Small Market, Early Movers: AI in New Zealand and the Business Adoption Landscape

AI in New Zealand Today

AI in New Zealand tracks adoption in a small, tight market, where a handful of survey series, not one official number, carry the national picture as of Q3 2025. It spreads like cloud accounting once did through New Zealand’s small firms, pulled in by scarce hands rather than pushed by scale. Labour scarcity, more than market size, is the lens this page uses to read the data.

The headline numbers disagree because their samples differ. Datacom’s 87% comes from a large-firm commercial panel of organisations with 100 or more employees, AI Forum’s 82% comes from a self-selected pool of already AI-engaged respondents, and MBIE’s 67% is a government-cited large-firm figure. None of the three is a representative population measure. Treat each one as directional, scoped and dated, not as “the New Zealand adoption rate.” MBIE published further business-AI research in May 2026, which post-dates the 2025 figures the tables here retain as comparators. Addendum: Datacom’s 2026 State of AI Index, which the comparator tables here do not include, reports 91% of New Zealand organisations using some form of AI, up from 87% in 2025 and 66% in 2024.

Why Small Markets Move Early

Labour scarcity is a leading constraint shaping how New Zealand businesses approach AI, though the cited sources show it as a barrier and a plausible incentive rather than a proven sole cause of adoption or evidence that New Zealand firms adopt earlier than firms in larger markets. A small population intensifies competition for talent, and the same shortage that motivates automation is also its largest barrier.

New Zealand’s population sets a hard ceiling on the domestic pool of AI talent. MBIE’s 2025 AI Strategy states plainly that New Zealand’s small population intensifies competition for AI talent with global markets. The same constraint cuts two ways: it pushes firms toward automation as a substitute for scarce hands, and, as a skills shortage, it holds many firms back from starting at all. Among businesses not yet using AI, 43% cite a lack of expertise as the main barrier, a figure MBIE’s barriers chapter draws from Datacom’s 2024 research. Awareness already outruns understanding: among surveyed New Zealand workers, 97% had heard of AI, yet only 34% felt they could clearly explain what it is, a gap Verian’s 2024 worker survey recorded and MBIE’s strategy repeats. The government frames the shortage as a design choice rather than a weakness, positioning New Zealand as a smart adopter that tailors global AI tools to local problems rather than building foundational models domestically. Among AI Forum’s surveyed organisations, 72% prefer off-the-shelf AI over a custom build, against 13% that build custom, a configuration-first pattern that fits a market short on specialist engineering hours.

Labour Scarcity as the Automation Driver

A small labour pool is the New Zealand market’s defining constraint, and it pushes firms toward automation while, as a skills shortage, it simultaneously holds many of them back.

MBIE states plainly that New Zealand’s small population intensifies competition for AI talent with global markets, and it names a lack of AI skills as a headline barrier, with 43% of non-users citing lack of expertise as their main reason for not adopting. The constraint cuts both ways, driver and barrier at once, which is why the honest reading of the New Zealand market stops short of claiming that its firms automate earlier than firms in larger markets.

The Cloud-Accounting Precedent

New Zealand’s small firms may have run a similar pattern before, when a low-cost software wave lowered the barrier to cloud-accounting adoption; the page treats this as an illustrative analogy, not a measured precedent.

The cloud-accounting shift sets the template rather than the statistic. A wave of low-friction, off-the-shelf software moved through New Zealand’s small firms once the cost of adoption dropped. No New Zealand-specific measured comparison of cloud-accounting adoption speed against other markets exists: an NZIER report commissioned by Xero (March 2020) states directly that “no such statistics are publicly available for New Zealand” and substitutes older Australian ABS data as an illustrative proxy rather than a New Zealand measurement. This page keeps the comparison illustrative for the same reason. NZIER reported in 2020 that the relevant public New Zealand statistics were unavailable, which limits what can be claimed about the pattern’s timing and does not establish that it was never measured privately or since. The directional pattern still shows in the current AI data: 72% of AI Forum’s surveyed firms reach for off-the-shelf AI over a custom build, the same low-friction preference the cloud-accounting wave relied on.

Short Decision Chains, Fast Pilots

Owner-led New Zealand firms plausibly carry shorter decision chains, so a pilot may clear fewer approvals before it starts; this is an operating hypothesis rather than a measured New Zealand finding. As of Q3 2025, 72% of surveyed businesses reach for off-the-shelf AI over custom builds, which shortens time to first use, though speed of decision differs from depth of deployment.

