reduce seasonality. It also said customers increasingly prioritize integrated, outcome-based digital solutions. That language matters because it describes the intended margin transition: away from one-off infrastructure deployment and toward service-led contracts.
The fifth layer is procurement aggregation. Malaysian enterprises may want public cloud but not want to manage every vendor directly. TM can position itself as the local integrator that handles cloud migration, connectivity, security, data-centre anchoring and operations. If it does that well, it can participate in hyperscaler growth rather than only compete with it. If it does it poorly, it becomes a pass-through channel squeezed by hyperscaler prices on one side and enterprise procurement teams on the other.
The economic challenge is that each revenue layer has a different margin and risk profile. Rack and power revenue is predictable but capital intensive. Network revenue can be sticky but faces price pressure. Managed security and cloud operations can be higher margin but require skilled labor and constant platform renewal. GPUaaS can be attractive when utilized but punishing when demand is lumpy. The best TMONE DC customer is therefore not a one-rack buyer. It is an enterprise or public agency that buys several layers together: co-location, private connectivity, cloud migration, security monitoring, backup, disaster recovery and perhaps AI compute. The more layers TM can attach, the more defensible the account becomes.
The cost base is power, cooling, land, equipment and trust
Data-centre cost discussions often start with construction cost, but the operating constraints are broader. Power is the first cost and the first growth limit. Cooling is the second, especially in a tropical climate where heat and humidity make efficient operations harder. Land is the third, because Johor, Cyberjaya, Klang Valley and Penang each carry different trade-offs around customer proximity, power availability, water, industrial zoning and cross-border routes. Equipment is the fourth, especially when AI infrastructure requires imported GPUs, switching, storage, transformers, UPS systems and cooling components. Trust is the fifth, because a certified facility with physical security, audits, cyber controls and regulatory comfort is more expensive to build and operate than a cheap server room.
Malaysia's own data-centre boom makes these costs more visible. Bank Negara Malaysia's 2025 box article on data-centre economics said approved data-centre investment amounted to RM144 billion between 2021 and the first half of 2025, that Malaysia's live capacity grew strongly between 2019 and the third quarter of 2025, and that data centres present opportunities through investment, digitalisation, GDP and jobs while also creating resource constraints and current-account pressure from specialized equipment imports (https://www.bnm.gov.my/documents/20124/19910400/qb25q3_en_box1.pdf). BNM also estimated that a 50 MW data centre would consume roughly the equivalent daily electricity usage of about 22,000 households and water usage of about 2,200 households. That is the economics behind every AI-capacity press release: growth consumes real infrastructure.
ISEAS - Yusof Ishak Institute put the pressure in sharper policy terms. Its 2025 paper said Malaysia attracted RM184.7 billion in data-centre-related investments from 2021 to 2024, that data centres are mainly concentrated in Johor and Klang Valley, and that energy consumption could surge to more than 5,000 MW by 2035, equal to 40 percent of Peninsular Malaysia's present power capacity or 11.1 percent of projected 2035 power capacity (https://www.iseas.edu.sg/wp-content/uploads/2025/05/ISEAS_Perspective_2025_43.pdf). It also noted 38 projects with Electricity Supply Agreements and 5.9 GW of secured maximum demand against only 405 MW of actual load utilization as of December 2024. The point is not that all requested demand will arrive at once. The point is that the queue for power is becoming a strategic variable.
Wood Mackenzie's 2026 Johor analysis shows the local version of the same pressure. It said Johor data centres could account for around 40 percent of the state's electricity demand by 2035, that current system-wide generation may be adequate in the near term, and that transmission and distribution constraints around clusters such as Sedenak and Nusajaya are becoming the bottleneck (https://www.woodmac.com/press-releases/jb-data-center-expansion/). This matters to TMONE DC because IPDC sits in Johor's competitive geography while KVDC sits in the Klang Valley/Cyberjaya geography. The race is not only who can announce capacity. It is who can secure the right utility path, substation capacity, cooling design and commissioning timeline.
Sustainability also has a commercial price. TM said in 2022 that KVDC, IPDC and KL City Data Centre had secured the Green Electricity Tariff from Tenaga Nasional Berhad, adding to Green Building Index and LEED certifications (https://tm.com.my/news/tm-data-centres-now-powered-by-clean-energy). The claim helps TMONE DC with multinational customers and regulated enterprises that need sustainability reporting. But green tariffs and LEED status do not remove the physical problem of growing demand. They help a provider position the energy mix and facility efficiency. They do not make dense AI racks cheap to cool or guarantee that local grids can keep pace with every proposed campus.
