Digital Infrastructure Bill

Resumption of Debate on Question [6 October 2026],

"That the Bill be now read a Second Time." – [Minister for Digital Development and Information].

Question again proposed.

Mr Speaker: Mr Lee Hong Chuang.

12.00 pm

Mr Lee Hong Chuang (Jurong East-Bukit Batok): Mr Speaker, I rise to support the Digital Infrastructure Bill. Before I begin, I would like to declare for the 28 years before I was elected into the House, I worked in the information technology (IT) industry, including with major multinational corporations (MNCs), such as Hewlett-Packard, IBM and Kyndryl.

For most of my professional life, I was on the other side of legislation like this. During those years, I managed and worked closely with teams responsible and that provide professional services for disaster recovery, crisis management, business continuity, data centre operation, resilience services, and other critical technology and infrastructure functions. I was involved in preparing organisations for disruptions, managing incidents when they occurred and ensuring that critical systems and services could continue operating under challenging circumstances. I have seen first-hand what it means when critical infrastructure is disrupted, the urgency of the situation, the operational decisions that have to be made and the real impact on businesses, organisations and the people who depend on these services and infrastructure.

I am no longer with any of these MNCs but those years in the technology industry gave me valuable experience of the operational realities behind the systems and infrastructure we are discussing here today.

Mr Speaker, many Members have spoken on energy, water and land. I would like to add a different perspective. I want to speak not about the operators whom the Ministry has consulted extensively, but about the people at the far end of the wire, whose experiences and concerns may not always be represented in the room.

In recent years, we have seen three major disruptions. In February 2023, a power surge tripped cooling systems and took down a cloud service for hours. In October 2023, a cooling failure at one data centre disrupted online banking at our major banks for more than 12 hours, 810,000 failed logins, 2.5 million failed payment and automated teller machine (ATM) transactions. In September 2024, a fire at a data centre took cloud services down with it.

Not one of these was a cyberattack. They were physical operational failures. And the Bill knows it.

Clause 15B now requires licensees to minimise the likelihood of unexpected interruptions. Clause 15C requires plans to recover from them. Resilience before the event and after it. I welcome both.

But the Bill's notification duties point in only one direction. Clause 15D requires a licensee to report cybersecurity incidents and service delivery disruptions to the Infocomm Media Development Authority (IMDA). Clause 17(3) lets IMDA direct a licensee to notify its users, but only of a prescribed cybersecurity threat or incident.

Mr Speaker, put those clauses beside the earlier three examples I mentioned. If the October 2023 cooling failure happened again tomorrow, under this Act, the operator must tell IMDA. IMDA would have no power to make the operator inform anyone else. The Bill's power to inform the public stops exactly where the Bill's own evident rights.

And the people never contract with the licensee. The chain runs from data centre, to bank or platform, to the people. The Bill regulates the first link and leaves the rest of the commercial terms. So, the operator has no duty to tell the bank, the bank cannot tell its customers anything useful and the people stand at the counter with a screen that says "Try again later". Mr Speaker, allow me to share a bit in Mandarin here.

(In Mandarin): When there is a problem with the power or water supply, people usually know whom to contact for help. But when digital systems or online services are disrupted, many people do not know who to turn to or how long they will have to wait.

For small businesses that rely on these systems for their day-to-day operations, the impact is often immediate. Customers may leave one by one, business may come to a standstill, and the business owner may not even be able to give customers a simple answer: how much longer will they have to wait before the service is restored?

That is why when we discuss these infrastructure systems, we must also remember the real people whose lives, work and businesses depend on them. They are often the very last to know what has happened.

This Bill requires operators to notify the authorities, and that is the right approach. But notifying the authorities is not the same as informing the public. If digital infrastructure is truly as essential as electricity and water, then when an interruption occurs, people should be told what has happened and when service is expected to be restored – just as they would be during an interruption to their power or water supply. I hope the Senior Minister of State can provide greater clarity on this point.

( In English ): Let me continue my speech in English. I have three requests here.

First, a drafting alignment. Extend clause 17(3) so that the power to direct user notification also covers a "service delivery disruption", the phrase the Bill already uses in clause 15D. Thus, no new concept enters the Act, the Bill's powers simply catch up with the Bill's evidence.

Second, a commitment on the Codes of Practice. Clause 16(1)(a)(iv) lets IMDA set how quickly a licensee must report to IMDA. Will the same code set minimum standards for what a licensee tells its own customers, such as scope, estimated time to restore and a named point of contact so that the bank in the middle of the chain has something concrete to pass on?

Third, transparency after the fact. Will IMDA publish anonymised summaries of major disruptions? It lets the smaller operators who were not in the consultation room learn from the ones who were, and it tells the public that when their phone went dark, someone in the Government saw it and acted on it.

I anticipate the reply that customer communication belongs to contracts and service agreements. With respect, Sir, this Bill has already crossed that line. Clause 17(3) asks Parliament to accept that IMDA may compel a licensee to speak to its users. The principle is settled today. The only question is why it stops at cyber when every incident the Ministry cited to us was not.

If cloud and data centres are now the utilities of the digital economy, the person at the end of the wire deserves to know when they fail and when they will be back. This Bill has built the machinery. I ask the Senior Minister of State to let it reach all the way to the people who depend on this infrastructure and these services. With these requests, I support the Bill.

Mr Speaker: Mr Edward Chia.

12.09 pm

Mr Edward Chia Bing Hui (Holland-Bukit Timah): Mr Speaker, I rise in support of the Digital Infrastructure Bill.

Digital infrastructure has become fundamental infrastructure for a modern economy. Cloud services and data centres support services that Singaporeans and businesses use every day, including digital payments, ride hailing, business management applications and generative artificial intelligence (AI) services.

