New Opportunities

Context

AI doesn't only displace — it creates new industries, roles, and economic possibilities. The question for NZ is whether we capture these opportunities proportionately, or whether the gains concentrate offshore while we absorb the disruption locally.

The evidence is genuinely mixed. NZ has real competitive advantages in specific areas, but also structural weaknesses — low R&D spending, small domestic market, cultural caution toward AI — that could prevent us from capturing the upside.

NZ companies building with AI

Halter — NZ's standout AI success story. Solar-powered GPS and AI collars for cattle, using a proprietary algorithm trained on 7 billion hours of animal behaviour data. Raised $220M in Series E funding (March 2026) led by Peter Thiel's Founders Fund at a $2 billion valuation — the largest venture capital raise in NZ history. One million collars deployed across 2,000+ farms in NZ, Australia, and the US. Founded by Craig Piggott, a former dairy farmer. Halter doesn't replace farmers — it augments them. One farmer can manage cattle movements that previously required multiple workers and physical fencing.

Xero — Wellington-founded, 4M+ subscribers globally. Actively deploying AI across the product: JAX (an AI financial agent for automated reconciliation, invoice drafting, and natural language queries), ML-powered transaction categorisation, anomaly detection, and predictive cash flow. Named a Leader in the IDC MarketScape for AI-Enabled Small Business Finance 2025–2026.

Volpara Health — AI for breast cancer screening. Analyses breast density, image quality, and workflow optimisation. Regulatory-approved and used internationally. Clinics using Volpara screen more patients per day with the same radiologist workforce — a clear example of AI augmenting rather than replacing skilled professionals.

LawVu — AI-powered in-house legal operations platform. Customers include Deloitte, PwC, and Microsoft/LinkedIn. Automating contract review, matter management, and legal spend analysis.

Datagrid — Building NZ's first "AI factory" at Makarewa, Southland. A 280MW hyperscale campus for GPU clusters. US$2 billion ($3.5B NZD) project with resource consent granted March 2026, construction expected to start June 2026, first phase operational by 2028. Would be NZ's second-largest electricity user after Tiwai Point. Over 1,200 construction jobs and approximately 50 full-time roles per 100MW in operation.

NZ's competitive advantages

Renewable energy — NZ's electricity grid hit 96.4% renewable in Q4 2025, a record. The full-year average is 85.5%, among the highest in the world. The government targets 100% renewable by 2030. This gives NZ a genuine structural advantage for AI compute, which is energy-intensive. Competitive industrial renewable pricing (NZD $90–120/MWh) is in line with Nordic markets and below EU benchmarks. Untapped geothermal potential adds a further 21 TWh from conventional resources plus 30 TWh from supercritical resources.

A BCG report (February 2026) found NZ could unlock NZD $70 billion in economic activity from data centre exports by 2035 in a base case of 600MW new capacity, rising to $160 billion in a high scenario. NZ has a "genuine right to win" in the global AI infrastructure buildout. Southland's average annual temperature of approximately 9.8°C provides natural cooling advantages over Australian and Asian data centre hubs.

Agricultural R&D — NZ has world-class domain expertise in precision agriculture, fisheries monitoring, biosecurity, and conservation ecology. These are areas where AI application requires deep local knowledge, not just compute power.

Te reo Māori and indigenous knowledge — NZ is one of the few places in the world capable of building AI tools for te reo Māori because the training data and cultural context for evaluation both live here. This is not a niche concern — it represents genuine IP that no other country can replicate.

Small, agile regulatory environment — NZ's small economy means fewer vested interests to block change. Regulatory reform can move faster than in the US or EU, potentially positioning NZ as a "trusted AI jurisdiction" for global services.

Weaknesses to be honest about: NZ spends approximately 1.3% of GDP on R&D versus the OECD average of 2.7%. The domestic market is small. Only 18% of New Zealanders trust companies to use AI responsibly, versus 29% globally. NZ was the last OECD country to establish a formal AI strategy (July 2025). Cultural caution is a real barrier to capturing the upside.

