New Zealand business has an AI governance problem, and it is not the one most directors think they have. The risk is not that firms are moving too slowly. It is that they are moving quickly without a plan, and the gap between the two is starting to show up in results.
AWS research cited in this year’s Deloitte Top 200 coverage puts the core number bluntly. Half of NZ businesses now consistently use AI, up from 42% in 2025, but only 15% have a formal AI strategy. Adoption is racing ahead. Governance is barely out of the blocks.
Two surveys, one uncomfortable answer
If this were a single vendor’s statistic, it would be easy to wave away. It isn’t. Datacom’s separate 2026 State of AI Index found 91% of NZ organisations use some form of AI, up from 66% in 2024. Plenty of them claim to have “some form” of strategy. Strip that vague category out and Datacom finds just 16% with a clearly defined standalone AI strategy, and only 22% with a dedicated AI leadership role.
Two research houses, different methods, the same answer. Roughly one in six firms has actually decided what it is doing with AI. The rest are improvising.
Agents are already acting, not advising
The sharpest tension sits in the Herald’s Mood of the Boardroom survey. It found 55% of respondents already using AI agents, software that executes whole workflows rather than helping with individual tasks, with a further 38% considering it. A third of CEOs list adopting AI effectively among their top three worries.
Put those numbers side by side. The most autonomous form of AI, the kind that takes actions on behalf of the business, is already inside most large organisations. A documented framework for how it should behave exists in a small minority. That is not an IT oversight. It is an unpriced liability sitting on the balance sheet.
Dame Therese Walsh, chair of Air New Zealand and ASB, says boards are registering that AI “could impact trust which is an organisation’s most important asset”, and that governing it touches strategy, operating models and capital allocation. She also warns that simply inserting AI into existing processes is unlikely to deliver its benefits.
Strategy is where the money is
This is not a compliance argument dressed up as business advice. The AWS data shows firms with a formal strategy are more than twice as likely to reach advanced AI use (41% versus 19%) and nearly twice as likely to confidently measure return on investment (63% versus 34%).
Size is not the dividing line. Startups lead on advanced use, ahead of large enterprises, with SMEs trailing. AWS New Zealand country manager Manuel Bohnet describes an emerging two-tier economy, warning that “a small group of AI leaders with the strategy, skills, and investment in AI is pulling significantly ahead of the rest of the market”.
The local examples make the point. Bohnet cites Zuru analysing 20,000 social media videos a day to cut product development from 18 months to five, and Tower deploying an AI contact centre in 10 weeks that has saved more than a million minutes of customer call time. Those are redesigns, not bolt-ons.
PwC’s advice to directors captures the balance: “AI is not an IT issue, it is a business one,” but “caution cannot become inertia.”
Don’t answer this with a 40-page policy
The wrong response is a governance freeze. Back in 2025, the Institute of Directors cautioned boards against letting perfectionism delay action, and that still holds. The government has deliberately taken a light-touch, principles-based approach rather than AI-specific legislation. That is the right call, but it means nobody in Wellington is going to write your guardrails for you. Responsibility sits squarely with boards.
The practical starting point is simpler than most consultants will admit. Ethos Advisory notes that AI exposure builds gradually through staff tools and team experiments, often “without deliberate organisational oversight”, and poses one question every director should ask at the next meeting: what AI use across this organisation would we expect to know about?
If the board cannot answer that, it does not have a strategy. It has a collection of experiments, some of which are now making decisions on its behalf.
The gap will close one way or another
The 85% without a plan are not doomed. But the window in which improvisation was a reasonable stance has shut. The early movers are now compounding their advantage in capability, data and measurable returns, and the agents are already in production. Firms will either close the strategy gap deliberately in the next year or have it closed for them by a competitor, a customer complaint or a regulator asking awkward questions after something goes wrong.
Sources
- NZ Herald: Deloitte Top 200: AI decision-making and the art of becoming a great company (2026-11-18)
- Reseller News: Two-tier AI economy is emerging, AWS warns (2026-09-25)
- NZ Herald: Mood of the Boardroom survey: AI ranks second among CEOs’ concerns (2026-10-22)
- About Amazon: Kiwi businesses are turning AI ambition into impact one year in (2026-09-22)
- NZ Herald: Therese Walsh: AI is a boardroom issue, not just a technology problem (2026-10-24)
- PwC: Governing AI without holding it back (2026-06-16)
- Ethos Advisory: When AI use requires an organisation level decision (2026-10-16)
- Institute of Directors: Getting started with AI: why the first conversation matters most (2025-04-14)
- NZ Government: New Zealand’s strategy for artificial intelligence: Investing with confidence (2025-07)
Join the discussion
Add useful context, ask a good question, or challenge an idea — keep it specific and respectful.