DATA CENTRES

New Zealand seems to be moving rapidly towards much larger data centres, expanded cloud computing, AI infrastructure and a more interconnected digital state. Yet these developments are being managed by different government sectors, each responsible for only part of the picture.

A series of Official Information Act 1982 (OIA) request responses has identified that, most fundamentally, there is no overarching oversight architecture, no attempt to balance the burdens potentially placed on local communities by foreign-owned assets, and no visible mechanism for resolving competing departmental objectives. Investment policy may therefore continue to advance even where other agencies have identified unresolved public-interest concerns.

In August 2026 PSGRNZ lead researcher, Jodie Bruning, lodged four Official Information Act requests with the agencies sitting at critical points in this emerging system: MCERT, MBIE, the Department of the Prime Minister and Cabinet (DPMC), and the Government Digital Delivery Agency/Public Service Commission.

The requests included questions that communities should be able to ask before these infrastructures becomes embedded: How much electricity, water and land might it require? Who pays for the supporting infrastructure? Who benefits? Who controls the data and physical infrastructure? What happens to local decision-making? Who is helping Government write the rules? What risks arise as government information and services become increasingly interconnected and dependent upon private infrastructure? How do officials identify excessive overreach across the government agencies when shifts may be incremental and fragmented?

The public may ask: Is anyone in Government responsible for putting all of these questions together?

It's important that the public keep asking questions, because confidentiality provisions have been widely drawn upon to prevent officials from answering questions. A new regulatory system designed to reduce consenting burdens for ordinary development may not necessarily provide an adequate framework for developments with unusually large, cumulative or nationally significant effects.

Data centres are long-lived strategic infrastructure whose cumulative demands may extend well beyond their immediate sites. Large data centres potentially place substantial demands upon electricity generation and transmission, groundwater and surface water, land and supporting infrastructure. Multiple developments may also compete with households, industry, agriculture and other users for finite resources and infrastructure capacity.

Data centres should not be treated as static, century-long infrastructure. Data centres have a relatively short economic and technological lifecycle, commonly estimated at around 15–25 years, although buildings may remain operational much longer through continual refurbishment and equipment replacement. The answers to the OIA requests show that officials have not reconciled public-interest stewardship to balance data centre ownership responsibilities and lifecycles with the potentially much longer life of electricity generation, transmission, water and other infrastructure built to support data centres. If demand falls, technology changes or facilities close or relocate, long-lived infrastructure costs and stranded assets may remain after the original data-centre demand has disappeared. Of course, when contracts involve private sector actors, much of the information can be black-boxed, even though local communities may be impacted.

Stewardship therefore extends beyond deciding whether an individual consent should be granted. Sound governance therefore requires assessment of long-term resource constraints, infrastructure dependencies, cumulative development, drought and water availability, technological change and other material risks. These are issues that affect central government planning and regional communities. The public cannot know whether government assessment frameworks consider horizons of 25 years or longer, rather than treating each development principally through the circumstances prevailing when consent is sought.

The public should be able to access all advice provided to officials, in order to identify potential conflicts of interest and whether any claimed benefits will justify potential costs. Data centres may attract foreign investment, generate tax revenue and create employment, but these benefits need to be weighed against infrastructure and environmental costs, effects upon communities and the possibility that substantial economic returns accrue offshore. We are yet to see cost-benefit analysis, economic modelling, value-for-money assessments and, importantly, distributional analysis: not simply whether economic activity occurs, but who receives the benefits and who carries the costs.

These shifts are occurring are not occurring in isolation, but amid a greater Government effort to consolidate public digital infrastructure and information. Government digital investment decisions must adhere to particular principles. The ‘Single View of the Customer’ and a 'Digital Target State' are embedded as mandatory Digital Investment and Procurement Principles that were overseen and approved by Minister for Digitising Government Judith Collins (See Appendix 1, principle 10).

The current Target State describes a centralised model, with a shared Digital Public Infrastructure through which agencies provide rules and data, connected government services, private-sector channels, agency clustering and authorised data sharing with third parties including businesses. There is no evidence that any assessment of the constitutional and public law implications of this centralised model has been undertaken.

Our major request to MCERT resulted in the disclosure of one July 23 Memorandum and the request to MBIE resulted in the disclosure of one July 2 Briefing. These documents can and do not assure communities that their quality of life and that the quality of resources will be protected. It's also interesting that MCERT forwarded many questions to the DPMC who had already, following a separate OIA request, transferred that request to MBIE.

As with MBIE, MCERT was created swiftly using secondary legislation (not through an Act of Parliament that would have required greater public input), ostensibly to provide joined-up advice across environment, infrastructure, cities, regions and transport, yet (like MBIE) there is no overarching statute establishing the purposes, principles or public-interest duties by which those competing responsibilities are to be integrated. The data-centre issue exposes the consequence: MCERT itself identifies national questions spanning electricity, water, land, infrastructure and environmental externalities, but the OIA trail shows responsibility fragmenting back across DPMC and individual councils, with little evidence that MCERT has undertaken the integrated national assessment its institutional design would appear to invite.

