One Year Later: What Changed After I Wrote to the Government About AI Infrastructure

September 2024:

I sent a letter to Norway's Department of Digitalization warning that our country was falling behind in AI infrastructure. A year later, I wanted to see what actually changed.

The short version? Norway attracted over $20 billion in private AI infrastructure investment. The government published comprehensive strategies. We got regulatory frameworks in place.

But my core concern remains unaddressed: this is foreign-owned infrastructure that happens to be built in Norway, not Norwegian infrastructure for Norwegians. We're providing cheap power and cold weather so OpenAI, Microsoft, and CoreWeave can build their datacenters here. That's not sovereignty—that's hosting.

The government's job should have been ensuring Norwegian enterprises and researchers have easy, affordable access to AI infrastructure. Instead, we're letting foreign companies own critical national assets because Norway is an attractive location. It's like Denmark's housing market—except we forgot to regulate foreign ownership.

AI infrastructure and the knowledge that comes with it is the future equivalent of oil. We're treating it like a real estate opportunity instead of strategic national wealth.

Let me walk through what actually happened.

The Government's Response

A few weeks after my letter, I received an official response. It was... diplomatic.

The Ministry thanked me for my engagement and pointed to the newly published National Digitalisation Strategy. They outlined ambitious goals: national AI infrastructure by 2030, Norway at the forefront of ethical AI use, world-class computing power for Norwegian and Sami language models.

All good objectives. But notice what was missing: any mention of direct government capital investment.

The strategy talks about "working to establish" infrastructure and "investigating" details.

It doesn't commit funding, doesn't propose public-private partnerships with government capital, doesn't announce specific projects.

Compare this to what I proposed in my original letter: strategic investment similar to how we fund railways or telecommunications—infrastructure Norway considers essential for national function. Whether through the Oil Fund, dedicated infrastructure funds, or other mechanisms, the point was that Norway as a country has the financial capacity to fund critical infrastructure when we recognize it as strategic.

We do this for roads, for railways, for telecommunications networks, for energy infrastructure. These aren't left purely to market forces because we understand they're foundational to everything else. We invest in them not just for immediate returns, but for long-term competency, experience, and competitive advantage.

AI infrastructure deserves the same strategic thinking. This is investment in people, industry expertise, and national capability—not just servers and cooling systems.

The government's response was essentially: "We agree this is important, we're making plans, good luck to the private sector."

What Actually Happened: $20 Billion in Private Investment

Here's where it gets interesting. Despite the lack of government capital, Norway attracted massive private investment:

OpenAI Chose Norway for Their First European Datacenter

Stargate Norway, announced July 2025, commits approximately $1 billion initially for 230 MW capacity near Narvik in Northern Norway. By end of 2026, they plan to deploy 100,000 NVIDIA GPUs with potential expansion to 520 MW.

This is OpenAI's first datacenter outside the US. They chose Norway.

Microsoft Followed With $6.2 Billion

Microsoft announced a five-year, $6.2 billion commitment for AI infrastructure in the same Narvik location, partnering with Nscale and Aker. They're deploying 52,000 NVIDIA GB300 GPUs.

This is one of Microsoft's largest European AI investments.

CoreWeave Committed Over $1 Billion

CoreWeave partnered with Bulk Infrastructure for NOK 13 billion (~$1.2 billion) to build what they describe as one of Europe's largest NVIDIA AI clusters—400 MW initially with potential for 1 GW.

Location: Vennesla in Southern Norway.

Google's €600 Million Datacenter Nears Completion

Google's Skien facility, announced earlier but now in final construction phases, represents €600 million investment with 240 MW capacity. Expected operational late 2025 or early 2026.

Why Norway? (The Fundamentals Were Right)

These companies didn't choose Norway randomly. The advantages I outlined in my original letter proved compelling:

Renewable Energy: 98% hydropower means genuinely sustainable AI operations. With AI systems consuming 150 kW per rack, energy sourcing matters enormously for both cost and ESG requirements.

Cold Climate: Natural cooling reduces operational costs significantly. When you're running 100,000 GPUs generating massive heat, ambient temperature isn't trivial.

Political Stability: In an increasingly uncertain geopolitical environment, Norway offers regulatory predictability and minimal political risk.

Grid Capacity: Northern Norway in particular has available power capacity that's constrained in traditional European datacenter hubs like Frankfurt or Amsterdam.

These fundamentals attracted capital even without direct government investment.