Singapore's digital economy accounted for 19.3% of our gross domestic product (GDP), or $144.1 billion, in 2025 – double its size since 2018. Singapore also has more than 1.6 gigawatts of data centre capacity. As our dependence on digital infrastructure grows, so do the consequences of disruption. At the same time, data centres draw on scarce resources, including electricity, water, land and our national carbon budget.

This Bill therefore addresses two important objectives: one, strengthening the security and resilience of foundational digital infrastructure and two, progressively raising the energy and water efficiency of our data centres.

I would like to raise four broad considerations.

First, we should turn higher standards into a competitive advantage for Singapore. Singapore need not compete on price alone. Our competitive proposition should be to offer trusted, resilient, connected and resource efficient compute, supported by a predictable regulatory environment. Scarcity and resilience can themselves be economic advantages. Companies handling important workloads will value confidence that their digital infrastructure is subject to credible cybersecurity, operational resilience and business continuity requirements.

Regulatory quality should be another part of our competitive proposition. For companies making substantial, long-term infrastructure commitments, regulatory consistency has real economic value. Clear rules, coordination between agencies, predictable implementation and visibility over the future direction of regulation all reduce investment uncertainty. Higher standards therefore do not necessarily make Singapore less competitive. If businesses understand the standards, how they will evolve and how they will be applied, our regulatory environment can itself become part of Singapore's competitive moat.

There are, however, several areas where implementation guidance will be important.

The first concerns the requirement for an applicant for a data centre licence to have at least one officer with the required connection to Singapore. The Bill defines an officer as a director or another person responsible for the management of the applicant. Global operators may rotate leadership and management responsibilities across jurisdictions. Companies therefore need practical guidance on which roles qualify, how ordinary residence will be assessed and how temporary vacancies or leadership transitions should be handled. Can the Senior Minister of State clarify what roles will satisfy the local officer requirement and whether IMDA will provide a reasonable process for handling leadership transitions?

The second issue concerns who is regarded as operating a data centre. Under clause 23, this turns on who controls the introduction and implementation of operating, maintenance or environmental policies. Where more than one person exercises such control, the person with the greatest authority is treated as the operator. That principle is understandable, but its application may be less straightforward in complex corporate structures. A global parent may determine group-wide environmental or operating policies while a Singapore subsidiary employs the site team and implements those policies locally. This distinction determines which entity must hold the licence.

Clause 54 may also extend liability to officers or persons involved in management in specified circumstances, including where they knowingly participate in an offence or fail to take reasonable steps despite knowing that an offence would be committed. The regulatory guidance should therefore include working examples covering common parent company, subsidiary, owner, tenant and outsourced management arrangements. Can the Senior Minister of State clarify how IMDA will identify the operator where policy authority and day-to-day operational control are divided between different entities?

A third issue is the time taken to decide on licence applications. The Bill gives existing operators six months after commencement to apply and allows them to continue operating while IMDA considers the application. This transitional protection is helpful, but companies must also plan capital expenditure, hiring, financing and contractual commitments.

I recommend that IMDA publish indicative service standards for licence decisions, distinguishing straightforward applications from more complex cases. The guidance should explain when the assessment clock starts, when it may be paused for further information and how applicants will be kept informed of the progress.

Regulatory streamlining will also be important. The Government has indicated where cybersecurity incident reporting requirements overlap, licensees will report to IMDA, which will then share the relevant information with the Cyber Security Agency of Singapore (CSA).

The same principle should be applied to sustainability reporting. Operators should, as far as possible, be able to use one underlying dataset for reporting under the Digital Infrastructure Act – existing carbon reporting frameworks, the Building and Construction Authority (BCA)-IMDA Green Mark scheme and other applicable sustainability requirements. Agencies may require different analyses, but companies should not have to recreate substantially the same information in several formats.

Codes of Practice should also be accompanied by reasonable implementation lead time. Clause 37 requires notice of a new or amended code to be published, but it does not prescribe a minimum notice period. Except where urgent action is needed, operators should have the time to change systems, retrain staff and obtain any necessary equipment. Will the Ministry consider adopting a "report once, use many times" approach across agencies, together with a default advance notice period for new or materially amended Codes of Practice?

Second, we need a fair and predictable transition for existing data centres. I support progressively raising sustainability standards, but implementation should recognise the practical differences between designing a new facility to meet standards and upgrading an existing facility.

The Government has stated that existing data centres will receive sufficient transition time and that the industry will continue to be engaged on the transition arrangements. This should be translated into a published glide path, with milestones that give operators adequate visibility to plan their upgrades.

Where practicable, the glide path should allow upgrades to be coordinated with planned equipment replacements and refurbishments. It should also take into account factors, such as the age and configuration of the facility, present efficiency, technological feasibility and the scale of investments required. Can the Senior Minister of State provide greater clarity on the principles that will determine the transition period and intermediate milestones for existing data centres?

There is also a regional dimension. Singapore should work with regional partners toward greater comparability in data centre sustainability, resilience and reporting. A practical starting point would be common definitions, metrics, comparable disclosures, followed progressively by stronger standards where appropriate. This would support Singapore's position as a trusted digital infrastructure hub while encouraging higher standards across the wider regional ecosystem. Will the Government explore greater regional alignment of data centre metrics, definitions, reporting standards through the Association of Southeast Asian Nations (ASEAN) or other regional platforms?

Third, we should address our resource constraints at a systems level. The question is not simply about how much additional data centre capacity Singapore can accommodate. We should consider what economic and societal value Singapore receives from every electron and every drop of water allocated to compute.