MCERT and the DPMC appear to characterise the DPMC's role rather differently: MCERT considered at least some of the national data-centre governance questions sufficiently connected with DPMC's functions to justify a statutory transfer under s14(b)(ii), while DPMC characterised itself largely as a consulted agency and directed its request towards MBIE. However, MCERT transferred part of its request to DPMC because the information was 'more closely connected with [DPMC’s] functions'. This sits awkwardly beside DPMC's response saying MBIE leads the work and DPMC holds essentially no additional substantive material.

We might ask how the public interest is represented when MCERT expresses concern that the economic benefits of data centres can be overstated and that foreign ownership can result in most of the value accruing elsewhere. These are issues MBIE hasn't publicly addressed, instead the Briefing notes that Invest New Zealand views data centres as a strategic priority area for attracting foreign investment.

The departmental responses reveal four consistent themes that reveal that the agencies are struggling with public interest impartiality: a presumption in favour of development, fragmented assessment of its consequences, no visible decision rule for navigating uncertainty and ignorance of existing expectations for good regulatory process.

Development is the presumption, not the question. The documents largely start from the premise that large-scale data-centre investment will occur and ask how government can facilitate it while managing its effects. They do not begin with the prior question of whether, at what scale, and under what conditions this development is in New Zealand's public interest. MBIE's preference for a 'light-touch approach' and concern not to diminish (foreign) investor attractiveness suggests an unwillingness of that agency to consider domestic trade-offs.
Risks are identified, but never consolidated. Electricity demand, transmission capacity, water, land, environmental effects, foreign capture of benefits, limited permanent employment and opportunity costs all appear in the documents. MCERT itself recognises many of these trade-offs. Yet there is no released integrated assessment weighing these costs, benefits and alternatives together to establish the net public benefit to New Zealand.
Uncertainty does not appear to operate as a reason for caution. The documents acknowledge important evidence gaps, uncertain productivity benefits, uncertain future demand, possible stranded assets and potentially substantial infrastructure consequences. But there is no disclosed framework explaining how much uncertainty is acceptable, what evidence would cause government to reconsider its approach, or what conditions would justify slowing or pausing development. Instead, uncertainty exists alongside continued policy development and investment facilitation.
The requirements of good public decision-making are not themselves identified. Government has established expectations for good regulatory practice, regulatory impact analysis and stewardship, while public law imposes broader disciplines of lawful, rational, accountable and procedurally fair decision-making. Yet the released documents do not identify an overarching process against which the data-centre work is being conducted. There is no visible framework requiring officials to define the problem, establish public-interest objectives, examine genuine alternatives, assess cumulative costs and benefits, address uncertainty, identify affected interests, manage conflicts and unequal influence, or determine how competing considerations across agencies are to be reconciled. This matters because good process cannot readily be reconstructed after major policy and infrastructure commitments have been made. Judicial review remains an important constitutional safeguard, but for developments involving long-lived commitments to land, electricity, transmission, water and other infrastructure, legal scrutiny may arrive only after decisions have become difficult and costly to reverse. The public-law discipline therefore needs to operate upstream, while the policy architecture itself is being designed.

DPMC [OIA-2026/27-0175, 29/08/26] Development of the Government's Data Centre Principles: Participants, Process and Industry Engagement. The DPMC transferred Parts 2, 3, and 5 to 8 of the request, along with aspects of Part 4, to MBIE. The DPMC then stated that the information was 'publicly available' and refused the request. DPMC’s response advises that MBIE is the lead policy agency and that the Prime Minister is being directly informed by advice generated through MBIE’s work. It is therefore surprising that DPMC, despite the Prime Minister publicly bringing it into the process, holds no additional policy-relevant information concerning the development of the proposed data-centre framework.

MCERT [26-OIAD-02576, 29/08/26] Data Centres: Resource and Environmental Governance, Public Participation and the Makarewa Datagrid Development.

Part A of the request (questions 1-10) was transferred by MCERT to the DPMC (OIA-2026/27-0214). DPMC refused to release the information so as to 'maintain the confidentiality of advice tendered by or to Ministers and officials' (9(2)(f)(iv)) and in order to 'maintain the effective conduct of public affairs through the free and frank expression of opinion' (section 9(2)(g)(i)).
After seeking refinement of Part A-5, MCERT released one document (in part) 'with some information withheld under section 9(2)(f)(iv) of the Act to maintain the constitutional conventions for the time being which protect the confidentiality of advice tendered by Ministers of the Crown and officials'. MCERT then refused to answer Part B of the request 'under section 18(e) of the Act as that the document alleged to contain the information requested does not exist in the Ministry’s records or, despite reasonable efforts to locate it, cannot be found'. MCERT then linked to information published by local authorities on the Datagrid project (ICC, ES, SDC).
Part B of the request was transferred by MCERT to Environment Southland, Invercargill City Council, and Southland District Council

MBIE [REQ-0038937 CRM:0257503, 29/08/26] Data Centre Strategy: Scale, Costs, Environmental Limits and Public-Interest Evidence. MBIE responded to link to one document (BRIEFING-REQ-0033553, July 2, 2026), while refusing to answer the balance of the request as 'as the information relates to recent and ongoing policy development and Ministerial consideration. Release at this stage could prejudice those processes'.

Public Service Commission / GDDA [OIA 2026-0282] Digital Government Architecture: Cloud Infrastructure, Digital Identity, Citizen Data & Public/Private Control. Upon request the OIA request was substantially refined (September 10), but was nevertheless refused.

