NZ’s AI Report Card: 6 Numbers That Tell a Bigger Story

Stat: 75% of firms now report setup costs under $5,000, with 58% spending less than $1,000. On the ongoing side, over 70% of firms spend less than $5,000 annually on AI operations. Only 4% now spend over $50,000 a year, compared to ~30% last year.

Insight: On the surface, this looks like a huge positive. AI is no longer reserved for corporates with deep pockets, it’s accessible to SMEs and startups alike. Off-the-shelf tools and subscription models mean nearly every business can now get a foot in the door.

My read: Let’s be real: most businesses can't become an AI-native business for $5,000 a year, the equivalent of $600 a month. That level of spend might cover an AI tool or two alongside the spend on the productivity workbench (Microsoft 365 or Google Suite), but it’s nowhere near enough to fund the tooling, integration, and workflow redesign required to truly outcompete in an AI-first world. And it’s worrying that such a significant volume of businesses are spending less than this each year. What’s happening in practice is most firms are dabbling, ticking the “AI adoption” box, and moving on. That’s fine for accessibility, but dangerous if leaders think the job ends there. Buying the tool is the start, not the finish line.

Stat: Efficiency improvements sit at 91% of firms, essentially unchanged from the 93% reported earlier this year.

Insight: The easy wins are believed to be complete according to respondents. Commodity AI tools like copilots and chatbots have delivered their first wave of time savings, but they’re not pushing the needle further.

My read: Firstly, I'm not convinced most leaders even know whether their efficiencies have improved. Very few from what I've talked to leaders about are measuring productivity with any rigour. Much of this 91% figure likely reflects perception bias: AI makes people feel faster, but without baselines or output tracking, it’s sentiment, not science. And even when efficiency does improve, it’s often doing the wrong tasks faster, which doesn’t equal effectiveness. This plateau also reveals the next constraint: integration debt. Workflows are still clunky, with double-handling, siloed data, and shallow integrations. What's counterintuitive is that efficiency is already maxed out, therefore shouldn’t financial impact be climbing? Instead, as we’ll see in the next section, it’s actually falling. That’s the warning sign we’re hitting the ceiling of shallow adoption, and surface-level gains that look good in surveys don’t translate into P&L impact.

Stat: Positive financial impact dropped from 56% to 50%, even as cost savings rose from 71% to 77%.

Insight: This is a split story. On the one hand, AI is proving itself as a cost-cutter. On the other, it’s not delivering the revenue growth that should come with higher efficiency.

My read: This paradox is interesting because if efficiency is maxed out at over 90%, why is financial impact dropping? The answer: firms are capturing savings but not reinvesting them into growth or differentiation. Leaders might be happy to pocket operational savings, but the hard work of using AI to open new markets, drive product innovation, or create unique experiences isn’t being done. I hope the answer is that it's yet to come and this is just where we are at in the cycle. But this is the danger of treating AI like a software subscription: it keeps the lights on, but it doesn’t move the company forward.

Stat: 14% of organisations now report job losses due to AI, up from 7%. At the same time, 45% say they need fewer new hires, but 55% see new opportunities emerging.

Insight: AI is clearly reshaping workforce demand faster than most expected. While job losses are climbing, the bigger story is the reduction in new hiring. Firms are stretching their current workforce further with AI, rather than expanding headcount.

My read: The reports frame this as partly a recessionary effect, but that underplays the structural shift. What we’re seeing is a “sinking-lid” approach to staffing: instead of replacing departing staff, businesses are upskilling existing employees with AI. That means fewer net jobs, but also entirely new types of roles. The winners won’t be those who set up centralised “AI teams", rather they’ll be the ones embedding AI operators and champions inside every function so that teams can redesign workflows in place, not bolt on another silo. the next frontier for organisations is to redesign their organisational structure and capabilities of their roles. A first step for inspiration is to check out Wade Foster's (CEO of Zapier) post on how Zapier's role definitions are shifting in the age of AI.

Stat: Only 44% of New Zealanders believe AI’s benefits outweigh the risks, which is the lowest trust rating globally. Māori and Pacific communities are particularly sceptical.

Insight: Adoption inside businesses is racing ahead, but public confidence is lagging badly. Without social licence, scaling AI will hit a wall.

My read: Clearly, the back-office of AI focus (finance, admin, compliance) isn’t making its way into the customer experience layer. Marketing, communications, and service are lagging in showing customers how AI is being used responsibly. This is a huge risk. Without visible, transparent comms, the trust gap could become the real bottleneck to adoption rather than the cost or efficiency.

Stat: Earlier this year, 72–73% of firms used off-the-shelf AI, with only 13% using custom solutions. The latest report shifts the narrative to “purpose-built AI” as the next edge.

Insight: The message is clear: you won’t differentiate by buying the same tools as everyone else. The edge lies in applying AI to proprietary data and industry-specific workflows.

My read: The rhetoric has moved, but the reality hasn’t caught up. Most firms are still leaning heavily on generic copilots and commodity products. This sets up a looming “AI capability divide”: companies that treat AI as a checkbox vs. those that invest in redesign and customisation. The former will plateau (or become obsolete), and the latter will build sustainable advantage.

The AI honeymoon is over. The novelty of chatbots and copilots has passed.

Costs aren’t the constraint. With setup costs near zero, the real bottleneck is creativity, execution and redesign.

Efficiency is maxing out. The only way forward is to rewire work patterns, not just plug in tools.

The workforce is shifting. AI is reducing new hiring but creating new roles inside functions, the challenge is redesigning jobs, not just saving them.

Trust is the Achilles’ heel. If customers don’t believe in the benefits, adoption will slow no matter how fast firms move internally.

Differentiation requires purpose-built. The era of buying generic copilots is tabele stakes. The next wave belongs to those who design around their data, their industry, and their processes.