Land and cooling are also becoming reputation issues. Public discussion of Malaysia's data-centre boom increasingly includes water use, power prices and whether data-centre benefits justify the infrastructure burden. A broad Reddit discussion of Malaysia data-centre water concerns is not evidence about TMONE DC, but it is a useful market signal because it shows how ordinary observers frame the sector when cooling and water become household topics (https://www.reddit.com/r/malaysia/comments/1iv7ym7/explained_why_malaysias_data_center_boom_faces/). A provider that wants a national-cloud premium cannot ignore that social license. TMONE DC's image as a domestic digital-infrastructure partner depends on showing that its facilities are efficient, secure and useful to Malaysian industries, not only profitable for cloud tenants.
Competition is not only hyperscale; it is also capital discipline
The obvious competitors are hyperscalers and large regional data-centre operators. AWS, Microsoft and Google bring cloud regions, AI services, storage ecosystems, developer tooling and global procurement credibility. Equinix, NTT, STT, AirTrunk, YTL, AIMS, TIME and other operators shape customer expectations around interconnection, campus scale, carrier neutrality and wholesale capacity. Singapore spillover has made Johor especially competitive. MDEC's digital-investment article argues that Malaysia offers proximity to Singapore, lower operational costs, strong digital infrastructure, government incentives and lower land and energy costs than Singapore (https://mydigitalinvestment.gov.my/resources/articles/malaysia-the-rise-of-data-centre-hub). That is good for Malaysia, but it also attracts competitors that can bid aggressively for power, land and tenants.
The more subtle competitor is the customer's own capital discipline. A CFO may support sovereign infrastructure but still ask why the workload cannot run in a hyperscaler region on a reserved contract. A CIO may like a domestic provider but worry about developer tooling, automation and talent availability. A regulator may require resilience but not require every workload to be domestically hosted. A business unit may push for rapid AI experimentation and choose the platform where models, GPUs and APIs are easiest to access. TMONE DC wins only when it can turn local control into measurable operational value.
That means TMONE DC has to compete on bundles, not slogans. A customer might accept a premium if TM offers a certified data centre, Malaysian data residency, private connectivity, disaster recovery, security operations, hybrid-cloud management and executive accountability in one package. The same customer may reject a premium if the offer is merely a local rack with higher prices. The difference is service design.
Kenanga's 2025 data-centre connectivity report captures the telco opportunity from another angle. It argued that hyperscale data-centre growth creates demand for international, terrestrial and enterprise connectivity, and that even Tier-1 hyperscalers with their own global backbones still rely on local carriers for redundancy and resilience; smaller hyperscalers and content platforms rely even more on local access and backbone capacity (https://www.kenanga.com.my/market-insights/esg-thematic-reports/data-centre-connectivity/). This is where TMONE DC can be more than a landlord. If Malaysia's data-centre market grows, TM can earn not only from its own facilities but also from connectivity into and between other facilities, cloud regions and content platforms.
The Nxera development changes the frame again. TM and Singtel's Nxera announced plans in 2024 to develop next-generation, hyperconnected, sustainable and AI-ready data centres in Malaysia, with submarine cable connectivity between Singapore and Johor to enhance digital connectivity (https://www.tm.com.my/news/tm_singtel_nxera_develop_next_gen_data_centres). That project is not the same as TMONE DC's existing facility business, but it matters because it shows TM choosing partnership and scale for the AI-ready campus market. The strategic implication is clear: TM knows that traditional enterprise data-centre capacity alone is not enough for the next wave. It needs larger campuses, cross-border routes, hyperscale tenants and AI-density design.
Competition therefore cuts in three directions. Hyperscalers compress the value of generic cloud. Specialist data-centre operators raise the bar on facility design and scale. Customers' internal teams pressure every managed-service premium. TMONE DC's defensible lane is the intersection of domestic trust, telco network depth, certified facilities, managed cyber, public-sector familiarity and regional interconnect. It should avoid pretending that all cloud demand is sovereign demand. Much of it is not. But the sovereign slice can still be profitable if TM runs it with discipline.