MCERT: RESOURCE, ENVIRONMENTAL & PLANNING GOVERNANCE OF DATA CENTRES

The Ministry for Cities, Environment, Regions and Transport (MCERT) holds a central stewardship function in data-centre development. Many of the most significant consequences of large data centres extend beyond concerns relating to digital technology. They include questions of land, electricity, water, infrastructure, environmental limits, planning, consenting and the allocation of decision-making authority between central and local government.

MCERT was not established via an act of Parliament. It was established as a Public Service department through secondary legislation, Order in Council 2026/6 under the Public Service Act 2020. Schedule 2, Part 1 now lists MCERT as a department. Separate legislation was required to disestablish the statutory Ministry for the Environment and preserve its environmental functions. The underlying decision was taken by the Cabinet Economic Policy Committee, December 10 2025, ECO-25-MIN-0218, and subsequently confirmed by Cabinet through CAB-25-MIN-0476 on 15 December 2025. However, any underlying broader function and purpose of MCERT is difficult to identify.

MCERT is a new ministry and New Zealand's resource-management system is itself undergoing major reform. Much is yet to be defined, particularly the long-term stewardship of resources facing pressures from shrinking aquifers and increasing contaminant pollution. In the absence of publicly available analysis, communities cannot simply presume that this work is being undertaken. The Ministry's December 2025 Resource Management consent data report estimated that up to 46% (±6%) of existing resource consents might become unnecessary under the proposed system and recorded decisions to raise thresholds for identifying affected parties and requiring public notification. Yet the principal sample underpinning that work excluded publicly notified consents, non-complying activities and the highest-cost 25% of consents, which were considered more likely to include large or complex developments.

MCERT occupies a key position between central direction and local democratic authority. National data-centre policy, national planning direction and Fast-track mechanisms may facilitate investment, while regional councils and territorial authorities remain responsible for assessing many of the resulting environmental and community effects. The extent of those effects, particularly without full disclosure, may be unknown. It is difficult to establish what responsibilities are being centralised, what responsibilities and implementation burdens remain with councils, and whether councils possess the scientific expertise, staffing, information and financial resources necessary to assess complex developments and monitor them over time.

Official Information Act request 26-OIAD-02576 sought to understand whether the new effects-management and notification thresholds have been tested specifically against large-scale developments; whether reduced notification could diminish meaningful public participation; and whether the system can identify cumulative pressures across electricity, water, land, infrastructure and environmental limits.

The proposed Makarewa Datagrid provides a practical test of this system. The request asked what MCERT knows about the development's approximately 280 MW anticipated electricity demand, groundwater abstraction, aquifer capacity and recharge, land-use opportunity costs, infrastructure requirements and effects on other resource users. It wanted to find out whether ownership and controlling interests, electricity-market implications and the overlapping roles of Mercury as generator, prospective electricity supplier and equity investor were considered in assessing public benefit.

Makarewa is useful as a governance case because substantial information about the development exists across Overseas Investment Office decisions, council consent files, Fast-track processes and corporate announcements, yet the principal resource-consent applications were not publicly notified. Our concern was not limited to whether individual agencies discharged their particular statutory functions, but to find out whether any institution is responsible for assembling the whole picture.

The OIA asked whether anyone has integrated Makarewa's electricity demand, water requirements, land use, infrastructure, ownership and control, environmental effects, economic costs and benefits, community consequences and public-participation implications into an assessment of its net public benefit to New Zealand.

MCERT transferred a large portion of the request to the DPMC, Environment Southland, Invercargill City Council, and Southland District Council.

MCERT subsequently refused the response with the exception of a Memorandum that identified a spectrum of governance challenges internally in July 2026. MCERT then refused the correspondence component under s18(f), stating that the documents sought under Part B otherwise do not exist in MCERT's records or could not be found.

The July Memorandum reveals MCERTs admission that data-centre economic benefits can be overstated, that much of the value may accrue to international investors, that permanent employment is modest relative to capital and electricity demand, and that the principal environmental/community issues include water, land-use opportunity costs and infrastructure. It expressly recognises that adverse impacts can be localised while benefits may occur regionally or nationally.

The July Memorandum acknowledges that electricity and network capacity can be diverted from residential, industrial and agricultural uses and notes that overseas governments increasingly consider precisely those trade-offs. Yet under 'key gaps' it says this 'can only be assessed on a site-by-site basis, rather than nationally'. However, its own preceding discussion recognises system-level and national consequences. Hence, a single consenting authority cannot determine the cumulative national opportunity cost of allocating scarce electricity/network capacity across multiple data centres, yet the local authorities require the resources to undertaken their own long-range analyses.

Yet its OIA response reveals little evidence that these questions have progressed beyond an internal scoping memorandum. Instead, substantive questions have been distributed between DPMC and individual local authorities, leaving unclear which agency is responsible for integrating national environmental limits, electricity and infrastructure opportunity costs, land and water pressures, and cumulative effects into the Government's emerging data-centre framework.

The underlying stewardship question extends beyond thorough consenting: does New Zealand have an institution capable of determining whether major data-centre development is environmentally sustainable, infrastructurally prudent and in the long-term public interest, or are these questions being fragmented across agencies and individual consenting decisions without anyone being responsible for the whole?