The Policy Progress (Impressive, Actually)

While the government didn't commit capital, they did deliver on regulatory frameworks:

1. Datacenter Registration Became Mandatory

Effective January 1, 2025, all datacenters exceeding 500 kW must register with Nkom by July 1, 2025. This isn't just paperwork—it establishes government visibility and control over critical infrastructure.

Operators must disclose power consumption, services offered, and designate physical representatives in Norway. Foreign shell companies can no longer operate datacenters without Norwegian presence.

2. Critical Infrastructure Designation

Datacenters received official "critical infrastructure" status. This means they're subject to security requirements similar to telecommunications and energy infrastructure.

Practically: government can mandate specific security measures, require contingency planning, and impose usage restrictions during national emergencies.

3. Data Centre Strategy 2025 Published

The government released a comprehensive strategy document in October 2025 laying out goals through 2031:

• Employment growing from 4,400 FTE to potentially 25,000 FTE

• NOK 28 billion in value creation

• Power consumption trajectory: 2.5 TWh (2025) → 6 TWh (2030) → 10 TWh (2040)

The strategy includes requirements for waste heat utilization (facilities over 2 MW must analyze heat reuse) and establishes datacenters as "fundamental national

functions" for security purposes.

4. NIS2 Implementation Underway

Norway implemented the Digital Security Act (Digitalsikkerhetsloven) on October 1, 2025—transposing NIS1 requirements with NIS2 following in 2026.

This establishes mandatory incident reporting (24-hour preliminary notifications), legally binding leadership responsibility, and penalties up to 4% of global turnover.

For datacenters, this means security isn't optional—it's legally mandated with serious consequences for failures.

The Scale Question (Still Unanswered)

Here's where my original concern persists.

Yes, $20+ billion in announced private investment is impressive. But "announced" isn't "deployed," and total potential capacity still raises questions.

The Math

If all announced projects reach maximum capacity, Norway could approach 1,400 MW total datacenter capacity. Current operational capacity is approximately 250-300 MW.

That's roughly 5x growth—substantial, but is it enough?

McKinsey projects global datacenter capacity growing from 60 GW currently to 171-219 GW by 2030. That's 3-4x growth in six years.

If Norway maintains proportional share of European capacity and aims to be "the most digitalized country in the world by 2030" (government's stated goal), 1,400 MW potential may prove insufficient.

Especially considering execution risk: not all announced projects will fully deploy on stated timelines.

The 2030 Timeline Concern

The government's National Digitalisation Strategy targets 2030 for comprehensive AI infrastructure. That's five years from now.

Ambitious goals include:

• 100% of government agencies using AI (currently 43%)

• Norwegian enterprises leading Nordics in AI adoption

• National infrastructure for Norwegian/Sami language models

Five years sounds reasonable until you consider:

• Google's €600 million facility takes 2+ years from announcement to operation

• Microsoft's $6.2 billion commitment "starts 2026" over five years

• OpenAI Stargate targets "end 2026" for initial deployment

Major projects require 2-3 years from announcement to full operation. If we're announcing projects now for 2026-2027 completion, what gets announced in 2027-2028 for 2030 readiness?

The timeline feels optimistic unless there's significant acceleration.

What About Nordic Competition?

While researching this article, I looked at what Norway's neighbors are doing. The comparison is instructive.

Sweden secured even larger commitments:

• Brookfield Asset Management: $10 billion for Strängnäs AI Center (750 MW)

• Microsoft: $3.2 billion including training for 250,000 people in AI skills

• Total Swedish announced investment exceeds $15 billion

Finland took a different approach:

• LUMI AI Factory: €250 million Finnish government + €306.4 million EU funding

• Direct government capital investment in research infrastructure

• Operational by 2027 with free access for startups and researchers

Denmark:

• Novo Nordisk Foundation invested DKK 600 million (~$87 million) in Gefion supercomputer

• Foundation funding (non-government) but represents institutional capital commitment

• Focused on life sciences and pharmaceuticals AI applications

Notice a pattern? Finland committed €250 million in direct government capital. Denmark secured foundation funding. Sweden's government partnered strategically with state-owned Vattenfall for power supply.

Norway? Zero direct government capital despite having a $2 trillion Oil Fund.

The Missing Piece: Government Capital

This is the core issue I raised in my original letter, and it remains unresolved.

The Oil Fund holds massive stakes in Microsoft, NVIDIA, Alphabet, Amazon, and other AI infrastructure companies globally. It profits enormously from the AI boom—recording $222 billion profit in 2024 alone.

Yet it invests nothing directly in Norwegian AI infrastructure, even as government strategies declare AI sovereignty a national priority.

