The Numbers Don’t Lie — And They’re Not Comfortable Reading for Wellington
Nearly half of Asia Pacific governments are now actively evaluating Sovereign AI technologies, and more than a third are already running proofs of concept. That’s not a forecast — that’s the finding from a Dell Technologies and IDC survey of 360 government IT decision-makers across eight APJ markets, released May 13, 2026.
Sovereign AI jumped from the seventh to the second-highest government investment priority in just one year. 76.9% of government leaders say investing in sovereign AI enhances resilience against geopolitical risks and supply chain disruptions.
And then there’s New Zealand.
What is Sovereign AI?
Sovereign AI is a nation’s ability to develop, deploy, and control AI systems within its own borders — keeping data, compute infrastructure, and governance under domestic jurisdiction rather than depending on foreign providers. It covers everything from running local data centres to training models on national datasets to ensuring AI decisions can be audited under domestic law.
The concept has moved from academic to urgent as countries realise that relying entirely on US or Chinese cloud providers for AI infrastructure creates the same strategic vulnerability as depending on foreign satellite networks for communications.
The APAC Sprint
The regional picture is startling:
- 46.1% of APAC governments are actively evaluating sovereign AI technologies
- 36.1% are running initial proofs of concept
- Only 3.1% are investing significantly — but only 1.7% have no plans to adopt at all
- 99% see agentic AI as an accelerator for government adoption
- Nearly 9 in 10 APJ government organisations report critical digital skills shortages
The preferred model? “Selective sovereignty” — hybrid deployments that keep sensitive data and critical workloads onshore while leveraging global technology ecosystems for innovation and scale. Not full autarky, but strategic control over what matters.
This isn’t theoretical. NEXTDC just launched its KL1 data centre in Kuala Lumpur — a AUD$1 billion, 65MW AI-ready facility built to Tier IV standards, specifically designed for sovereign AI workloads in Southeast Asia. Malaysia’s Minister of Digital called it a milestone in making Malaysia “the premier digital hub of Southeast Asia and a key driver of the regional AI economy.”
And NZ? The Blueprint That Missed the Point
Last week, the AI Forum of New Zealand released its AI Blueprint for Aotearoa — a “refreshed national programme of work” with a vision to 2030. It talks about five strategic pillars: opportunities, innovation, adoption and risk, talent, and global reach.
What it doesn’t talk about — at all — is sovereign AI infrastructure.
The Blueprint describes NZ as sitting in an “uncomfortable position: high-use, low-trust.” It acknowledges agentic AI is moving “faster than almost anyone predicted.” It mentions quantum computing timelines collapsing.
But when it comes to the question every APAC neighbour is now building data centres to answer — where does our AI compute run, and who controls it? — the Blueprint offers silence.
We’ve been here before. As we wrote in NZ’s Pollyanna Policy, the pattern is familiar: upbeat language, broad vision, no concrete infrastructure commitment.
Why This Matters for NZ
New Zealand’s position is uniquely exposed:
-
Geographic isolation. We can’t physically plug into the Asian sovereign AI infrastructure being built in Singapore, Malaysia, and Japan without latency and sovereignty compromises. As we explored in Undersea Cables at Risk, our digital lifelines are fragile.
-
Small market, big dependency. NZ organisations overwhelmingly use US-based cloud providers (AWS, Azure, Google Cloud). That’s fine until it isn’t — ask any European government currently rewriting procurement rules.
-
The talent gap is real. The IDC study found the hardest-to-hire roles in APAC are AI safety researchers (42.5%), data architects (35%), and sovereign data governance specialists (30%). NZ is competing for the same people with a fraction of the budget.
-
The French model works. France’s €2.73B Q1 2026 funding surge (a 79% increase) shows what sovereign AI investment looks like when a mid-sized country gets serious. As we covered in French AI Sovereignty, it’s paying off.
The Real Question
The APAC sovereign AI race isn’t about nationalism — it’s about leverage. Countries that control their AI infrastructure control their economic future. They can set terms with big tech, protect citizen data under their own laws, and ensure critical services don’t get switched off during a geopolitical dispute.
NZ doesn’t need to build its own foundation model. But it does need an answer to the question: when the sovereign AI infrastructure maps of the Asia Pacific are drawn, where does Aotearoa sit?
Right now, the answer seems to be: dependent on everyone else’s infrastructure, hoping they’ll keep the lights on.
That’s not a strategy. That’s a prayer.
❓ Frequently Asked Questions
Q: What does sovereign AI actually mean for an average NZ business? It means whether your customer data, financial records, and AI models are processed on servers you can audit under NZ law — or on servers in Virginia that are subject to the US CLOUD Act. For most businesses, it’s invisible until something goes wrong.
Q: Isn’t NZ’s AI Blueprint addressing this? The Blueprint addresses adoption, trust, and talent — important areas. But it doesn’t include sovereign infrastructure as a strategic pillar, which is the gap every APAC neighbour is now filling. It’s like writing a transport strategy that covers driver training but not roads.
Q: What should NZ do? Three things: (1) Mandate sovereign data residency for government AI workloads, (2) incentivise local compute infrastructure investment (the Southland data centre proposal we covered in NZ Data Centre Haven is a start), and (3) negotiate data sovereignty provisions in trade agreements before they’re locked in without them.
🔍 THE BOTTOM LINE
Half of APAC governments are building sovereign AI as strategic infrastructure. NZ’s new Blueprint talks about trust and talent but skips the compute question entirely. In a region racing to control its own AI destiny, that silence is deafening.
Sources
- Dell Technologies / IDC Sovereign AI Study (APJ), May 2026
- Vietnam Investment Review — Asia Pacific governments prioritize sovereign AI
- AI Forum NZ — AI Blueprint for Aotearoa, May 2026
- NEXTDC KL1 Kuala Lumpur Launch, May 2026
