At a glance
- Nine in ten New Zealand organisations now use AI in some form, on Datacom's annual index of 207 senior business leaders.
- Datacom's index puts organisations using AI to change core operations at 4%, and those scaling it across the business at 15%.
- Investment rose at 79% of organisations over the past year, and 73% plan to raise it again.
- Dedicated AI leaders sit inside 22% of the organisations surveyed, while 26% have added AI to an existing executive job.
- A government survey counted 272 uses of AI across 70 agencies, and a 2026 review puts a fifth of them in service.
Datacom's State of AI Index puts New Zealand AI use at 91%
Datacom runs five of the data centres in New Zealand and sells cloud, managed services and AI consulting to enterprise and government customers. It also surveys New Zealand organisations once a year on how far they have got with AI, and released the 2026 index in August.
Datacom's index puts AI use at nine in ten New Zealand organisations, and the share using it to change core operations at 4 per cent. Adoption was 87 per cent in 2025. In 2024 it was 66 per cent. Another 15 per cent of organisations describe themselves as scaling AI across the business, and the report puts the remaining 81 per cent in the exploratory or implementation phases. The New Zealand Herald reported the core operations measure at 8 per cent a year earlier.
Dedicated AI leaders sit inside 22 per cent of the organisations surveyed. Twenty-six per cent have added AI to an existing executive job, and 19 per cent share it across several executives.
Lou Compagnone, Datacom's director of AI, states the finding in the report: "There is a significant difference between AI adoption and AI adaptation. Many organisations are using AI, but relatively few have started redesigning workflows, services and operating models around what AI makes possible."
In a written Q&A with Certified Strategic published on 19 August 2026 she named what closes that gap: a chief AI officer who owns strategy, governance and business outcomes within six months, and workload requirements such as latency, data sensitivity and sovereignty defined while a use case is being scoped rather than after a proof of concept succeeds.
The government has counted its own AI work too. A 2025 government survey recorded 272 AI use cases across 70 agencies. A review of digital delivery released on 2 July 2026 says only 20 per cent of them are deployed and operational.
| Datacom's 2026 State of AI Index, New Zealand | Share of organisations |
|---|---|
| Use AI in some form | 91% |
| Exploring or implementing | 81% |
| Scaling AI across the organisation | 15% |
| Using AI to change core operations | 4% |
| Have a dedicated AI leader | 22% |
| Have a chief AI officer | 13% |
| Added AI to an existing executive role | 26% |
| Have a clearly defined standalone AI strategy | 16% |
| Increased their AI budget in the past 12 months | 79% |
Source: Datacom, 2026 State of AI Index. Survey of 207 senior business leaders at New Zealand organisations of 100 or more staff, conducted by Curia Market Research.
Peter Nelson links model choice to what the software costs to run
Seventy-nine per cent of organisations increased their AI budget over the past 12 months, and 73 per cent plan to increase it again. On payback, 33 per cent said returns had exceeded costs, and 30 per cent said benefits and costs were broadly balanced. Nearly four in ten reported significant or transformational impact on the business.
General-purpose assistants including ChatGPT, Claude and Gemini are the AI most commonly used, at 68 per cent. The report flags what buying more of them will cost: 20 per cent of organisations have not addressed AI cost management at all, 15 per cent have been surprised by AI costs exceeding expectations, and another 15 per cent say they are aware of the shift in pricing but have not adapted their budgeting.
Datacom New Zealand managing director Peter Nelson told the New Zealand Herald what that shift is. The large AI platforms are moving bigger customers off a flat fee per employee each month and onto charging for how much the software is used, which puts a price on the choice of model. "You wouldn't hire a 50-seater bus to transport two people from Hamilton to Wellington," Nelson said, "so why use an expensive frontier AI model to improve an email or create basic content for a sales proposal?"
Datacom's index puts dedicated AI leaders at 22% of organisations
Thirteen per cent of organisations have a chief AI officer and 9 per cent another full-time AI leadership position. Twenty-six per cent have added AI to an existing executive job, 19 per cent share it across several executives, and 20 per cent have no plans for an AI leader.