Augmented roles

Not every AI impact is displacement. Many roles become more productive rather than redundant:

Agriculture — Halter's system shifts the farmer's role from manual herding to smartphone-based herd management. The farmer is still essential; the nature of the work changes.

Accounting and SMEs — AI scanning bank transactions to identify tax-deductible expenses has reduced accounting workload by 80% for over 5,000 NZ businesses. Xero's AI features automate bookkeeping tasks. The accountant's role shifts from data entry to advisory — but the total number of accountants needed declines.

Healthcare — Volpara's AI assists radiologists, enabling the same workforce to screen more patients. AI decision support in medication management has reduced errors by 40% in trials.

Broad adoption — 93% of NZ firms using AI report improved worker productivity. 72–73% use off-the-shelf tools (ChatGPT, Copilot, Gemini) rather than custom solutions. Only 13% have built custom AI.

The augmentation story is real but comes with a caveat: "more productive per worker" often means "fewer workers needed for the same output." Augmentation and displacement are not opposites — they are two sides of the same coin.

Research strengths

NZ cannot compete for dominance in foundational AI research — training frontier models requires capital and talent concentrations that small economies do not naturally produce. But NZ has genuine comparative advantages in applied AI research:

University of Waikato AI Institute — Created WEKA, the most popular open-source machine learning tool in Java, with over 10.5 million downloads and citations in 18,000+ publications. Practical NZ applications include dairy herd optimisation, soil chemistry prediction, and bioinformatics.

Te Hiku Media — Built te reo Māori automatic speech recognition with 92% accuracy using NVIDIA NeMo. Crowdsourced 300+ hours of labelled speech data in 10 days from 2,500+ volunteers. Established the Kaitiakitanga licence — a data sovereignty protocol ensuring Māori community control over their language data, now a global model for indigenous data governance. Papa Reo, a 7-year MBIE-funded national data science project, is building a multilingual NLP platform for indigenous communities.

Cacophony Project and DOC — AI cameras with machine learning that automatically identify invasive predators (rats, stoats, possums, hedgehogs, cats), detecting 3–20x more predators than conventional trail cameras. Supports the Predator Free 2050 national goal. DOC's Tools to Market programme allocates $1M annually to develop new predator-free tools.

The data centre opportunity

The convergence of NZ's renewable energy, cool climate, and political stability creates a genuine infrastructure opportunity. Microsoft opened its NZ North data centre region in late 2024. Google and AWS are building hyperscale facilities around Auckland. Datagrid's $3.5B Southland campus would be purpose-built for AI workloads.

But NZ should be deliberate about the terms. As the data sovereignty section of this document argues, becoming a data centre hub on terms that benefit offshore investors at the expense of grid stability or carbon budgets is not an opportunity — it is a subsidy. The value of NZ's renewable energy for AI compute should accrue primarily to NZ.

Technology as an export industry

Technology is already NZ's third-largest export sector at approximately 11% of total exports. The top 200 tech exporters generate over $10.5 billion in revenue. ICT is projected to become NZ's leading export industry. NZ has 98,290 ICT professionals across all sectors, including 33,250 software developers and programmers.

Microsoft and Mandala Partners estimate the NZ AI tech stack could generate $3.4 billion in revenue by 2035, with AI applications expected to account for $2.1 billion of that figure.

The honest assessment

The opportunities are real but conditional. They require:

• Active investment in R&D (currently below OECD average)
• Regulatory frameworks that enable rather than obstruct responsible AI
• Workforce development that moves at the speed of the technology
• Deliberate policy to ensure gains accrue domestically rather than flowing offshore
• Overcoming cultural caution that sees NZ lag global adoption rates

The risk is that NZ captures the disruption (job displacement in services) without capturing the upside (new industries, productivity gains, export revenue), because the upside requires proactive adoption that NZ's institutional and cultural caution may delay.

Questions for contributors

1. What NZ AI companies or projects are we missing?
2. Are there examples of NZ workers or firms using AI to compete for international work they couldn't access before?
3. How should NZ balance the data centre opportunity with energy sovereignty and grid stability?