The Bill allows licence conditions to reflect the strategic, economic and sustainability commitments made by operators. The Bill identifies potential contributions, such as fixed-asset investments, total business expenditure, employment, research and development (R&D), innovation, advanced computing for AI and broader ecosystem development. These considerations should form part of a transparent and consistent assessment of new data centre proposals.

Importantly, as electricity demand grows, the associated generation, grid and infrastructure costs should also be allocated fairly. Large additional loads should not result in households and smaller consumers disproportionately bearing the cost of infrastructure built to serve these loads. Can the Senior Minister of State explain how the Government intends to allocate the additional electricity system costs associated with major data centre demand? How will this impact households?

The same system-level approach should apply to cooling. Future data centre should be assessed for precinct level solutions, such as district cooling, thermal energy storage, shared water infrastructure and advanced liquid cooling. Such infrastructure may create efficiencies that are difficult to achieve if every operator is required to optimise its facilities independently.

We should go further and explore industrial co-location, where waste heat from one facility becomes a resource for another activity. International guidance identifies potential applications in food processing and drying, and certain industrial cleaning or heating processes. The suitability of each use will depend on the temperature and the consistency of the heat, the distance between facilities and whether heat upgrading is required. This should therefore not become a blanket obligation. Instead, land use and infrastructure planning for future data centre precincts should identify nearby activities that could productively use recovered heat.

The Government can help by mapping potential heat sources and uses, reserve spaces for shared infrastructure and enabling pilot projects. Will the Government study precinct-level cooling and industrial co-location for future data centre clusters, including opportunities to use recovered heat in compatible industries?

Energy and water efficiency should also be considered together. An improvement in one metric should not be treated as an unqualified success if it causes disproportionate deterioration in another. The regulations and Codes of Practice should therefore establish clear measurement methodologies, adequate transition periods and a combined assessment of energy and water outcomes.

Enforcement should remain proportionate and distinguish persistent or deliberate non-compliance from genuine technical constraints. Can the Senior Minister of State confirm that future energy and water efficiency requirements will be assessed together, and that operators will receive clear methodologies and sufficient transition time before new requirements become enforceable?

Fourth, our constraints should become opportunities to build Singapore's capabilities. As operators invest in more efficient and resilient infrastructure, Singapore should use this demand to develop capabilities in advanced cooling, power management, cybersecurity, water recycling, carbon accounting, energy optimisation and low-carbon projects.

We should look beyond an operator's direct capital investment and employment numbers to the broader ecosystem that projects create. Where appropriate, access to scarce data centre capacity should be accompanied by meaningful opportunities for Singapore enterprises, startups and research institutes to participate in trials, procurement and capability development. Will the Government incorporate local ecosystem development and opportunities for test-bedding into assessment of the broader value created by new data centre proposals?

Finally, the framework should recognise enabling technologies where their benefits can be demonstrated. Energy storage is not automatically equivalent to low-carbon electricity. Its value depends on how and when it is charged and how it supports the electricity system. However, where an operator can demonstrate that battery or thermal storage increases the utilisation of quality low-carbon energy, reduces peak demand or strengthens operational resilience, those outcomes should receive appropriate recognition. Will the Government consider recognising energy storage and thermal storage where operators can demonstrate that they support qualifying lower carbon energy, reduce peak system demand or improve resilience?

Mr Speaker, Singapore's objective should not simply be hosting more data centres. Our ambition should be to strengthen Singapore's position as a trusted, resilient, connected, resource-efficient and predictable digital infrastructure hub.

Strong regulation is not merely a cost of doing business. If the rules are clear, coordinated and implemented predictably, regulation can be part of the reason companies choose to locate important digital infrastructure here. In a resource-constrained Singapore, we should ensure that every electron, every drop of water committed to compute, generates the greatest possible economic and societal value for Singapore. Sir, I support the Bill.

Mr Speaker: Dr Choo Pei Ling.

12.24 pm

Dr Choo Pei Ling (Chua Chu Kang): Mr Speaker, Sir, we do not think about digital infrastructure until it stops, like water and electricity, and we are a small country. There is no next state where the system is still running.

So, this Bill is the right step. IMDA already has resilience guidelines and a Green Data Centre Roadmap. The Bill makes good practice the baseline. I will ask about three goals: resilience, sustainability and value.

Firstly, on resilience. The Bill requires each major operator to have security and continuity plans. That is right. But my worry is what sits underneath. Think of five shops, each with its own fire plan, sharing one staircase. Each passes inspection. But if the staircase is blocked, nobody gets out. Banks, clinics and Government services can each comply and still depend on the same cloud provider, substation, cooling system or cable route. Engineers call this common mode failure, that is, a shared point of failure.

The Bill defines a major service by whether its failure would disrupt many businesses and organisations. The schedule then applies thresholds: 10 megawatts for data centres, and $100 million in Singapore revenue for cloud providers. Thresholds give industry certainty, and I understand why that matters. But a smaller facility can carry something critical, for example, a hospital group's imaging archive. And a large provider can pose systemic risk because many essential users can depend on it, even if no single customer looks critical.

Hence, how will IMDA assess criticality, looking at substitutability, concentration of users and cross-border dependence? In the European Union (EU), for example, quantitative screens are combined with qualitative assessments of systemic impact and substitutability.

Regulators are expected to supervise their own sectors well. Would the Government consider a risk-based cross-sector exercise programme involving IMDA, CSA and sector regulators where the dependency involves critical infrastructure sectors? The test is not only whether the provider recovers, but whether critical sectors keep running while it does. In addition, will the Government publish a short anonymised annual review of incidents, root causes and recovery times? People trust what they can see being checked.