MBIE: INVESTMENT, INFRASTRUCTURE & THE PUBLIC-INTEREST GOVERNANCE OF DATA CENTRES

The Ministry of Business, Innovation and Employment (MBIE) occupies a key position in data-centre policy because it sits at the intersection of economic development, investment, energy, infrastructure and national productivity. While environmental and planning agencies are principally concerned with how developments are consented and environmental effects managed, MBIE has responsibility for the larger economic question: why should New Zealand encourage data-centre development, at what scale, and under what conditions would doing so be in the national public interest?

The Government is developing principles or rules for data centres, and MBIE has already undertaken policy work concerning both Data centre investment in New Zealand and options to capture benefits and manage risks. [BRIEFING-REQ-0037878].

Our OIA [26-OIAD-02576] sought the underlying analysis supporting that work, rather than simply the resulting policy position. We were provided with a July 2026 Briefing that is much narrower in its scope and that does not address the OIA questions.

Large data centres are unusual investments because private commercial benefits may be accompanied by substantial demands upon publicly shared or regulated infrastructure. Electricity generation, transmission and distribution capacity, water, land and supporting infrastructure are finite resources. Decisions to facilitate data centres can therefore determine not merely whether individual investments proceed, but how national infrastructure capacity is allocated between competing uses, what other uses may be displaced, and who ultimately bears the costs and risks.

MBIE's stewardship role should extend beyond any investment coordination objectives. Public and administrative law conventions should mean that MBIE is impartially capable of establishing whether the proposed industry represents a sound allocation of New Zealand's scarce electricity, infrastructure, water and land resources, and whether the economic gains justify the public and private costs.

Norms of transparency and accountability should be displayed: before New Zealand creates policy intended to attract and facilitate large-scale data-centre investment, has Government established what scale of development it wants, what resources that development will require, who will pay for them, what alternative uses may be displaced, and whether the overall result will produce a demonstrable net public benefit for New Zealand?

As the lead agency for the Government's data-centre policy work, MBIE is central to understanding the anticipated scale of development. Before Government creates a favourable policy environment, it is reasonable to know whether it anticipates several relatively modest facilities or a substantial hyperscale and AI-computing industry, and what that development could require over medium and long-term horizons.

The request [26-OIAD-02576] sought scenarios and modelling of electricity demand and additional generation, transmission and distribution requirements, together with water demand and availability, land requirements and associated infrastructure.

A second key issue concerned who pays - the public do not want the costs to be made public and the profits to be privatised. Infrastructure required by data centres does not necessarily remain a private cost to their operators. New generation, transmission lines, network upgrades, water infrastructure or other supporting investment may ultimately affect electricity consumers, ratepayers or taxpayers. The request therefore sought analysis of public expenditure and infrastructure investment associated with facilitating the sector, together with evidence identifying where those costs are expected to fall.

The energy draw makes MBIE's stewardship role particularly significant and large data centres can represent very substantial new loads. The request asked whether operators should provide or finance the additional generation, transmission or storage their operations require; whether allocating electricity-system capacity to data centres carries opportunity costs for households, existing businesses or other productive activities; and whether major data-centre demand could affect electricity prices or security of supply.

Finally, the request asked MBIE to identify uncertainties and evidence gaps before the policy framework is established, rather than allowing investment policy to outrun the evidence required to govern its consequences. It also sought information on the resources Government itself has committed to developing the framework, including commissioned research and external expertise.

The only substantive document released, a July 2026 briefing, does not resolve these concerns because it begins from a substantially narrower policy question: how to attract data-centre investment, capture economic 'spillovers' and manage risks without materially diminishing investor returns. Although the briefing recognises sovereignty, skills, innovation and wider infrastructure issues, its substantive risk analysis is concentrated principally on electricity-system effects, alongside the policy objective of attracting investment and capturing economic spillovers.

The July Briefing downplayed broader responsibilities, and appeared to expressly frames its task around attracting data centres to respond to global demand, rather than first undertaking a wider assessment of New Zealand's domestic infrastructure needs and competing priorities.

The broader international analysis sought was not provided. MBIE's released Briefing considered Australia and Ireland, but does not provide the wider comparative assessment requested across the Netherlands, Japan, South Korea, the United States and Nordic countries, including how foreign investment concerns have been navigated.

MBIEs July Briefing is fundamentally different from the public-interest analysis sought in our OIA. We asked who bears infrastructure costs; where the economic benefits ultimately accrue; what opportunity costs arise from allocating electricity, water, land and network capacity to data centres; what cumulative environmental and freshwater effects may result; how communities may be affected; how foreign ownership and offshore returns alter the economic equation; and, ultimately, whether the development produces a net public benefit to New Zealand.

Yet MBIE concludes that substantial intervention is presently unwarranted and favours 'modest preparatory actions' and a 'light-touch approach', while cautioning against measures that could diminish New Zealand's attractiveness to investors.

Although it is the 'lead agency' the information disclosed in MBIE's response does not include a comprehensive public-interest assessment establishing whether, at what scale, and on what terms large-scale data-centre development is desirable for New Zealand. What it discloses is principally an investment-policy briefing concerned with how anticipated investment can be accommodated and shaped. The prior question, whether that investment represents a demonstrably sound allocation of New Zealand's resources in the long-term public interest, remains unanswered.

