New Zealand's economy runs on exports. Dairy, wine, forestry, tech services. These industries employ hundreds of thousands of people and generate the foreign exchange that keeps the country running. AI is starting to change how some of them operate. Here's where it's real and where it's still marketing.
What You Need to Know
- Dairy is the most advanced NZ export sector for AI adoption, driven by Fonterra's scale and the sector's data-rich operations. Herd management, processing optimisation, and supply chain prediction are in production or advanced pilot.
- Wine is using AI for precision viticulture, but adoption is fragmented. Larger producers are investing. Smaller estates are watching.
- Forestry has the most untapped potential. Massive datasets from satellite imagery, LiDAR, and operational sensors. Limited AI adoption so far.
- Tech services (NZ's fourth-largest export) face a different AI question: not "how do we use AI in our operations?" but "how do we embed AI in what we sell?"
- Across all sectors, the common barrier is not technology. It's the gap between data collection and data usefulness. NZ export industries collect enormous amounts of data. Most of it isn't in a state where AI can use it.
Dairy: Leading the Pack
New Zealand's dairy industry is, by necessity, one of the most efficient in the world. We produce more dairy per person than almost any country, and we export over 95% of it. Efficiency isn't a strategy. It's a survival requirement.
That efficiency imperative has made dairy the most advanced adopter of AI among NZ's export sectors.
What's working:
Herd management. AI systems that monitor individual cow health, predict calving, detect lameness, and optimise feeding schedules. These systems combine sensor data (wearables, milking station sensors) with historical records and weather data. The ROI is clear and measurable: healthier cows produce more milk and require fewer interventions.
Processing optimisation. Dairy processing plants generate enormous volumes of sensor data. AI-driven optimisation of processing parameters (temperature, timing, pressure) improves yield and consistency. Fonterra has been investing in this for several years, and the results are flowing through to operational efficiency.
Supply chain prediction. Predicting milk supply based on weather, pasture conditions, calving patterns, and historical trends. This helps processing plants plan capacity and helps the cooperative manage its supply chain across thousands of farms.
What's still early:
Market prediction. Using AI to forecast global dairy commodity prices. The models exist, but dairy markets are influenced by geopolitical factors, trade policy, and currency movements that no AI can predict reliably. Useful as one input among many. Not a replacement for market analysis.
Pasture management from satellite imagery. The technology works in trials. Scaling it to thousands of farms with different topography, soil types, and microclimates is harder than the demos suggest.
95%
of NZ dairy production is exported, making efficiency improvements from AI directly relevant to international competitiveness
Source: Dairy NZ, Industry Statistics, 2023
Wine: Precision in the Vineyard
New Zealand wine is a premium export. Our reputation is built on quality, not volume. AI's application in wine is about protecting and enhancing that quality.
What's working:
Precision viticulture. Drone-mounted sensors and satellite imagery analyse vine health, canopy density, and soil moisture at individual-vine resolution. AI processes this data to recommend variable-rate irrigation, targeted pest management, and harvest timing.
Larger producers (the ones with the budget for drones and the data infrastructure to process the imagery) are seeing real results. More consistent grape quality. Lower water and chemical usage. Better yield prediction.
Disease prediction. Botrytis, powdery mildew, and other vineyard diseases are influenced by weather patterns, vine health, and local microclimate. AI models that combine weather forecasts with vineyard sensor data can predict disease risk days before visible symptoms appear, enabling preventive treatment rather than reactive spraying.
What's still early:
Flavour prediction. Using soil, weather, and vine data to predict flavour profiles before harvest. The wine industry is deeply interested in this. The science is not yet reliable enough for production use. Flavour is influenced by factors that current sensors don't capture.
Small estate adoption. Most NZ wineries are small. They don't have data infrastructure, drone budgets, or the technical capacity to implement AI tools. Until the tools become simpler and cheaper (and they will), adoption will be concentrated among larger producers.
Forestry: The Untapped Giant
New Zealand's forestry sector exports roughly $7 billion annually. It generates massive amounts of data: satellite imagery, LiDAR scans, harvest records, logistics data, sensor feeds from processing mills. And yet AI adoption in forestry is notably behind dairy and wine.
The opportunity:
Inventory and growth prediction. LiDAR and satellite data can estimate forest inventory (volume, species composition, growth rate) across vast areas. AI models that predict growth based on soil, weather, and silvicultural history could transform forest management planning. The data exists. The models are feasible. The adoption is slow.
Harvest optimisation. AI-driven optimisation of harvest scheduling, log allocation, and transport logistics. When you're managing harvest across hundreds of stands, shipping to multiple ports, and balancing domestic and export demand, the optimisation problem is enormous and well-suited to AI.
Fire and pest risk. Predicting wildfire risk from weather, fuel moisture, and historical patterns. Identifying pest incursions from satellite imagery before they spread. Both are active research areas in New Zealand.
Why it's slow:
Fragmented data. Forestry data is spread across forest owners, harvest contractors, transport companies, and processing mills. No single entity has the complete picture. Until the data is integrated (or at least interoperable), AI tools that span the value chain are difficult to build.
Conservative culture. Forestry operates on 25-30 year investment cycles. The industry culture favours proven approaches over experimental ones. This is rational given the timeframes but means AI adoption lags behind sectors with shorter feedback loops.
Tech Services: A Different Question
New Zealand's tech services sector (SaaS, consulting, managed services) faces a different AI challenge. These companies don't need AI for their production processes. They need AI in their products.
The question is existential: if your software product doesn't incorporate AI within the next two to three years, will it still be competitive? For many NZ tech exporters, the answer is increasingly no.
What's working:
NZ tech companies that have embedded AI into their products early are seeing stronger growth and better customer retention. AI-powered features (intelligent search, automated analysis, predictive alerts) are becoming expected, not exceptional.
What's hard:
Building AI features requires skills that most NZ tech companies don't have in-house. ML engineers, data scientists, prompt engineers. The talent market is tight globally and acutely so in New Zealand. Many NZ tech companies are partnering with AI-focused firms (hello) to build these capabilities rather than trying to hire directly.
NZ exports depend on being world-class at what we do. The sectors that figure this out first will set the pace for the rest.
Isaac Rolfe
Managing Director
The Common Thread
Across all four sectors, the same pattern emerges: NZ export industries collect enormous amounts of data. Most of that data is not in a state where AI can use it. The gap between data collection and data usefulness is the single biggest barrier to AI adoption across the export economy.
Closing this gap requires investment in data infrastructure: integration, cleaning, standardisation, and governance. It's not glamorous. It's not AI. But it's the foundation that AI requires.
The sectors that invest in their data infrastructure now will be the ones that capture AI's value first. The ones that wait for the AI to "figure out" their messy data will be waiting for a long time.