The next part I am going to talk about is sustainability. The power usage effectiveness (PUE) shows how much energy goes to overhead for each unit that reaches the computers. It measures the efficiency of the building and cooling systems. A low score is good, but it does not show the efficiency or carbon intensity of the computing being done inside.

Clause 35 establishes the statutory basis for sustainability duties. The Government proposes PUE requirements first, while IT equipment energy efficiency and water efficiency requirements will be developed subsequently.

The Ministry will consult industry first, and I understand why. Operators do not control their customers' servers. My question is, when? Can the Senior Minister of State share the timetable and say how existing data centres will move towards the standard already asked of new capacity? I would also welcome a public scorecard, introduced progressively, covering what operators' control: PUE, water usage effectiveness and perhaps the share of low-carbon electricity.

Carbon credits are a sensible safeguard, but they are a bridge, not a routine alternative, to low-carbon energy. Can the criteria be published before any licence condition relies on them? Can IMDA report each year how much has been met by clean energy delivered, and how much by carbon credits?

Finally, on value. Every megawatt given to one operator is a megawatt another user cannot have. Land and power are limited. When we award capacity, Singaporeans are, in effect, the shareholders.

In August, the Economic Development Board (EDB) and IMDA provisionally awarded 200 megawatts to four operators. They committed to green energy for over half the capacity, advanced compute for AI and research partnerships in Singapore. The Bill gives IMDA broad powers over the grant and conditions of data centre licences, while the Government has stated it may consider a data centre's economic and strategic importance.

Where commitments on jobs, local investment, AI compute, research or green energy were material to the award of scarce capacity, would the Ministry be able to clarify which commitments will be incorporated into enforceable instruments, how they will be measured, and what consequences follow from material non-delivery?

I would also like to request for greater transparency. Before licences are finalised, publish a standard framework on the types of commitment, how each is measured, when it is reported and what follows a material shortfall. No commercial terms need be disclosed. If an operator promises jobs, are they net new Singapore-based roles or training places? If it promises AI compute, how much, by when and for whom?

An aggregate report should be published after each round, covering capacity awarded, commitments delivered, local investment, jobs, research partnerships and enforcement actions. That is how Singapore shows it is not merely hosting infrastructure, but converting scarce resources into opportunities for Singaporeans.

Mr Speaker, Sir, in a digital economy, resilience and sustainability are part of national security and competitiveness. I support the Bill, which will keep Singapore a digital hub that is trusted, resilient and resource-responsible.

Mr Speaker: Mr Victor Lye.

12.31 pm

Mr Victor Lye (Ang Mo Kio): Mr Speaker, I wish to declare my interest as an adviser to companies involved in data centre operations.

Sir, in supporting the Digital Infrastructure Bill, I wish to highlight three considerations: one, concentration risk; two, value; and three, capabilities.

First, concentration risk. In addressing concentration risk, we have to identify critical services and map structural concentration and common dependencies. For example, where banks, healthcare providers and Government systems depend on the same underlying infrastructure. The Bill provides for a critical IT load threshold of 10 megawatts. However, is it possible that data centres below the 10-megawatt critical IT load threshold may be systemically important? We must therefore ensure that our system monitoring can capture such concentration risks from the perspective of critical function and interconnectedness, rather than simply the scale of the IT load.

The second consideration is value. Singapore has finite land, power, water and carbon budgets. New capacity should therefore generate broader economic and strategic value for Singapore. Hence, we should measure value outcomes, such as Singapore IT-created R&D, AI capabilities created here, anchored in Singapore and ultimately, productivity. The objective should not merely be value promise per megawatt, but the value actually realised per megawatt. Ultimately, the value must mean good jobs for Singaporeans.

The third consideration would be capabilities. Singapore cannot and need not own every layer of the technology stack. Therefore, we should distinguish what digital capabilities Singapore must host, what we must have access to and what we can safely procure through diversified global networks. In particular, we need to help our small and medium enterprises (SMEs), who provide the most jobs for Singaporeans, to use the compute connectivity and intelligence available through Singapore to create products, participate in global networks and reach markets far beyond Singapore.

Mr Speaker, in concluding, I seek three clarifications. First, how will we understand and how can we manage correlated dependencies across our digital infrastructure? Second, can we measure the economic and strategic value actually realised from every megawatt? And third, what are the economic and strategic capabilities Singapore aims to reap from the compute infrastructure that we are willing to support?

Mr Speaker, notwithstanding these clarifications, I support the Bill.

Mr Speaker: Mr Ng Shi Xuan.

12.35 pm

Mr Ng Shi Xuan (Sembawang): Mr Speaker, the digital world may seem weightless, yet the systems that support it use land, power and water. They also produce carbon emissions. I will raise three points on this Bill: first, data centre firms must keep investing after they set up in Singapore; second, data centres should help support our power grid where they can; third, their growth must lead to good jobs and new chances for our firms and our workers.

My first point is about the need to keep pace with change. The Chinese phrase, "严进宽出", means being strict at the point of entry, but less strict after that. For data centres, our approach should be, "严进，更要严管" – we should be strict at entry, but just as firm in the years that follow.

Data centres are built to last for many years, but the tech within them can change much faster. A data centre that is green and efficient today may fall behind in five or 10 years. The needs of our firms will also change as AI becomes more common. Approval should therefore not become a right to operate forever, based only on pledges made at the start.

Operators should keep investing to use less power and water. They should also keep their sites safe, secure and up-to-date. Could the Senior Minister of State explain how an operator's past record will affect the renewal of its licence and its right to grow? If a firm falls behind in the new rules, what steps can IMDA take before its licence comes up for renewal?

