Norway's Data Center Problem

In July 2025, OpenAI announced they're building a massive AI data center in Narvik. 500 MW, 100,000 NVIDIA GPUs, eventually scaling to 250,000. First AI data center in Europe. Sounds great, right?

Maybe. But here's what's strange: Norway has these intense political debates about offshore electrification. People argue endlessly about whether we should use our electricity to power oil rigs. But when OpenAI shows up wanting 500 MW, about the same as many offshore projects, nobody seems to care.

Why the asymmetry? Electricity is limited. And according to Sam Altman himself, future LLM costs will converge to the price of electricity. So power isn't just an input. It's becoming the bottleneck. If anything, that makes it more valuable, not less.

The Promises

What does Norway get out of this? OpenAI's commitments are vague:

• "Collaborate with countries to build data center capacity"
• "Deliver customized ChatGPT to residents"
• "Prioritized access" for Norwegian researchers and startups

But what does "access" actually mean? At what price? With what guarantees? These are just words in a press release. I suspect they'll be distant memories once the facility is operational.

You don't need local data centers to deliver customized ChatGPT. An AI safety center could handle that requirement just fine. The data center is about compute, not compliance.

How This Happened

The most surprising thing I learned at an NTNU - Energy Transition Initiative, debate in late October: data center decisions are apparently made at the local level, not national. Municipalities negotiate these deals individually with companies.

Think about that. We're talking about infrastructure that affects national energy strategy and digital sovereignty. But it's being decided municipality by municipality, by people who probably lack the leverage to demand better terms.

If I heard that right (and I hope I didn't), we should expect more deals like this across Norway in the coming years.

Why Data Centers Are Different

Someone at the NTNU debate proposed a specific tax on data centers, calculated on computational operations (FLOPS) rather than just electricity usage.

This makes sense. Data centers don't consume electricity the way normal companies do. Regular businesses need power for heating, lighting, machinery: energy is an input to their work. For AI data centers, electricity essentially is the product.

The big tech companies will say they're just like any other business. But they're not. Data centers create few local jobs. Most value gets generated by software developers working anywhere in the world. And the whole point of building in Norway is to get cheap electricity.

A FLOPS-based tax would:

• Tax the computational capacity these facilities export globally
• Be legally distinct from regular business taxation
• Capture appropriate rent from Norway's comparative advantage

Those governing Norwegian energy policy should embrace data centers, but as their own category of business with their own tax structure. Otherwise we're just watching our power advantage flow to California shareholders while municipalities compete to offer the sweetest deals.

The Window Is Closing

Norway's position is good. We have renewable energy, cooling climate, and industrial expertise. These are real advantages.

But the question isn't whether to participate in AI infrastructure. It's on what terms.

Right now, while facilities are under construction, we have leverage. Companies need our cooperation. Once they're built and operational, that leverage disappears entirely.

The difference between getting this right and getting it wrong will come down to whether we can move from vague promises to concrete commitments. From local negotiations to national strategy. From treating data centers like any other business to recognizing what they actually are.

We need written commitments about what "access" means. Real pricing. Capacity guarantees. Governance structures. Not someday, but now, before the concrete is poured.

Otherwise this becomes a cautionary tale about selling critical resources too cheaply. And we'll have plenty of time to regret it while watching our electricity power Silicon Valley's next trillion-dollar industry.