The customer base: banks, public agencies, regulated enterprises and AI adopters
The strongest customer logic for TMONE DC is regulated, domestic and operationally sensitive. Banks and financial institutions need resilience, auditability, secure connectivity and clear incident response. TM Global markets compliance readiness including BNM Risk Management in Technology data-centre resiliency assessment, PCI DSS, ISO/IEC 27001, SOC 2 Type II, business continuity and environmental and energy management standards. TM's 2025 B2B review specifically noted increased data-centre colocation demand from the banking sector. That is not surprising. Banking workloads create the right combination of latency, compliance, uptime and risk aversion.
Government is the second natural base. MyGovCloud and the Cloud Framework Agreement gave cloud adoption a public-sector procurement path. TM One's sovereign-cloud language is designed for this buyer: citizen data, Malaysian law, privacy, security, trust and local operation. Government accounts can be slow and demanding, but they can also be large and sticky if the service becomes embedded in public platforms. The risk is political and budgetary. If procurement changes, if a public platform moves to a global cloud under a different framework, or if service quality disappoints, the customer base can shift.
The third base is enterprise digitalisation across manufacturing, logistics, energy, healthcare, education and travel. TM One's annual-review language mentions BFSI, manufacturing, energy, utilities, transport and logistics. These customers may not require full sovereign control for every workload, but they often need managed migration, network integration, security monitoring and predictable support. A manufacturer may want local latency to plants and suppliers. A hospital group may need data protection and imaging support. A logistics platform may need uptime and cross-border connectivity. A travel platform may need content, booking and customer-service resilience.
The fourth base is AI adopters. Here the market is less mature. TM's GPUaaS offer targets public services, healthcare diagnostics, autonomous vehicles and immersive media. These examples are plausible, but the real demand will depend on whether Malaysian organizations have data readiness, budgets, use cases and talent. AI infrastructure is not valuable just because GPUs exist. It is valuable when customers can translate compute into inference, automation, analytics or model training that improves operations. TMONE DC benefits if Malaysian customers want AI capacity that is local, secure and bundled with connectivity. It has less advantage if customers want access to the deepest global model platforms and can keep data governance under control elsewhere.
Customer dependence is therefore partly a portfolio question. A healthy TMONE DC book should not depend too heavily on one category of buyer. Banking gives compliance gravity. Public sector gives sovereign logic. Enterprise gives diversification. AI gives growth. Hyperscaler and OTT demand gives wholesale scale but can carry tougher pricing. The more balanced the book, the less exposed TM is to a single procurement cycle or technology fashion.
The available public evidence does not disclose enough customer concentration, facility-level utilization, average contract tenor or margin by service line. That limits the precision of any outside judgment. The public signals are still meaningful: TM says B2B data centre and cloud are growing; TM is expanding KVDC and IPDC; TM is selling GPUaaS; TM is marketing data-centre connectivity and cyber services as an integrated platform. Those facts point to a business that is moving from classic enterprise hosting toward a broader cloud-and-AI infrastructure role.
Non-official signals: the market believes in Malaysia, but worries about overbuild
The non-official signal around Malaysia data centres is not hard to read. Investors, analysts and trade publications see Malaysia as a major winner from Singapore spillover, AI demand and cloud-region localization. They also worry about power queues, water, underutilized commitments, imported equipment and whether local economic benefits will match headline investment numbers. That tension is exactly the environment in which TMONE DC must operate.
MIDA said at Data Centre Nexus 2025 that Malaysia had approved RM278 billion in digital investments from 2021 to 2024, with RM184.7 billion from data-centre and cloud-related projects, and that the data-centre market was projected to grow from USD4.04 billion in 2024 to USD13.57 billion by 2030 at a 22.38 percent annual growth rate (https://www.mida.gov.my/media-release/mida-powers-up-malaysias-digital-future-at-data-centre-nexus/). That is the bullish story: Malaysia becomes a regional digital hub, attracts cloud and AI capital, and develops a stronger local supply chain.
BNM's box article gives the more cautious economic story. Data centres can support GDP, services exports, jobs, supply chains and digitalization, but they also require imported capital goods, electricity, water and land. The benefits arrive over time when data centres are operational and integrated into local value chains; the costs can arrive earlier through construction, grid upgrades and equipment imports. For TMONE DC, this means growth is attractive only if utilization and local service value are real. Capacity that is announced but not filled is not strategic infrastructure. It is capital tied up in an expensive shell.