I'm not suggesting the Oil Fund should fund everything. Its mandate is global diversification, not domestic infrastructure. But the disconnect is striking:

Strategy documents say: AI infrastructure is critical to national competitiveness and security.

Budget reality says: Private sector figures it out, government provides regulatory framework.

Compare to Luxembourg's model (which I cited in my original letter): public-private partnerships where government retains ownership and leases capacity to trusted operators. Or Finland's direct government funding for LUMI.

Norway chose pure private sector approach, relying on energy advantages and regulatory environment to attract capital.

It worked—for now. But it means we're dependent on foreign companies' strategic decisions rather than controlling our own AI infrastructure destiny.

What This Means for Norwegian Enterprises

For enterprises making cloud and AI strategy decisions in 2025-2026, here's what the last year tells us:

1. Sovereign Infrastructure Options Are Emerging

If your compliance requirements demand data residency in Norway, options are expanding. Skygard (Telenor, Hafslund, HitecVision joint venture) provides Norwegian-owned capacity. OpenAI, Microsoft, and CoreWeave facilities will be physically in Norway even if foreign-owned.

The critical infrastructure designation means these facilities are subject to Norwegian security oversight and emergency control measures—not purely at foreign operators' discretion.

2. AI Compute Capacity Will Be Available (Eventually)

If you're planning significant AI workloads, Norway will have capacity. 100,000+ NVIDIA GPUs coming online by end 2026 (OpenAI) plus Microsoft and CoreWeave deployments mean compute availability improves dramatically.

Question is timing. If you need capacity now, you're still dependent on existing Azure/AWS/GCP. If you can wait until 2026-2027, more options emerge.

3. But Don't Assume Government Will Subsidize

Unlike Finland's LUMI (free for researchers and startups) or Denmark's Gefion (foundation-funded for life sciences), Norwegian infrastructure is commercially operated. Don't expect subsidized access or government grants for compute time.

If you're a startup needing AI infrastructure, you'll pay market rates or need to look at EU/Nordic programs rather than Norwegian government support.

4. Energy Pricing Remains Advantage

Even without government capital, Norway's renewable energy pricing provides operational cost advantage. If you're comparing TCO for sustained AI workloads, Norwegian facilities' energy costs run 40-50% below most European alternatives.

For workloads you control (not SaaS), this matters significantly at scale.

The Sovereignty Problem Nobody's Talking About

Here's what bothers me most about the last year's developments.

Yes, Norway attracted $20 billion in AI infrastructure investment. But let's be clear about what that actually means: foreign companies are building their infrastructure on Norwegian soil because we have cheap power and cold weather.

That's not the same as Norway building AI infrastructure for Norwegians.

And this isn't just about national pride or abstract sovereignty concerns. Just this week—October 20th, 2025—AWS experienced significant service disruptions affecting customers globally. Before that, we had the Microsoft Azure outage in summer 2024 that I referenced in my original letter.

These aren't theoretical risks; they're regular occurrences that remind us what dependency actually means.

When critical infrastructure is owned and operated by foreign entities, you're subject to their operational issues, their strategic priorities, and their business decisions. You're in the queue with everyone else when things go wrong, hoping for updates and praying for quick restoration.

When I wrote to the government, my core argument wasn't just "we need more datacenters." It was that Norway needs to own and control AI infrastructure so Norwegian enterprises and researchers can access it easily and affordably—and reliably, under Norwegian operational control when it matters most.

AI infrastructure and the knowledge that comes with it represents the future equivalent of oil—and we're treating it like we're just another attractive location for foreign real estate investment.

Think about Denmark's housing market. They regulate foreign ownership heavily specifically to prevent external investors from buying up property just because Denmark is an attractive location. The Danes understood that letting foreigners own critical national assets—even something as simple as housing—creates dependency and removes local control.

We should be applying similar thinking to AI infrastructure.

What we're building now is mostly foreign-owned capacity that happens to be physically located in Norway. OpenAI owns their datacenter. Microsoft controls their infrastructure. CoreWeave operates their facility. These companies made strategic decisions to build here, and they can make strategic decisions to prioritize other markets, restrict access, or change terms whenever their business interests shift.

The smarter approach? Norway invests directly—potentially through the Oil Fund or dedicated infrastructure funds—and maintains ownership while allowing foreign companies to invest in Norwegian currency or take equity stakes in Norwegian operators. We supply the power, the land, the regulatory framework, and the capital. Foreign tech companies bring expertise and become customers or minority partners.

That's sovereignty. What we have now is just hosting.