DPMC: WHOLE-OF-GOVERNMENT COORDINATION, THE PUBLIC INTEREST & THE INTEGRITY OF DATA-CENTRE POLICYMAKING

The Department of the Prime Minister and Cabinet (DPMC) has been directly tasked by the Prime Minister with helping develop the Government's proposed data-centre principles or rules. The OIA request records the Prime Minister's July 2026 direction involving DPMC, the Minister for Economic Growth and MBIE, alongside an intention to involve energy companies, data-centre operators and other interested parties in shaping the rules.

The DPMC's governance role is different from that of an environmental regulator, consenting authority or economic-development agency. Its significance lies in whole-of-government coordination and the integrity of the policy-development process itself. At this formative stage, decisions about the purpose, scope and objectives of the framework can determine which questions government subsequently asks, which risks it examines, whose evidence it considers and ultimately what constitutes an acceptable data-centre development.

The DPMC's stewardship responsibility is therefore fundamentally about the quality and legitimacy of the governing framework. We might think that the DPMC has a strategically important role because data centres cut across conventional departmental boundaries. Their development potentially engages energy security, infrastructure, freshwater and environmental management, foreign investment, competition, privacy, digital governance, local government and community interests. No single sectoral agency necessarily sees all these dimensions.

The DPMC is the agency in this set of OIAs that is most directly positioned to ensure that policy development is coordinated across government, that relevant but competing interests are represented, that commercial expertise does not become commercial dominance, and that constitutional and public-interest considerations are incorporated before consequential policy settings become embedded. The DPMC's OIA request [OIA-2026/27-0175] therefore looked upstream of individual policy decisions to strategy. It asked when the Prime Minister first directed the DPMC to participate and seeks the terms of reference, project brief, objectives, work programme, deliverables and milestones.

However, in response to the request the DPMC responded to state that MBIE was the lead agency responsible for developing the Government's data-centre framework, and that DPMC held essentially nothing additional beyond the material transferred to MBIE.

However, if MBIE leads development of the data-centre policy, but MCERT considers the overarching governance questions more closely connected with DPMC's functions, what exactly is DPMC's role, and where is the corresponding analysis?

GDDA & THE PUBLIC SERVICE COMMISSION: DIGITAL ARCHITECTURE, SOVEREIGNTY & THE PUBLIC CONTROL OF DATA-CENTRE INFRASTRUCTURE

The Government Digital Delivery Agency (GDDA), located within the Public Service Commission (PSC), occupies a critical position in data-centre governance because government is itself becoming a major user of interconnected cloud, computing and data-centre infrastructure. Following the transfer of Government Chief Digital Office functions from DIA to GDDA on 1 April 2026, the Government has pursued a more centralised and coordinated approach to digital investment, procurement and delivery.

GDDA's role therefore extends beyond purchasing information technology. Following the Hon Judith Collins digital government policy reforms the government's Digital Government Target State establishes a direction towards consolidation, common infrastructure and greater interoperability, progressively grouping agencies around shared ICT platforms and systems. It was underpinned and justified by an extensive evaluation of cost savings. The Digital Investment and Procurement Principles similarly encourage agencies to treat data as a strategic asset, enable connected data flows and support a ‘Single View of the Customer’ (Appendix One, principles). Meanwhile, Cloud First policy encourages public-cloud services.

The GDDA/PSC is therefore a key data-centre stewardship agency. Government information, AI-enabled computing, digital identity and essential public services ultimately require physical infrastructure through which information is stored, processed and transmitted. Government is therefore not merely regulating the emerging data-centre ecosystem: it is designing its own digital architecture in ways that may make government increasingly dependent upon it.

The implications extend beyond procurement. Decisions about digital architecture can shape future dependence upon particular infrastructure, providers and jurisdictions, while increasing the technical capacity to connect government agencies, services and citizen information.

The original OIA request [OIA 2026-0282] asked for an integrated assessment bringing together the Digital Government Target State, Cloud First, cloud and AI/data-centre infrastructure, digital identity, interoperability, citizen data, private-provider dependency, jurisdictional risk and the Digital Investment and Procurement Principles. It also sought the independent expertise used to challenge that architecture and analysis of alternatives, including decentralised or federated systems, deliberate separation, multiple providers and publicly controlled infrastructure.

We are concerned about industry concentration and dependency and the concentration of private citizen information. Reliance upon a small number of cloud or data-centre providers may create technological and contractual lock-in, switching costs and dependencies that become difficult to reverse. The request therefore sought analysis of provider concentration, ownership and control of infrastructure, bargaining power and the strategic and operational risks arising where essential public functions depend upon particular private providers.

A related stewardship question concerns data sovereignty. Physical location within New Zealand does not necessarily answer questions about legal or effective control. The OIA therefore sought analysis distinguishing physical location from foreign ownership, corporate control, foreign jurisdiction and governmental access powers.

The request also raised the implications of increasing interoperability. Different agencies exercise different statutory functions and hold information for different purposes. Common platforms, digital identity, cross-agency data exchange and a ‘single view of the customer’ may make previously separate information and functions increasingly capable of being connected. The OIA therefore asked not only what systems are currently intended to do, but whether Government had considered what the resulting architecture could become capable of doing if permissions, software, policy or legislation subsequently changed.

