The AI race: can New Zealand carve its own path?

The race for AI supremacy is the new space race, and it's heating up fast.

US President Donald Trump's $500 billion “Stargate” initiative was designed to secure America's lead in AI through high-profile partnerships with OpenAI, Oracle, and SoftBank. But this grand plan faces mounting scepticism following the disruptive entrance of China's DeepSeek - a lean, low-cost AI disruptor creating shockwaves in global markets.

DeepSeek's budget-friendly, supposedly high-performing AI models, developed at a fraction of the cost of US giants like OpenAI, have already wiped $200b off Nvidia's market value and rattled investor confidence. Experts are now asking: is massive spending like Stargate necessary, or is China's more cost-effective approach the future of AI dominance?

Meanwhile, smaller nations like New Zealand are considering their place in the race, and while the world fixates on AI's role in economic power struggles, another battleground on the horizon may be ethics.

In 2024, New Zealand took a bold step by introducing its first AI Bill, tackling concerns like bias, misinformation and job displacement, particularly in critical sectors like tourism and healthcare. Australia, too, is ramping up efforts to regulate AI responsibly.

As tech giants chase AI dominance, countries like New Zealand could become unexpected leaders in setting global standards for ethical AI. Like environmental champions shape global climate policy, countries with effective AI regulations could wield influence in this fast-evolving race.

DeepSeek's rise signals that the next big breakthrough may favour efficiency over expensive moonshots. The question is, who will adapt faster?

Founded in 2023 and based in Hangzhou, Zhejiang, DeepSeek operates under the ownership of the Chinese hedge fund High-Flyer. The company's flagship model, DeepSeek-R1, delivers performance comparable to leading AI models like OpenAI's ChatGPT-4 but with substantially lower training expenses. DeepSeek-R1 was trained with a budget of approximately $6 million, significantly less than the estimated $100m required for GPT-4's training in 2023.

DeepSeek cut costs with innovative training methods, using reinforcement learning to minimise human involvement. This allows the model to self-correct and improve without requiring extensive manual data labelling. This approach not only reduces costs but also accelerates the development process. Additionally, DeepSeek's models are designed to operate efficiently on less advanced hardware, decreasing reliance on high-end chips and further cutting expenses.

The open-source nature of DeepSeek's models has attracted interest from various organisations. Being open source rather than closed means that the source codes and underlying algorithms are available, can be modified and built upon. DeepSeek's cost-effective models democratise access to advanced AI capabilities, enabling a wider range of applications across various sectors.

With DeepSeek's open-source potential, New Zealand can innovate on the shoulders of AI successes and create fairer, more equitable large language models tailored to local needs.

By combining global breakthroughs with local ingenuity, New Zealand could set a standard for small nations leveraging AI to drive growth and inclusion.

Kiwi businesses, especially SMEs in agriculture, logistics and tourism, could adopt cost-effective AI to boost productivity. Imagine AI-driven crop management tools tailored to New Zealand's unique conditions, or smart chatbots offering real-time guidance to tourists, all at accessible prices.

Breaking to build

AI-driven automation may displace traditional roles globally, but it's also creating new industries. A report by Goldman Sachs estimates that 300m jobs worldwide could be disrupted by AI by 2035.

In Australia, sectors like customer service and administrative work are already seeing significant cuts, while new roles such as AI safety officers, prompt engineers and data curators are rapidly emerging.

In the US, tech workers laid off in traditional roles are quickly shifting to AI product management and chatbot customisation.

Meanwhile, China's government-backed AI development means that their AI engineers are being absorbed into public sector projects focused on smart cities and infrastructure.

As innovations dive deeper and become cheaper, nations that invest in skills, infrastructure and ethical AI governance could rise as future powerhouses.

By riding the wave of global AI disruption, New Zealand might use DeepSeek’s success to bolster our own AI innovation—because the future doesn’t wait, and neither should we.