State of AI in New Zealand
Updated: 2 April 2026
Executive Summary
March 2026 did not bring another headline-grabbing public-service rollout on the scale of February’s emergency-department scribe push. Instead, the month shifted the centre of gravity of New Zealand’s AI story toward the underlying machinery: compute infrastructure, research-platform selection, privacy guidance, and enterprise risk management. The clearest new signal was Southland’s Datagrid project moving closer to implementation through full resource consent and a long-term power deal, while MBIE’s national AI Research Platform hit its 31 March phase-two deadline. At the same time, regulators and ecosystem bodies sharpened the trust conversation: the Privacy Commissioner refreshed its AI guidance, InternetNZ reported both very high public usage and high concern, and Kordia’s latest cyber survey showed “shadow AI” becoming a mainstream business-security issue. (rnz.co.nz)
The result is a month that looks serious rather than spectacular. New Zealand’s AI ecosystem still is not showing a flood of local startup capital or frontier-model activity. But it is showing something important: more institutions are now working on the conditions required for scaled adoption, from power and data residency to governance, assurance, and research commercialisation. (mbie.govt.nz)
What Happened in the Past Month
Infrastructure became the strongest new story
The most material new development was Datagrid New Zealand’s move toward implementation of its planned Southland hyperscale campus. In March, the company secured full resource consent for the Makarewa project, which it describes as New Zealand’s first “AI factory,” alongside approval progress for its associated connectivity infrastructure. Datagrid says the site is designed for 280MW of hyperscale capacity and positions Southland as a renewable-powered AI and high-density compute hub. (rnz.co.nz)
That story became more concrete again when Datagrid signed a 15-year, 140MW power purchase option agreement with Mercury for the project’s first phase. Even allowing for execution risk, this is one of the clearest signs yet that New Zealand’s “sovereign AI” conversation is moving beyond vendor positioning and into actual infrastructure, energy, and location decisions. (datacenterdynamics.com)
The national AI research platform entered its decision month
The other important national development was procedural but meaningful: MBIE’s Artificial Intelligence Research Platform reached its phase-two submission deadline on 31 March 2026, with a platform funding decision due in late April and investment decisions scheduled for May. That matters because one of the biggest AI commitments announced in 2025 is now passing from strategy into selection and funding. (mbie.govt.nz)
One visible example of that competition came from the University of Waikato and University of Canterbury, which publicly pitched an “Outdoor AI” proposal for primary industries on 31 March. The consortium said it is one of the five concepts shortlisted from 108 submissions, underscoring that New Zealand’s research push is now narrowing from broad ambition to a small number of nationally backed bets. (waikato.ac.nz)
Health AI remained active, but the month was more about capability-building than a fresh nationwide rollout
Health was still moving, but the signal changed. Rather than another all-system deployment announcement, March showed Health New Zealand building internal capacity to identify and scale AI-enabled delivery. Health New Zealand advertised for an AI Lab Research Fellow to support nationally significant research on safe implementation, and HealthX called for a Clinical Innovation Advisory Group to help identify and scale digital and AI-enabled innovations that reduce workforce pressure and improve access. (jobs.tewhatuora.govt.nz)
That makes March look like a consolidation month for health AI: February’s story was frontline deployment and breast-screening exploration; late March’s story was the quieter work of creating structures that can test, govern, and spread new tools across the system. (beehive.govt.nz)
Privacy, trust, and cyber risk moved from side issue to core operating constraint
The Office of the Privacy Commissioner updated its AI privacy guidance in the past week, broadening the frame beyond earlier generative-AI-specific advice to a wider set of AI tools and automation uses. The guidance stresses privacy impact assessments, accuracy, human review in higher-risk decision-making, and engagement with Māori perspectives. It also explicitly warns that many public-facing AI tools are developed overseas and may not be sufficiently relevant or reliable for Aotearoa New Zealand contexts. (privacy.org.nz)
That tougher governance tone lines up with public sentiment. InternetNZ’s latest Internet Insights release, published on 2 March, said 79% of New Zealanders had used AI in the previous 12 months, but 52% were very or extremely concerned about AI’s impact on society. The top concerns were harmful content such as deepfakes (68%), malicious use (65%), and insufficient regulation or law (64%). In other words, adoption is rising faster than public confidence. (internetnz.nz)
Business risk data is telling a similar story. Kordia’s 2026 New Zealand Business Cyber Security Report, released on 9 March, found 24% of surveyed businesses now rank improper staff use of AI among their top cyber-security challenges, up from 16% a year earlier, while attacks carried out through AI vulnerabilities rose from 6% in 2024 to 14% in 2025. For New Zealand firms, “AI governance” is increasingly becoming a practical security and assurance issue, not just an ethics discussion. (kordia.co.nz)
Private-sector product signals were present, but still selective
There was at least one notable private-sector product signal from a New Zealand-founded listed company. On 10 March, Serko used its investor day to set out its “AI advantage” and introduce serko.ai, a new multi-agent AI offering as part of its longer-term growth plan. Serko said the product would enter a limited US user trial in April, which does not transform the domestic market on its own but does show a local software company treating agentic AI as central to product strategy rather than a side experiment. (nzx.com)
Trend Line Across Recent Snapshots
Compared with the 11 March 2026 snapshot, the emphasis has shifted. The previous month’s strongest signals came from frontline health deployment and AI-specific harm response. This month’s strongest signals came from infrastructure, funding pipeline mechanics, privacy guidance, and cyber governance. That is a real change in texture: less visible service-delivery theatre, more back-end institution-building.
The broader trend still looks like operationalisation. The public sector is continuing to build shared AI machinery; the science system is moving toward a funded national platform; health is building internal AI capability; and private-sector adoption is increasingly tied to security, sovereignty, and implementation discipline. What is still missing is a comparable surge in local venture funding, major new AI law, or a wave of large-scale New Zealand product launches. (digital.govt.nz)
What Looks Quiet, Unchanged, or Early
New Zealand still does not have a new AI-specific law, a specialist AI regulator, or a newly announced blockbuster funding round that would change the commercial landscape overnight. The research-platform process is active, but the actual winner and funding allocations have not yet been announced. (mbie.govt.nz)
Education and leadership development are moving, but mostly in early-stage capability-building rather than system-wide transformation. The Ministry of Education signalled that ERO is surveying schools on how AI is being used, while Otago Business School is marketing what it calls the country’s first board-level AI governance intensive. Those are useful indicators of demand and institutional preparation, but they are still upstream of nationwide change. (education.govt.nz)
It is also worth noting that March’s enterprise story was thinner than its infrastructure and governance story. Serko stood out, but there was no obvious wave of major New Zealand private-sector AI launches to match the attention on Datagrid, MBIE’s research process, or the regulator-and-risk discussion. Evidence of momentum is real; evidence of broad-based commercial breakout is still mixed. (nzx.com)
Overall Assessment
The simplest reading of March 2026 is that AI in Aotearoa New Zealand is becoming more concrete, more infrastructural, and more governed. This was a month when power contracts, research-funding deadlines, privacy principles, and cyber controls mattered at least as much as new end-user tools. That is not as flashy as a national rollout, but it is arguably what a maturing AI ecosystem should look like. (datacenterdynamics.com)
So the national picture is still broadening, but unevenly. Infrastructure and governance are accelerating; health is still active but quieter in public; research is approaching a funding inflection point; and commercial momentum remains selective rather than explosive. If February showed AI arriving on the frontline, March showed New Zealand working on the harder question beneath that progress: whether the country has the power, institutions, safeguards, and research pipeline to support adoption at scale. (rnz.co.nz)