Resilience provides another direct connection to data-centre stewardship. Concentrating government information, computing capability and essential services within common providers or interconnected infrastructure can potentially create systemic dependencies or single points of failure. The request consequently sought the analysis undertaken of these risks and of continuity of government and essential public services.

But the PSC’s response [OIA 2026-0282] give rise to a number of constitutional and public law concerns relating to the independence and separation of powers.

PSC initially responded (September 9, 2026) with a letter of clarification that asked for the request to be refined. The OIA request was subsequently substantially narrowed (September 10, 2026). The PSC then refused the refined request (September 25, 2026) under section 18(f), stating that locating, reviewing and preparing the relevant information would still require 'substantial collation and research'.

Several of the documents the PSC itself identified in this letter are not independent assessments of the direction Government has chosen, but policies, standards and assurance mechanisms: the Digital Government Target State, Cloud First policy, Cloud Jurisdictional Risk guidance, an API standard, third-party information-sharing standards, procurement principles and data-centre certification.

The PSC letter of clarification raises several concerning questions and does not address our concerns regarding aggregation, linkage and function creep, and separately about private-provider use, accountability and control. The clarification points in both places to the Law Commission's automated-decision-making review for questions in part 7: Aggregation, linkage and function creep Public/private access, interoperability and integration, and part 10 Private-provider use, accountability and control of the request.

For Part 11, PSC directs the public to the February 2026 Digital Investment and Procurement Principles that expressly require agencies to align investment to Judith Collins’ developed Digital Government Target State, interoperability, Cloud First, connected services and Collins ‘single view of the customer’. The Digital Government website states that:

These principles were approved by the Minister for Digitising Government and noted by Cabinet in December 2025.

The Cabinet paper released February 2026 gives effect to this.

Driving down the cost of digital in government — Implementation — Public Service Commission (PDF 4.5MB)

Yet PSC points to those same principles in response to questions about resilience and continuity. Public Cloud Data Centre Certification similarly provides security assurance within the Cloud First model, but does not constitute a system-level assessment of whether increasing government dependence upon cloud and data-centre infrastructure creates concentration or continuity risks.

When asked for analysis of alternatives and deliberate limits (part 17), PSC identified a 2021 Digital Identity Trust Framework release rather than an identifiable comparative assessment of centralised, federated, decentralised, multiple-provider or publicly controlled alternatives to the current Target State.

From a public law perspective, this is most concerning, documents explaining how to implement, secure or govern a chosen architecture are not evidence that Government first subjected that architecture to a structured assessment against materially different alternatives.

A collection of standards governing particular components is not the same thing as a structured policy process requiring officials to define the problem, identify public-interest objectives, examine genuine alternatives, assess cumulative costs and benefits, identify uncertainty and dependencies, consider reversibility and determine whether the resulting architecture is justified before committing to it.

[OIA 2026-0282]

While the response letter and it's hyperlinked articles does not establish that the relevant analysis does not exist, it raises a significant stewardship question relating to how public officials inside government might understand the state of play relating to not only data centre oversight and stewardship but how this might work when the Digital Government Target State and the Single View of the Customer is considered.

If the policy reasoning underpinning an increasingly interconnected digital state cannot readily be identified and retrieved, where does an official working on one part of that system go to understand the whole and how does civic autonomy, sovereignty and rights operate in such a system, especially over time?

The Government's established regulatory-policy processes which we highlighted earlier, expect the clear and considered analysis of problems, objectives, options, impacts, risks and implementation. Yet the PSC/GDDA response does not identify an equivalent overarching pathway for examining the cumulative implications of the Digital Government Target State.

There is an unusual institutional circularity here. Judith Collins held both the Public Service and Digitising Government portfolios while the Government established GDDA within the Public Service Commission and developed the more centralised Digital Government Target State. She subsequently became President of the independent Law Commission. The automated-decision-making review was commissioned by Justice Minister Paul Goldsmith before Collins assumed that office, so it would be wrong to suggest that she commissioned the review herself.

Judith Collins KC role as future president of the Law Commission was announced January 28, her office was releasing substantive Target State and Single View of the Customer decisions in February, she relinquished her ministerial portfolios on April 7, on April 21 the Law Commission automated-decision-making review was announced, on May 14 Collins left Parliament and on June 1 2026, Collins commenced her duties as Law Commission president.

When PSC was asked in this OIA for analysis concerning aggregation, linkage, function creep and private-provider accountability within the developing digital architecture, it pointed to that Law Commission review.

The Law Commission review concerns automated decision-making and is intended to develop a coherent legal framework in an area where the Commission itself says government currently lacks overarching law, standards or guidance. It should not substitute for contemporaneous policy analysis by PSC/GDDA of the broader architecture it is already implementing, including aggregation, interoperability, provider dependency and function creep. Nor should prospective independent law reform be treated as equivalent to evidence that these questions were systematically examined before the architecture was adopted.

The PSC OIA response reveals to the public that the PSC does not hold a readily identifiable structured process through which an official can determine whether increasing consolidation, interoperability and dependence upon private digital infrastructure is, as a whole, in the long-term public interest.