The Resource Efficiency Grant for Emissions (REGE) administered by EDB may help firms make this shift. This Bill sets the basic rules, REGE can help firms invest in new tools that cut emissions. The Grant supports projects that cut at least 250 tonnes of carbon emissions each year. It can fund up to half of the cost that can be claimed. Data centres may use it for better cooling, more efficient uninterruptible power supply systems, smart controls, liquid cooling and lower emission cooling gases.

Could the Senior Minister of State state whether REGE will fund work needed to meet these new rules, or will only fund works that go beyond them? For projects that deliver large and clear carbon cuts, will the Ministry of Digital Development and Information (MDDI) work with the Ministry of Energy, Trade and Industry (METI) and EDB to consider a higher rate of support so that firms adopt better equipment sooner?

Such support should favour projects that lead to direct and verified carbon cuts in Singapore. These projects create demand for equipment, design, installation, testing and long-term support. They can produce local investments, jobs and skills that may not arise when firms rely mainly on carbon credits or other forms of carbon accounting.

Data centres are often called industry enablers. They give firms the computing power, cloud services and digital links they need, but they should also be industry developers. They should help grow new skills, firms and forms of works around them.

My second point is about energy. Under the second data centre call for application, four projects were awarded – about 200 megawatts of new capacity. Each has pledged to obtain more than half of its power through eligible green energy routes. These routes may include solar power built into the site, biomethane, low-carbon hydrogen or ammonia, fuel cells and carbon capture.

Many of these options are still new. Some may also cost much more than power from natural gas. This creates a chance for the four data centres to serve as anchor buyers. Their large and steady demand can give clean energy firms more reason to invest. It can also help form new supply chains, set clearer contract terms and bring down costs over time.

But we need to know what the green energy pledge will achieve in practice. Could the Senior Minister of State explain how IMDA will test whether the energy is real, new and able to reach Singapore when it is needed, rather than relying mainly on carbon credits or accounting measures that do not add new clean energy supply?

There is also the issue of cost. Some of these fuels may cost far more than natural gas, at least in the early years. The green premium may be borne by data centre firms passed on to their client or spread across the wider power system. Could MDDI track how these rules affect the cost of cloud hosting and AI services for our SMEs?

If cost rise sharply, smaller firms may need help so that our green shift does not widen our digital gap. We should not only look at where power comes from, we should also look at when and how it is used. The Energy Market Authority's (EMA's) demand-side flexibility road map points to batteries, flexible loads and virtual power plants as ways to balance demand and supply. Data centres are well placed to help because they use large and steady amounts of power, while their batteries and backup systems may sit idle, except during tests or outages.

Battery Energy Storage System (BESS) is not classified as renewable energy itself. A battery stores power. It does not decide how that power was made, but BESS can help us make better use of clean energy. It can store solar power for later, reduce the use of diesel backup sets and support the grid when demand is high.

Data centre batteries could, therefore, serve two roles without putting the safety of the site at risk. Their first role must remain providing backup power. But when there is enough spare capacity, they could also take part in demand response, provide reserves and help the grid during times of stress.

The cost of large-scale batteries is also falling fast. The Financial Times reports recently shared that battery pack costs for grid storage fell by 45% in 2025, while global storage capacity is now 11 times what it was five years ago. Could BESS qualify for REGE when it leads to clear and verified carbon cuts?

In assessing such projects, EDB could also take into account the source of the power used to charge the battery, the power loss during use and the diesel used that it replaces. More broadly, will IMDA, EMA and EDB set up one clear framework for data centre BESS? Operator should be able to understand how the same battery will be treated under their licence, REGE, their green energy pledge and EMA's demand response plans.

A data centre will still use more power than what a BESS can supply. But where it makes sense, they can also help the grid. Each new megawatt of computing power should support both digital growth and energy resilience.

My final point is about jobs for Singaporeans. A recent article in The Economist found that America's AI build-up is creating work far beyond software jobs. Data centres need electricians, cooling experts, grid engineers and site staff who keep the system safe and running. The article found that job openings in America's data centre sector had more than doubled in two years, even as job ads as a whole fell. It also found that jobs to install and maintain data centre systems offered pay above 40% to similar work in other fields.

Singapore is a much smaller market, and we should not assume that the same gains will arise here on their own. But the article gives us a useful view of where the jobs may come from. We should start preparing our workers before the new data centres open. There are three groups of jobs to consider. First, the jobs needed to build the data centres. These may grow fast, but some may end when the sites are completed. Second, long-term jobs needed to run, secure and maintain the sites. And third, the jobs created when firms use this computing power to build and apply AI.

Our aim should be to secure more of the second and third groups. These jobs can give Singaporeans long careers, useful skills and a path to better pay. We should also help workers in the first group move into longer-term roles when the building work ends. I, therefore, ask MDDI to work with METI and our workforce agencies on the jobs and business roadmap for this sector. It should show which jobs will grow, which may fade, what skills firms may need and how Singaporeans can move into these roles.

The roadmap should set out clear paths for workers with related skills. An electrician could move into data centre power systems or BESS work. An aircon technician could learn to work with liquid cooling. A site manager can move into smart energy control. Our cyber and software staff could gain the skills needed to run and protect AI systems.

The shift to electric vehicles offers a useful lesson. As electric vehicle use grew, the Government, schools and firms had to train workers in battery safety, charging systems and high voltage equipment. Existing workers could not be expected to make the shift on their own.

We should take the same approach for data centres. Training should begin before the new sites open. It should be tied to real jobs, known skill needs, on-the-job training and a clear path to better pay. Workers should also have their past skills recognised, so that they do not have to start again from zero.

