Tor Langøy, founder of Viking Digital, does not describe his company in terms of data centres or infrastructure. He frames it as a shift in economic logic.
“We will not sell the electron,” he says. “We will sell the AI.”
That distinction sits at the centre of the conversation with Langøy and his colleagues Tom Yeh and Venkat Thummisi. Norway, long established as a net exporter of energy, is now being positioned as something else entirely.
“You take energy, you build AI factories, and you export AI,” Yeh explains.
The implication is direct. Rather than exporting raw power at relatively low value, the country can convert that power into high-value computational output.
Langøy returns repeatedly to efficiency as the determining factor. “In order to win, you need the cheapest electrons, the cheapest cost of capital, and the most efficient computer,” he says. His argument is that first-generation cloud and data centre operators are structurally constrained. They outsource critical components, carry high cost bases and operate with limited control over power.
Viking Digital, by contrast, is being built as a fully integrated model. “We have our own electrons. We have our own tech stack. We design, procure, build, operate and optimise,” he says.
Thummisi, drawing on his experience in AI infrastructure, describes the current market as overly simplistic. “What you see today is distribution of GPUs,” he says. “Customers are asked how many they want, not what they need.” His contention is that this model is inefficient and misaligned with real-world use cases. “We are selling outcomes, not GPUs,” he continues. “For any given use case, we can deliver what the customer needs and still be about 30 per cent cheaper.” The advantage, he argues, comes from system-level optimisation rather than hardware scale alone.
“Two things will always be constrained, GPUs and power. The question is how efficiently you use both.”
Yeh places those efficiencies in a broader economic context. “If you export energy, you might make around $100,000 per megawatt per year,” he says. “With AI data centres, that can move into the millions per megawatt.” The comparison is not theoretical. It underpins the company’s claim that Norway can materially alter its economic output by moving up the value chain. “This is transformational,” Yeh says. “You are not just improving margins, you are changing what the country exports.”
The scale of the ambition is reflected in the funding strategy. Langøy confirms that the company is beginning with an initial raise of $50 million, but that this is only the first step. “This is just the beginning,” he says, pointing to a roadmap that moves into multi-billion-dollar investment as projects scale and acquisitions are executed. Part of that strategy involves acquiring existing data centres and upgrading them. “We take older facilities, bring them under our umbrella, and optimise them for AI,” he explains. This allows Viking Digital to accelerate deployment while improving the efficiency and revenue profile of underutilised assets.
Location is central to the thesis. Thummisi points to Norway’s natural advantages. “Power is everything, and the geography here is highly resourceful,” he says. The combination of renewable energy, climate conditions and political stability creates a foundation for large-scale AI capacity. That capacity, in turn, becomes exportable. “We should not just distribute power positively, we should distribute AI capability,” he adds.
Yeh emphasises that the model is not solely about efficiency or profit. “Every data centre should create a positive impact,” he says. He points to job creation, local economic development and the reuse of excess heat as integral to the design. “We are looking at how to use that heat to warm communities,” he explains. The intention is to move beyond the traditional perception of data centres as extractive infrastructure. “Tor is not just building for profit, he is building for the community as well,” Yeh adds.
There is also a strategic dimension tied to data sovereignty. Langøy notes that European companies are increasingly cautious about where their data resides. “Many are no longer comfortable sharing data externally,” he says, referencing growing concern around jurisdiction and control. Viking Digital’s model incorporates sovereign AI principles, ensuring that data and compute remain within trusted environments.
The conversation returns, in its closing moments, to the central idea. Yeh summarises it plainly. “Norway is already a net exporter of energy. Now it can become a net exporter of AI.” Langøy’s version is more succinct. “At some point, we stop selling electricity and start selling intelligence,” he says.
It is a shift from raw resource to refined output, from commodity to capability. If Viking Digital succeeds, the impact will not be limited to a company or even a sector. It will alter how a country defines what it produces, and what it sells to the world.