ISEAS and Tech For Good Institute both highlight the policy problem of resource allocation. The Tech For Good Institute piece, based on the same broad data-centre boom, noted 38 projects with Electricity Supply Agreements totaling 5.9 GW maximum demand and only 405 MW actual load utilization as of December 2024, and raised questions about electricity prices and sustainability (https://techforgoodinstitute.org/insights/country-spotlights/malaysias-data-centre-boom-can-growth-and-sustainability-coexist/). TransitionZero later wrote that TNB reports showed data-centre electricity consumption reaching 3 percent of total demand in the first nine months of 2025, with signed ESAs representing 7.1 GW of future demand and official projections of 7.7 GW by 2030 and 20.9 GW by 2040 (https://www.transitionzero.org/insights/malaysia-data-centres-can-clean-energy-keep-up). These numbers are not TMONE DC-specific, but they define the market's carrying capacity.
There is also a subtle market perception issue around local providers. Hyperscalers are seen as technologically deep but foreign. Domestic telcos are seen as trusted and locally embedded but sometimes slower or less cloud-native. TMONE DC's market challenge is to take the trust advantage without inheriting the slow-provider stereotype. The GPUaaS move is one attempt to show technical ambition. The facility expansion is another. Cybersecurity services are another. The credibility test will come from actual customer experience: provisioning speed, support quality, uptime, transparent pricing, skilled engineers, documentation, cloud integrations and incident handling.
Unofficial signals should not be treated as audited facts, but they matter because infrastructure markets are reputation markets. If customers believe Malaysia is power constrained, they will ask whether capacity is real. If customers believe local providers move slowly, they will test provisioning times. If customers believe hyperscalers are cheaper, they will demand evidence of total-cost value. If local communities believe data centres strain water and electricity, projects face political friction. TMONE DC's advantage is domestic trust; its risk is that domestic trust can erode if the market sees data centres as extracting scarce power without enough local benefit.
What would change the judgment
The positive case for TMONE DC would strengthen if several facts become clear. First, KVDC and IPDC expansions need to be commercially operational, filled with paying customers and tied to visible workloads rather than only announced capacity. Second, GPUaaS needs evidence of sustained utilization, named or verifiable enterprise uptake, and competitive performance against hyperscaler alternatives. Third, TM One's B2B disclosures should continue showing data centre, cloud, cybersecurity and managed services growth, ideally with clearer recurring-revenue contribution. Fourth, TM should demonstrate that its data-centre energy strategy is more than green tariff participation: efficient cooling, measurable PUE, renewable sourcing, grid coordination and transparent sustainability reporting all matter.
The negative case would strengthen if several risks show up. If new capacity slips, the AI-ready story loses timing advantage. If power or interconnection constraints delay customer deployments, the facility advantage weakens. If hyperscaler local regions satisfy most sovereignty and latency requirements at better developer convenience, TMONE DC's premium narrows. If GPUaaS utilization is shallow, the AI strategy becomes capital-heavy marketing rather than a real profit pool. If customer service is slow or pricing opaque, the domestic trust advantage becomes less valuable. If public debate turns harder against data-centre power and water consumption, expansion projects may face more scrutiny and higher mitigation costs.
The most important fact to watch is not the headline number of megawatts. It is the quality of revenue per megawatt. A megawatt sold to a thin-margin wholesale tenant is not the same as a megawatt supporting a bank's private cloud, cyber monitoring, disaster recovery, network services and AI inference. A telco-owned data-centre business should be measured by how much service revenue, customer retention and strategic account control it earns from each unit of capacity. That is where TMONE DC can outperform a pure landlord. It is also where it can underperform if sales execution is weak.
There is a broader national question behind the company question. Malaysia wants to be a digital hub, but not every data-centre investment creates the same local value. The best projects deepen local cloud adoption, strengthen regulated-sector resilience, build skills, support AI use cases, improve connectivity and create service exports. The weaker projects consume power and imported equipment while leaving limited domestic capability. TMONE DC has a credible claim to be in the first category because it is tied to a Malaysian telco, public-sector cloud, enterprise services, cyber operations, fibre and local AI infrastructure. That claim still has to be proven through execution.
The final judgment is that TMONE DC is strategically useful precisely because it is not trying to be only a hyperscaler clone. Its economic role is to sell Malaysian control around cloud and data-centre capacity: local data handling, low-latency fibre, certified facilities, cyber services, managed operations and AI capacity under a familiar domestic owner. That role is valuable in a market where sovereignty, AI and power scarcity are converging. It is also easy to overprice. The company has to keep proving that local control reduces real risk and friction for customers. If it does, TMONE DC can defend a premium in Malaysia's cloud economy. If it does not, hyperscale cloud regions and specialist campuses will turn the national-cloud story into a narrowing niche.