Large data centre firms can offer trainee roles, work-study places and mid-career entry paths. They can also help their SME vendors train staff, meet the new rules and move into work of higher value. Such plans should be considered when new data centre capacity is awarded.

Projects backed by REGE can also serve as test sites for Singapore firms working on cooling, batteries, sensors and energy controls. Successful trials would give these SMEs a track record to win work overseas. Grant recipients and firms awarded scarce data centre capacity should therefore report on local trials, local purchases and the growth of their Singapore suppliers.

Sir, our success should not be judged only by how many megawatts of data centre space we build. We should ask how well those megawatts are used, how much they help our power grid, how many Singaporeans move into better jobs and how many local firms gain new skills. If Singapore provides scarce land, power and firm rules, operators should keep improving, help our energy shift and create lasting value for our firms and workers. That is how data centres can be both an industry enabler and industry developer. It is also how each new megawatt can work harder for Singapore. Sir, I support the Bill.

Mr Speaker: Ms Nadia Samdin.

12.47 pm

Ms Nadia Ahmad Samdin (Ang Mo Kio): Mr Speaker, Sir, thank you for the opportunity to speak on this Bill. One of the things which Singaporeans value is efficiency and immediacy – whether it is a transaction going through, a download to our phones or an answer from ChatGPT. The backend digital infrastructure which keeps all of these things running can feel almost invisible.

However, the physical infrastructure required is very real. Data centres occupy land, draw electricity and require cooling. Within them is stored the data and functions on which our digital lives increasingly depend, forming the backbone of modern-day businesses.

For Singapore, one of these constraints, in particular, feels familiar to us – land. In a time when access to compute is becoming increasingly important to economic competitiveness and national capability, how we grow matters as much as how much and how fast we grow.

We have confronted this tension before. In 2019, Singapore paused new data centre development while considering how the sector could grow more sustainably. We subsequently resumed growth in a measured manner.

The International Energy Agency (IEA) projects that global data centre electricity consumption will rise from 485 terawatt-hours in 2025 to about 950 terawatt-hours in 2030, almost doubling in five years. This makes the choices we take today about infrastructure, resources and resilience, especially consequential.

This Bill is an important next chapter in that journey to enable sustainable growth, supporting our AI ambitions, connectivity and the wider digital economy. My speech will ask how do we provide a computing capacity that the digital economy requires without placing unsustainable demands on land, energy and water? As our dependence on digital infrastructure grows, how do we ensure that it remains secure and resilient during a crisis?

Sir, I welcome the decision to make environmental sustainability a stronger part of the statutory framework for data centres. Our green data centre roadmap already looks beyond facility efficiency towards more efficient IT equipment, cooling, server utilisation and green energy. The question is how that broader conception of sustainability will eventually be translated into the environmental requirements under clause 35, reporting under clause 36 and Codes of Practice under clause 37, as well as the corresponding reporting requirements, penalties in the event of contravention and enforcement measures.

PUE measures the energy used by the facility relative to its IT equipment. It does not tell us how much useful computing work that equipment performs, nor does it tell us the facilities' total demand for electricity, water or the emissions associated with that demand. While energy per AI task has been falling, while hardware and software become more efficient, globally, data centre electricity consumption still grew by 17% in 2025, while electricity consumption from AI focused data centres grew by 50%. For example, Ireland saw data centres account for 23% of its metered electricity in 2025.

While our circumstances differ, the planning question is still relevant. How do we align the pace of data centre growth with the additional generation and grid infrastructure needed? I would also like to ask how the additional potential demand has been incorporated into Singapore's emissions trajectory and power sector decarbonisation plans towards our 2050 climate targets? Will facility level PUE targets be complemented by measures of total energy use, and Scopes 1 and 2 emissions?

This is also reflected in the World Green Building Council's recent policy principles, which caution that relative indicators, such as PUE and water efficiency, do not, on their own, tell us the total impact on local electricity and water systems. I was glad to hear in the Senior Minister of State's speech that the Government will consider progressively developing a basket of indicators to complement PUE and water efficiency. I would like to ask whether total water consumption and water stress impacts will also be considered.

There is also a practical opportunity under clause 36. Such licensees will already submit regular information to IMDA. Will the Government consider publishing an annual aggregated picture of the sector's environmental footprint, including overall capacity, electricity, water consumption and relevant indicators? Commercially sensitive operator level information can be protected, and there is a sample precedent in the EU.

While green energy commitments can be enforced as licence conditions, how will the carbon accounting be handled? For example, how will additionality, lifecycle emissions, deliverability of supply be assessed and monitored? I also note that the Bill sets out that IMDA may require the transfer of eligible international carbon credits to the Government to offset resulting additional emissions, where a licensee uses more brown energy than committed because a green energy project has failed to deliver.

Sir, quality international carbon credits may not be so easy to come by. They may address residual emissions but are ultimately not a substitute for delivery of the promised energy pathway. Will these remedies be a backstop after exhausting all efforts to secure replacement, low carbon supply? How will carbon credit quality limits and public disclosure be handled?

My second point is that as data centres become larger and more concentrated, beyond efficiency, we must study how it fits into the wider precinct and systems around it. For example, in terms of security, the impact of energy consumption, heat and noise. As a small and densely built island city state, integrated planning has been one of our strengths. Unlike colder countries, Singapore cannot rely on district heating to make use of waste heat. As we confront the urban heat island effect, we are enhancing the digital urban climate twin to understand how urban form and heat affects temperatures.

Data centres also reject substantial amounts of heat. As we contemplate larger concentrations, could their cumulative heat rejection be assessed through our urban climate modelling alongside local wind conditions, humidity and surrounding land uses? How will such findings inform siting and cooling choices?

