OpenAI's Arctic gambit reveals Silicon Valley's new resource colonialism
How America's AI champion is rewriting the rules of technological sovereignty by exploiting Nordic energy abundance whilst European governments chase digital dreams
The announcement arrives with familiar Silicon Valley grandiosity: a $1 billion investment in Arctic Norway, promising 100,000 GPUs by 2026, all powered by pristine hydroelectricity. Sam Altman frames it as helping Europe "realise the full potential of AI," but the reality emerging in Kvandal, a remote town outside Narvik, reveals something more calculated—and more troubling.
This isn't partnership. It's sophisticated resource colonialism, dressed in sustainability rhetoric and democratic ideals.
The desperation behind the geography
OpenAI's Nordic expansion exposes an uncomfortable truth: American AI development has hit fundamental infrastructure limits faster than anyone anticipated. Gartner predicts that 40% of existing AI data centres will face power constraints by 2027, whilst European grid infrastructure requires €584 billion in modernisation. The geographical expansion reflects operational necessity rather than strategic vision.
Power demand from AI data centres is expected to surge 160% by 2030. One-third of Europe's grid infrastructure exceeds 40 years of age. Ireland imposed moratoriums on new Dublin data centres in 2022. Virginia, the world's largest data centre hub, faces grid strain despite massive construction programmes.
Norway generates over 90% of its electricity from hydropower, achieving carbon intensity of just 24g compared to 267g in the UK. For OpenAI, grappling with the energy-intensive demands of training frontier AI models, Norway isn't merely attractive—it's essential.
Observe the power dynamics carefully. OpenAI retains ownership of AI models, intellectual property, and high-value technological assets. Norway provides land, power, and infrastructure. This partnership resembles historical resource extraction more than technological collaboration.
Resource extraction in sustainable clothing
Sustainability rhetoric obscures fundamental asymmetry. Nscale, the London-based partner, positions itself as building "sovereign, sustainable" infrastructure, yet sovereignty remains distinctly American. OpenAI coordinates international deployments with the US government, maintaining control over where its technology can be deployed through export restrictions and "democratic AI" frameworks.
Josh Payne, Nscale's CEO and co-founder of a $357 million NASDAQ-listed battery metals SPAC, brings experience from renewable energy mergers and cryptocurrency mining—industries adept at extracting value from natural resources and regulatory arbitrage. Aker ASA, controlled by Norwegian billionaire Kjell Inge Røkke, represents Norway's industrial establishment embracing the role of power supplier for American technological dominance.
Economic logic proves stark. Norway contributes renewable energy infrastructure worth hundreds of millions whilst OpenAI retains models generating billions in revenue. This arrangement mirrors historical patterns where resource-rich regions provide inputs whilst value creation and capture occur elsewhere.
Consider Stargate's broader scope: $500 billion investment focused on American infrastructure, creating "hundreds of thousands of American jobs" whilst securing "American leadership in AI." International extensions through "OpenAI for Countries" explicitly reinforce US technological leadership whilst accessing foreign resources.
Democratic AI as imperial ideology
The "democratic AI" framing deserves particular scrutiny. OpenAI positions international partnerships as spreading democratic values against authoritarian alternatives, but the structure reveals different priorities. Partner countries invest in expanding the global Stargate Project, "thus in continued US-led AI leadership," according to OpenAI's own documentation.
This isn't technology transfer or capability building. European partners provide resources to strengthen American AI dominance whilst receiving customised ChatGPT services—the digital equivalent of exporting raw materials and importing finished goods.
The intellectual dishonesty runs deeper. OpenAI claims to help countries build "their own Stargates," but the infrastructure serves American AI models under American oversight. True technological sovereignty would involve European countries developing independent AI capabilities, not facilitating American resource extraction through renewable energy partnerships.
European governments' strategic blindness
European complicity proves most damaging. Norwegian authorities explicitly encourage data centre establishment, removing regulatory obstacles whilst marketing the country as providing "pure, renewable energy" to global tech giants. The government's Data Centre Strategy positions facilities as "sustainable, digital power centres" without questioning who controls the digital outputs.
Strategic myopia reaches breathtaking proportions. At a moment when technological sovereignty determines economic competitiveness and national security, European governments facilitate their own subordination. Rather than building indigenous AI capabilities or demanding technology transfer, they compete to provide the cheapest renewable energy for American AI development.
Patterns extend beyond Norway. Sweden, Finland, and Iceland similarly market themselves as sustainable destinations for global tech infrastructure. Nordic countries collectively become power suppliers for American digital dominance, replicating colonial resource relationships in the AI economy.
The infrastructure constraint revelation
Norwegian expansion reveals uncomfortable truths about AI development's physical requirements. Despite years of efficiency improvements, training frontier AI models requires vast computational resources consuming enormous amounts of electricity. Geographical dispersion isn't strategic positioning—it's acknowledgment that no single region can provide sufficient power infrastructure.
Seamless digital transformation proves illusory. AI development faces hard physical constraints requiring massive infrastructure investments and energy supplies. The technology's revolutionary potential depends on decidedly non-revolutionary resource extraction from countries willing to provide cheap power.
Implications extend beyond individual partnerships. AI development requires global resource extraction to overcome infrastructure constraints. The industry's concentration in a few American companies creates technological dependencies spanning continents. Countries providing resources become stakeholders in American technological success without corresponding control or ownership.
The emerging geography of dependency
OpenAI's Arctic gambit foreshadows a new geography of technological relationships. American companies access global resources whilst maintaining technological control, creating dependencies that extend far beyond commercial arrangements. Countries compete to provide inputs for American AI development rather than building independent capabilities.
This pattern proves particularly insidious because it appears collaborative. Norway willingly provides renewable energy infrastructure, viewing itself as promoting sustainable AI development. The reality involves Norwegian resources strengthening American technological dominance whilst Norwegian institutions remain dependent on American AI services.
Resource extraction has evolved sophisticatedly for the digital economy. Rather than seizing physical assets, American companies create technological dependencies whilst accessing foreign resources through partnerships framed as mutual benefit. The colonialism proves harder to recognise because it operates through apparently voluntary arrangements.
Kvandal's Arctic data centre may run on pristine hydroelectricity, cooling servers with naturally frigid air whilst generating minimal carbon emissions. Yet the arrangement it represents—European resources serving American technological imperialism—suggests that in the AI economy, sustainability rhetoric often masks much older patterns of extraction and control.
The question European policymakers should confront: is providing renewable energy for American AI development really partnership, or simply a greener form of technological colonisation?