It is unclear who prompted the Minister of Justice to step in to direct the Law Commission to undertake an analysis of risks that should already have formed part of the structured policy analysis underpinning the Government's digital architecture, particularly when the Law Commission is now headed by the former Minister who held responsibility for Digitising Government while key elements of that architecture were designed and adopted.

WHO IS CREATING THE RULES?

Governance begins with problem definition. A framework established principally to facilitate investment may develop very differently from one established to determine whether, where and under what conditions data-centre development serves New Zealand's long-term public interest. A central coordinating agency is therefore well placed to ensure that the Government's framework is not constructed through one institutional lens while important consequences residing elsewhere in government remain outside consideration.

Yet responsibility for developing the Government's approach remains surprisingly nebulous. DPMC has told us that MBIE is leading the work. The principal MBIE document released to us is a July 2026 Briefing entitled Data centres: options to capture benefits and manage risks. It records consultation with DPMC, the Government Digital Delivery Agency and Invest New Zealand. Its stated purpose is to advise on levers to capture spillovers from data-centre investment and manage potential risks.

MBIE records that Invest New Zealand has identified data centres as a strategic priority for attracting foreign investment and expressly states that its briefing focuses on the levers available to capture benefits and manage risks associated with attracting data-centre investment to respond to global demand. Later, officials recommend a predominantly 'light-touch, market-led framework', while cautioning that new policies should not diminish New Zealand's attractiveness to investors.

MCERT's July Memorandum Data Centres in NZ provides a different perspective. It warns that the economic opportunity can be overstated, that much of the value from hosting foreign-owned data and AI may accrue to international investors, and identifies electricity infrastructure, water, land-use opportunity costs and local environmental effects as significant considerations. It also asks key questions about the role of the planning system, whether data centres should be regarded as ordinary industrial developments or nationally significant infrastructure, the possibility of stranded energy assets, and competition with other land uses.

These are precisely the questions that might be expected to form part of an overarching public-interest framework. The Government's own Government Expectations for Good Regulatory Practice (2017) provide a useful benchmark. Before substantive regulatory change, agencies are expected to clearly identify the underlying problem, systematically analyse impacts and risks, consider alternative regulatory and non-regulatory options, make a genuine effort to identify and estimate costs and benefits, address implementation and resourcing, and provide affected and interested parties appropriate opportunities to participate. Current Regulatory Impact Analysis guidance similarly emphasises the rationale for intervention, alternative options, costs and benefits, and the views of those consulted. Regulatory stewardship - see Starting out with regulatory stewardship: A resource - is explicitly concerned with ensuring that the different parts of a regulatory system work together and remain fit for purpose over the long term.

Yet the MCERT memorandum describes them only as 'questions we could consider'. On the material released to the public, it is difficult to identify the overarching governance process through which MBIE's investment objectives, MCERT's environmental and resource concerns, electricity-system consequences, local-government responsibilities, digital sovereignty, competition, infrastructure investment and wider public-interest considerations are to be reconciled.

Against all of this advice a broad spectrum of elementary governance questions remain unanswered. Where is the overarching problem definition? What are the objectives? What alternatives have been considered? Where is the net public-benefit assessment? How are long-term infrastructure and opportunity costs being evaluated? How are uncertainty and potentially stranded assets being treated? Who is responsible for reconciling competing objectives across agencies? And who has been allowed to participate in answering these questions?

Following responses from these agencies, we still do not understand who participates in creating the rules. For example, the OIA request asked the DPMC to identify government agencies and Ministers, but also consultants, energy companies, electricity-sector interests, data-centre operators, technology companies, investors, experts and other external organisations involved in developing the principles. It additionally sought meeting records and documents showing the substance of engagement between DPMC and external organisations. Such information was not provided.

We would think that there would be files that could be drawn upon, where all of the agencies lodge relevant analysis, advice and information to a decision, and create an informational pathway for officials to work from.

Industry knowledge is plainly relevant to workable policy. However, firms proposing to build, supply or finance data centres have strong commercial incentives concerning the resulting regulatory environment. The governance issue is not whether industry should participate, but whether its access, advice and influence are balanced by interests that may lack equivalent resources, organisation and direct financial incentives to engage.

We therefore asked whether DPMC had developed a stakeholder map or consultation strategy and what processes exist to represent affected communities, local government, electricity consumers, environmental and freshwater interests, privacy and civil-liberties interests, Māori, independent scientists, competition and consumer interests, and public-law or constitutional expertise.

These are fundamental stewardship questions. Decisions about large infrastructure projects can distribute costs, benefits, risks and decision-making power across generations and communities. Public participation after the fundamental policy settings have already been determined is not equivalent to participation while the objectives and rules themselves are being formulated.

The request also sought advice concerning conflicts of interest and whether officials had considered regulatory capture, policy capture, lobbying and unequal stakeholder influence, together with safeguards intended to ensure that the resulting framework reflects the wider public interest.

Ministerial secrecy is more than a transparency problem. It a problem of knowledge flows and access. Agencies may legitimately protect some advice to preserve confidentiality or the free and frank expression of opinion. But where those provisions prevent the public from seeing significant portions of the policy record, while the institutional framework itself remains unclear, the public cannot readily determine who supplied the advice, whose evidence was relied upon, which interests were represented, which alternatives were considered, or whether the advice was sufficiently independent and comprehensive.