Uptake Across NZ Firms

New Zealand’s firm-size split reads more clearly next to peer markets, where cross-country data benchmarks New Zealand’s uptake against larger markets rather than leaving one country’s figures to stand alone.

The Large-Firm Lead

Large New Zealand firms lead AI adoption. As of July 2025, 87% of organisations with 100 or more employees reported using AI in some form, yet only 12% had scaled it across the whole business, with most still in an exploratory phase. Reach runs wide; depth stays thin.

Datacom’s State of AI Index 2025 records 87% use and 12% rolled out across the entire business, with most businesses still in an exploratory phase, drawn from a sample of 200 senior leaders in organisations with 100 or more employees.

The SME Adoption Gap

A large share of New Zealand’s small and medium firms sit outside AI entirely. MBIE’s 2025 AI Strategy reports that 68% of SMEs have no plans to evaluate or invest in AI, against 67% of large firms already using it, the clearest divide in the national picture.

MBIE cites 67% of large businesses using AI, up from 48% in 2023 (drawn from Datacom’s 2024 research), and 68% of SMEs with no plans (from NZIER/Spark 2024 research), describing the knowledge gap as particularly pronounced among SMEs; government republication does not make these government-origin measurements. This is the datum that rules out any claim that New Zealand’s small firms move onto AI faster than firms in larger markets.

Labour-Driven Automation

The top surveyed AI applications are administration, marketing, software development, project management and design, led by generative tools, according to AI Forum’s research. Across the same surveyed firms, 72% use off-the-shelf AI against 13% that build custom, so automation lands through configuration rather than construction for most New Zealand businesses. As a third-party market signal rather than an synthesised.ai rate, 70% of surveyed firms report AI setup costs under NZ$5,000 (AI Forum NZ, August 2025). No peer-reviewed or government study measures task-level AI automation impact in New Zealand specifically (time saved or error rates per task); the closest available evidence is AI Forum NZ’s Productivity Report (August 2025), which documents illustrative case examples from its surveyed members rather than a systematic measurement. For example, a reported drop from around 8 hours to 20 minutes for a report-preparation task, and from around 6 weeks to 6 minutes for a course-creation task. These are attributed, self-reported examples from named respondents in that survey, not an independent or peer-reviewed measurement, and are not rendered here as a general productivity multiplier.

Because labour constraint decides the first automation target rather than trend, the same reasoning explains how AI workflow automation removes the handoffs that consume scarce hours once a firm moves past a single tool.

Where NZ Businesses Go Next

New Zealand takes a light-touch, principles-based, technology-neutral approach to AI, relying on existing privacy, consumer-protection and human-rights law rather than a dedicated AI Act, and it aligns to the OECD AI Principles, endorsed by 47 jurisdictions including the EU as at the OECD’s May 2024 update. New Zealand published its first national AI strategy on 8 July 2025, the last OECD country to do so. This policy context describes the government’s operating stance, not a certification a business or its AI adviser holds, and no New Zealand regulatory instrument functions as a compliance anchor anywhere on this page. MBIE cites Microsoft-commissioned Accenture modelling projecting generative AI could add around NZ$76 billion in annual GDP by 2038, a conditional projection rather than a measured outcome. Microsoft and Mandala’s June 2025 research projects AI applications and data centre infrastructure reaching a NZ$3.4 billion industry by 2035, with generative AI adoption raising GDP per capita by around 14%. Both figures stay attributed projections rather than realised outcomes. Adoption momentum still runs ahead of trust: KPMG and the University of Melbourne’s 2025 trust survey finds that only 34% of New Zealanders trust AI systems, while 81% believe AI regulation is needed, a gap that bounds how fast the SME adoption picture closes.

New Zealand’s market reads differently once set beside a regulated hub, where the picture contrasts against New Zealand’s light-touch policy stance rather than sitting in isolation; the fuller cross-country benchmarking sits on that page and on the Australian adoption page, not here.

From Market Reading to a Working System

The same market reading compares against the picture across the Tasman, for a reader benchmarking New Zealand’s position next to Australia’s. The AI resources hub catalogues every other adoption and definitional page this page connects to. No pricing figure appears anywhere on this page.