Sixteen per cent have a clearly defined standalone AI strategy. A further 23 per cent have formally embedded AI inside a broader business, digital or technology strategy. Another 23 per cent say their strategy is high-level or not fully operationalised.
Compagnone argues the job spans the whole executive rather than the technology function. In the report she says AI "is no longer just a technology initiative, it touches strategy, governance, workforce capability and organisational change". She told the Herald: "For me, it's a socio-technical technology. It's something that requires transformation skills because of its impact on people and roles." Putting the role inside the IT department or treating it as an extension of the chief technology officer, she said, ignores that it needs to consider governance, and the role "needs to be an orchestrator across the C-suite".
In that Q&A she described the first phase of AI in organisations as experimentation, "AI confetti" scattered across the business, and the next as operationalising it: "what scales, what stops, and what changes in the operating model."
Asked in the second part of that interview what belongs inside a government AI office, she set out the five-part model Datacom runs internally and says government would be smart to adopt: a central team running intake and prioritisation of use cases, an academy for capability and change management, a council accountable for investment and risk, a lab for larger opportunities, and an owner for data readiness and running costs. Datacom's separate Australian research puts a clear mandate from the chief executive or executive team at the top of the conditions respondents named for the role delivering.
A government survey counted 272 uses of AI across 70 agencies
New Zealand's 2025 cross-agency AI survey recorded "70 agencies reporting a total of 272 use cases". The year before, 37 agencies reported 108.
The Digital Reset Plan, a rapid review of digital delivery released by Public Service Commissioner Sir Brian Roche on 2 July 2026, records the same total and says "only 20% are deployed and operational and many are duplicate use cases". On the same page it says the government "has AI activity rather than an AI strategy". The review was led by Adrian Littlewood, Matt Crockett and Justin Gray, a former Datacom managing director.
The plan sets out a draft scope for the Government Digital Delivery Agency, under the heading "Draft GDDA scope for discussion. To be finalised post diagnostic and prioritisation phase". One line in that list is an "AI lead" function: "Expert advice and lead on AI strategy and ethics for government". The Public Service Commission said on releasing the review that Sir Brian would consider the findings before making decisions on a way forward.
One AI leadership role is named in the government's own AI guidance. The Government Chief Digital Officer "leads the safe and trustworthy adoption and use of AI in the public service", sets standards and guides agencies on adopting and using it. Those functions moved into the newly established Government Digital Delivery Agency inside the Public Service Commission on 1 April 2026. The Public Service AI Framework issued by that office, last updated 29 January 2025, sets five principles and states that "agencies are encouraged to align with the direction set by this Framework, however it's not binding".
Asked whether the incoming government should introduce specific legislation or controls on the use of AI within government or the public sector, 64 per cent of the organisations Datacom surveyed said yes. Asked whether New Zealand should establish a dedicated national AI framework and co-ordinating body similar to Australia's Office of AI, 74 per cent said yes. Forty per cent are not confident the government is prepared to manage AI's risks and opportunities, against 31 per cent a year earlier. The share who are very confident rose from 6 per cent to 14 per cent.
What to watch
The Commissioner's decision. The Public Service Commission said on 2 July 2026 that Sir Brian Roche would consider the review before deciding a way forward.
The investment principles. 1News reported on 27 July 2026 that Christopher Luxon had tasked Nicola Willis, the Department of the Prime Minister and Cabinet and MBIE with developing a set of principles for hyperscale data centre operators, including an additionality requirement on electricity. No deadline was given and nothing has been published.
The next index. The report says its exploring and implementing figures are largely unchanged on 2025, so the movement to watch is out of those two phases and into scaling. On the supply side, Transpower is assessing about a gigawatt of data centre connection enquiries, Datagrid has started construction on its 280MW Southland campus, and Mercury's chief executive has said data centre demand will outrun what the company can build.