The concern is not that withheld advice must therefore be deficient or industry-driven. It is that the public presently lacks the information necessary to test whether it is not.

Amidst all of the process-based concerns sits a deeper constitutional and public-law concern. As Philip Joseph's account of New Zealand's constitutional arrangements emphasises, the rule of law, separation of powers, executive accountability and judicial review operate together as constraints upon public power. Yet judicial review is principally concerned with the lawfulness of executive action and the process by which decisions are reached, rather than providing a substitute for sound policy formation. For developments of this scale, waiting for judicial review may also be waiting too long. By the time a reviewable decision crystallises, consents may have been granted, contracts entered, land committed and generation or transmission infrastructure planned or constructed around assumptions that become economically and politically difficult to reverse.

The constitutional safeguard must therefore begin upstream, with transparent problem definition, clearly allocated responsibility, adequate evidence, impartial advice, meaningful participation and accountable decision-making before long-lived commitments are made. Judicial review remains an essential backstop, but it is a poor substitute for good public administration at the point when the policy architecture itself is being designed.

The central governance question posed by our OIA requests is therefore: before Government and commercially interested parties begin shaping the rules for data centres, has Government established a sufficiently broad and independent public-interest framework to determine what those rules should achieve, whose interests must be protected, which risks and alternatives must be considered, how costs and benefits should be distributed, and who should have a meaningful voice in making them?

CONCLUSION: THE PUBLIC ARE LEFT TRYING TO STITCH TOGETHER FRAGMENTED INFORMATION

Data centres are increasingly systemically important digital infrastructure and should not be treated merely as commercial developments. As government, businesses and essential services become increasingly dependent upon them, New Zealanders have a legitimate interest in how this infrastructure is owned, governed and controlled.

Public-interest stewardship should therefore include genuine consideration of alternative ownership and architectural models, including New Zealand ownership, public part-ownership, sovereign or publicly controlled infrastructure, multiple-provider arrangements and deliberate limits on the offshore storage, processing or control of sensitive information. These are choices about national resilience, sovereignty and the distribution of long-term benefits, not simply questions of investment attraction.

There is little transparency over who is providing advice to these agencies, and whether the advisors have financial or political conflicts of interest.

These are reasonable public concerns. Before infrastructure of this scale becomes embedded, government should be able to demonstrate that it has asked the right questions, brought the relevant evidence together, considered genuine alternatives, identified uncertainty and competing interests, and established that the resulting direction serves the long-term public interest.

The departmental responses do not yet provide that assurance. Instead, they reveal four recurring problems: development is largely presumed rather than tested; its consequences are fragmented across agencies rather than assessed together; significant uncertainty is acknowledged without a visible decision rule for responding to it; and the requirements of good public decision-making are not themselves clearly identified. MBIE approaches the issue principally through investment and economic growth; MCERT identifies significant environmental, resource and infrastructure questions but has disclosed little evidence of an integrated assessment; DPMC points to MBIE as lead agency; and important questions concerning the architecture of digital government remain elsewhere.

The result is an institutional landscape in which transparency and accountability is poor, where many agencies hold pieces of the problem, and where institutional power - and responsibility is nebulous.

Concerningly, we cannot identify a mechanism for resolving conflicts between departmental objectives. MCERT has identified concerns including potentially overstated economic benefits, benefits accruing offshore, modest permanent employment and opportunity costs involving electricity, water and land. MBIE, meanwhile, approaches data centres principally through economic growth, investment attraction and management of the risks accompanying anticipated development. The released material does not identify an overarching decision framework explaining how these competing considerations are to be weighed, which agency's assessment prevails, or what evidence could require the Government to slow, modify or decline a proposed direction. Without such a framework, there is a risk that investment facilitation becomes the de facto decision rule simply because it is the policy process that continues to move forward.

That makes transparency over who is creating the rules, who is advising government, what evidence is being relied upon, whose interests have been represented and what alternatives have been considered particularly important. Confidentiality and 'free and frank' protections have legitimate constitutional purposes, but extensive withholding also limits the public's ability to test whether the underlying advice is independent, sufficiently broad and consistent with government's own expectations for good regulatory practice. The concern is not that undisclosed advice must be deficient or commercially influenced. It is that the public cannot presently establish that it is not.

For infrastructure capable of committing electricity generation and transmission, water, land, public and private capital, digital systems and communities for decades, accountability cannot sensibly depend upon correcting mistakes afterwards. Judicial review remains an essential constitutional safeguard, but by the time a reviewable decision reaches the courts, consents may have been granted, contracts signed, infrastructure commissioned and billions of dollars committed. Good public law and good regulatory stewardship therefore have to operate upstream, while the problem is being defined, alternatives remain genuinely available and the rules themselves are being written.

New Zealanders should not have to reconstruct the governing architecture from council consent files, Cabinet papers, corporate announcements and successive Official Information Act requests. The question these four OIA requests ultimately pose is remarkably simple: before the architecture is built around us, who in Government is responsible for ensuring that the whole thing is actually in the public interest?

Download

Data Centre Governance: 2026 Official Information Act request series