The electricity grid raises a similar question. The IEA has highlighted a mismatch between fast moving data centre investment and the longer lead times required for generation and grid infrastructure. Actual data centre load may initially reserve more grid capacity than is effectively used. As we allocate these new data centre capacity, how will the Government ensure that the corresponding demand for power and the infrastructure needed to support it does not affect costs or impose risks on other users? Beyond adding demand, is there potential for our next generation of data centres to contribute to power system resilience? For example, through storage, demand response or flexible load when operationally feasible.

Beyond regulating what individual of operators can reasonably control, I would also like to ask how the Government plans to coordinate the overall needs and demands as considerations across different agencies. A new data centre is simultaneously a digital infrastructure decision, concerns land use, a major electricity load, significant water user and part of our longer-term emissions pathway.

Third is about the value which Singapore receives from the capacity. We know there is no shortage of demand. In 2023, after Singapore announced the resumption of data centre capacity allocation after the earlier pause, our first pilot call attracted more than 20 proposals. Ultimately, four proposals were selected. Competition around us has also intensified, for example in Malaysia and Thailand.

Our objective is not simply to host more data centres, but for greater economic and strategic value from what we choose to host here. Could the Government consider periodically publishing, at least in aggregate, how much capacity has been allocated against how much was actually built and utilised, and whether the material commitments and value associated with that capacity has been delivered?

Resilience also extends beyond Singapore. I would like to ask where a major service used in Singapore depends materially on overseas services, how will the Government assess these dependencies at the point of application, ensure timely information is obtained during a disruption and set standards for recovery? We cannot mistake multiple providers for genuine redundancy.

A business or public service may have a primary and backup provider, but both could ultimately depend on the same cloud control pain, network route or cooling infrastructure. For the people and businesses relying on these services, infrastructure resilience has a very practical consequence. Will the resilience codes therefore look at common points of failure and test scenarios in which several facilities or services are disrupted simultaneously beyond ensuring that individual providers have backup plans? Can the Codes of Practice encourage low carbon backup power and cooling options, while also ensuring that they remain reliable during grid stress or extreme weather considerations? How will companies be supported to implement these?

My fourth and final point. Much of the substantive regime will appropriately sit in regulations, license conditions and Codes of Practice. I hope that the Codes of Practice will specify audit frequency, whether there is a need for independent verification, corrective action paths and consequences for repeated incidents of non-compliance. Singapore has also progressively been developing standards, such as SS 726:2026 for liquid cooling in tropical data centres. How does the Government envision existing SS interacting with the Codes of Practice and mandatory requirements?

Clarity is important, just as is flexibility, to help operators plan more efficiently and support compliance. I therefore welcome the Government's commitment following the consultation to provide sufficient transition time and to continue consulting industry. I am behind being ambitious to raise our standards but believe we should be predictable about how we do so.

I also seek clarification on the 5% ownership reporting threshold. As stakeholders noted during consultation, a passive institutional investor crossing 5% may not materially alter control. Conversely, contractual or voting arrangements can sometimes confer significant influence without a dramatic movement in headline shareholding.

I note that the Bill already recognises indirect control. As reporting is streamlined for transactions that do not affect material control, could the Government provide practical examples of how passive investment holdings, voting arrangements and indirect influence will be treated. I would also like to ask the Government how it plans to address the structural relationship between the state and the firms operating this infrastructure.

The Bill gives IMDA powers to impose conditions and take action when they are breached, but as a provider becomes more indispensable to services, suspending its licence or reducing its capacity could ultimately cause inconvenience or have a negative impact on users. IMDA will have a range of enforcement tool. I would like to ask how it would use this progressively, require independently verified remediation and ensure continuity for users if serious or persistent breaches eventually require a licence to be suspended?

In conclusion, Mr Speaker, Sir, the infrastructure underpinning the next chapter of our economy will increasingly include data centres, cloud and compute. Singapore was prepared to pause and ask what sustainable data centre growth should look like, and that discipline has served us well.

As we enter another phase of growth, I hope the Bill's underlying Codes of Practices will embed that same discipline into our regulatory framework. Not just better PUE, but a better overall resource outcome, not simply greener individual buildings, but infrastructure planned with the energy, water and other systems around it. Capacity that is actually utilised and delivers lasting value for Singaporeans.

For a small country where every megawatt, every drop of water and every hectare of land matters, we must be strategic about the partnerships that we forge and clear about the position we wish to occupy. If there is any nation which can do it, sensitively weighing Singaporeans needs, economic considerations and precious resources, I would like to think that it can be our little red dot.

Notwithstanding these clarifications, I support the Bill.

Mr Speaker: Mr Yip Hon Weng.

1.01 pm

Mr Yip Hon Weng (Yio Chu Kang): Mr Speaker, Sir, I declare that I work for a global investment firm with investments in data centres. I am also the chairman of a non-profit organisation looking at digital defence.

In October 2023, a cooling system malfunction at one data centre disrupted the banking and payment services of both DBS and Citibank. Recovery plans were activated, but problems at their back-up sites delayed recovery further. By the time services were fully restored, about 2.5 million payment and ATM transactions could not be completed.

I raised this matter in Parliament then because the lesson was clear: two organisations can each have a recovery plan and still fail together when they depended on the same critical infrastructure.

In Yio Chu Kang, this is not an abstract concern. Our coffee shops, minimarts, clinics, tuition centres and neighbourhood retailers receive payments, take orders, manage stock and arrange deliveries through digital systems. A customer is ready to pay. The goods are ready to go. But the payment cannot be processed. For the customer, it is an inconvenience. For the shopkeeper, it may mean a lost sale, a delayed delivery or a queue that does not move. The shopkeeper may never know what system failed or which provider was responsible. But the rent still has to be paid. Wages are still due. And the customer may not return.