What This Means Going Forward

I'm not advocating for closing borders or blocking foreign investment. The OpenAI, Microsoft, and CoreWeave projects are genuinely positive developments. Physical infrastructure in Norway is better than no infrastructure at all.

But watching the last year unfold confirmed my original concern: without direct government capital and ownership structures that ensure Norwegian control, we're building dependency rather than sovereignty.

The critical infrastructure designation helps—government can mandate security measures and impose restrictions during emergencies. The registration requirements create visibility. These are good steps.

But they're defensive measures. They don't change the fundamental reality that when strategic decisions get made about AI infrastructure allocation, capacity priorities, or service terms, those decisions happen in boardrooms in San Francisco and Seattle, not Oslo.

What happens when OpenAI needs to prioritize capacity for a major commercial customer over Norwegian research institutions? What happens when Microsoft's global strategy shifts and Norwegian expansion becomes lower priority? What happens if geopolitical tensions make US tech companies cautious about operations in Europe?

We'll find out, because we've structured everything around hoping foreign companies' interests align with Norwegian national interests.

Maybe they will. But that's hope, not strategy.

The Question That Keeps Me Up

Here's what I keep thinking about: Norway has the financial capability to fund strategic infrastructure when we decide it's important. We have a $2 trillion Oil Fund, yes, but more fundamentally, we're a wealthy nation with strong public finances and the demonstrated ability to make long-term infrastructure investments.

The Oil Fund is one possible mechanism, but there are others: dedicated infrastructure funds, strategic investment vehicles, public-private partnerships where government maintains ownership stakes.

The specific mechanism matters less than the recognition that this is infrastructure worth direct public investment.

When we built our telecommunications networks, energy infrastructure, and transportation systems, we didn't just hope private companies would find it profitable enough to build what Norway needed. We recognized these as strategic national capabilities and invested accordingly—not just in physical infrastructure, but in competency, expertise, and industrial capacity that benefits Norway for generations.

That's what makes the AI infrastructure question so frustrating. We're treating it as a real estate opportunity instead of recognizing it as the digital equivalent of what oil represented for the physical economy.

We understood with oil that natural resources represent strategic national wealth requiring long-term Norwegian control. We didn't just let foreign oil companies extract resources and leave. We built Equinor, maintained state ownership, developed Norwegian expertise, and captured value for Norwegians.

Now we face a similar moment with AI infrastructure—arguably the critical resource for the digital economy—and we're taking the opposite approach. We're saying: "Foreign companies can build and own it, we'll just provide the land and power."

This isn't just about money. It's about building Norwegian expertise in AI infrastructure operations, developing local industry capability, creating employment with deep technical knowledge, and ensuring that when AI becomes as fundamental to the economy as electricity, we control our own destiny.

The government response to my letter mentioned investigating details and working toward 2030 goals. But investigation and working toward aren't the same as committing capital and building ownership structures that ensure long-term Norwegian benefit.

I wrote that letter because I saw Norway falling behind. A year later, we've made real progress on policy and attracted significant investment. That's genuinely good news.

But the core question I raised—whether Norway will own and control AI infrastructure or just host it—remains unanswered.

And the longer we wait to address it, the harder it becomes to change course when foreign ownership is already established.

We still have time to get this right. But it requires recognizing that attracting investment isn't the same as securing sovereignty, and that critical national infrastructure shouldn't be left purely to market forces—especially when every other Nordic country understood that and invested accordingly.

The 2026-2027 period will show us whether Norway's bet on pure private sector approach actually delivers. I hope it does. But hope isn't a strategy, and sovereignty isn't something you can retrofit after foreigners already own the infrastructure.

We had a chance to lead. Instead, we're watching from the sidelines as foreign companies build what should have been Norwegian-owned and operated from the start.

That's what bothers me most about the last year. Not that nothing happened—but that what happened reinforces dependency rather than building sovereignty.

Christian Medewitz

Cloud Solutions Architect

CCC-Consulting AS

Bernt Ankers Gate 17, 1534 Moss

Want to discuss your organization's AI infrastructure strategy or sovereign cloud architecture?

Book a consultation or reach me at cm@ccc-consult.com

Sources & Further Reading

• Norwegian Government: Data Centre Strategy 2025 (PDF)

• National Digitalisation Strategy: The Digital Norway of the Future 2024-2030

• OpenAI Stargate Norway: [Official announcement, July 2025]

• Microsoft-Nscale partnership: [Press release, 2025]

• Energy Act consultation: [Ministry of Energy, June 2025]