This Bill will not prevent every payment failure nor is it intended to. But where a major digital infrastructure service fails and disrupts a large number of businesses, the consequences spread far beyond any one shop.

I support the Digital Infrastructure Bill. The Bill establishes licensing regimes for major foundational digital infrastructure services and data centre operators. It gives us the opportunity to establish a clear principle: where dependence is concentrated, accountability must be demonstrable.

My central proposal is simple. Where providers depend on the same critical systems, they must prove that they can recover together. I have three proposals for the Bill.

First, Mr Speaker, Sir, test shared recovery before businesses are disrupted. Clause 15 requires major foundational digital infrastructure licensees to minimise unexpected interruptions and maintain business continuity and disaster recovery plans. That is necessary. However, a plan on paper is not the same as recovery in practice.

A provider may have a sound plan for its own systems, yet still depend on a shared facility, power supply, connectivity pathway or overseas operational team. If that dependency fails, several individually sound plans may fail simultaneously. So, the question is not only: can each provider recover? It is: can the services that our businesses rely on recover when a shared dependency fails?

I ask IMDA to use the Bill's Codes of Practice and licensing powers to require regular, risk-based joint recovery exercises for major foundational digital infrastructure licensees with significant shared dependencies. These exercises should test whether essential services can resume within recovery timeframes appropriately to the likely harm caused by disruption.

They should involve the relevant sector regulators and where necessary, critical upstream operators. These exercises must not simply conclude with just a written report. Every significant weakness found should have a responsible party, a deadline for correction and independent verification that the remedial action has worked. Restoring a service should close the outage and work towards preventing the next one.

I ask the Senior Minister of State whether IMDA will issue the relevant Code within 12 months of the Bill's commencement, conduct the first sector-wide exercise thereafter and require independently verified remediation for serious weaknesses found. For the business owner, resilience is not a compliance report. It is being able to serve and deliver to the next customer.

Second, Mr Speaker, Sir, account publicly for promises made in return for scarce resources. Data centres are essential to Singapore's digital future. But they also draw on scarce national resources, including electricity capacity and water. When Singapore allocates these resources, we must be clear about the public value we expect in return.

Clause 27 allows commitments made by data centre applicants, where they are economically or strategically important, to become licence conditions. Clause 36 requires compliance reporting, which IMDA may require to be audited. A compelling application must become a fulfilled promise.

I propose an annual Digital Infrastructure Accountability Report. Within proper confidentiality limits, IMDA should report in aggregate on the material commitments attached to data centre licences: what was due, what was delivered, what remains overdue and what corrective or enforcement action has been taken.

The report should distinguish a minor administrative delay from a significant unfulfilled commitment. It should also distinguish new economic benefits created specifically because Singapore allocated capacity, from routine business activities that the company would have carried out regardless.

This is not about disclosing commercially sensitive information or creating a fresh reporting burden. It is about using the Bill's existing reporting and audit framework to give Singaporeans confidence that scarce resources are being right sited with discipline.

Clause 35 also establishes minimum energy and water efficiency requirements. How will IMDA secure timely correction when a provider falls short, without causing a second disruption to businesses and consumers? Firm enforcement and service continuity must go together.

Third, Mr Speaker, Sir, ensure accountability survives borders. Singapore is deeply connected to the world. That is a strength. Our businesses benefit from global cloud services, global expertise and globally distributed infrastructure. We should not respond to every risk by insisting that every system be duplicated here. But where recovery depends on a decision made overseas, accountability cannot stop at the border.

The Bill covers major foundational digital infrastructure services provided from outside Singapore to users here. Clause 10 requires an overseas major foundational digital infrastructure licensee to appoint a person in Singapore to accept notices, directions and orders. But a local mailbox is not the same as local accountability. When a critical service is disrupted, can the local representative secure the access, authorisation and cooperation needed to restore it? Can it obtain necessary information? Can it manage a conflict between a direction here and an obligation elsewhere?

I ask IMDA to require major foundational digital infrastructure licensees, through continuity arrangements and licence conditions where appropriate, to demonstrate three things. First, who has the authority to make recovery decisions during a major disruption? Second, how access to the necessary people, systems and information will be secured? Third, how overseas legal or operational constraints will be anticipated and managed?

These arrangements should be updated whenever material cross-border dependencies change. This is not a blanket call for data localisation. It is a call for clarity before a crisis, not improvisation during one. Singapore cannot control every system we depend on. But we can insist on knowing who is answerable, what they can do, and how their obligations will be enforced.

In conclusion, Mr Speaker, Sir, let us return to that shop in Yio Chu Kang. A customer is at the counter. The goods are ready. The business owner has done everything advised and encouraged by our national digitalisation drives: he has digitalised, invested in the equipment and learnt new ways of serving his customers. Then the payment fails. He cannot see the data centre. He cannot inspect the cloud service. He cannot call the overseas team responsible for restoring it. But he bears the consequence nonetheless.

That is why shared recovery must be tested before livelihoods are interrupted. That is why promises made in return for scarce national resources must be checked against delivery. And that is why responsibility must remain effective even when decisions are made overseas.

No law can promise uninterrupted service. But this Bill can require providers to prove that recovery arrangements work, require the regulator to verify the evidence and require those who fall short to answer for it.

When the next customer comes to that counter, the business owner should be able to do the straightforward thing he has worked so hard to do – serve the customer, receive payment and keep business moving. His reliance on these digital systems is a daily reality. The accountability behind those systems must be just as real. Mr Speaker, Sir, I support the Bill.

Mr Speaker: Leader.