At the recent AIM Summit in London, I had the pleasure of chairing a panel on the timely topic of “The Infrastructure of Power: Energy, Sovereignty and the AI Economy.”
I was joined by a distinguished group of leaders representing the technology, energy and investment sectors: Sam V. Tabar (WhiteFiber), Kerri Langlais (TeraWulf), Karl Rendenbach (AgenticScale.ai) and Ozan Özkural (Tanto Capital Partners). The blend of analytical insight and practical experience made for a thought-provoking discussion throughout.
Our conversation confirmed what is now impossible to ignore: AI has shifted decisively from a software story into a strategic infrastructure story.
Recent numbers have only sharpened the point. Hyperscaler capital expenditure on AI infrastructure in the United States is projected at between US$650 billion and US$725 billion in 2026 alone, arguably the largest peacetime industrial mobilisation in American history. Microsoft is reportedly carrying US$80 billion in unfulfilled Azure orders that cannot be delivered because of power constraints. That single data point tells the entire story.
The primary constraint on the AI economy is no longer capital, nor even silicon. It is power. Every major hyperscaler has now signed at least one nuclear power agreement, with thirteen projects committing more than 9.8 gigawatts of nuclear capacity. Yet even this is proving insufficient. Northern Virginia has effectively halted new data centre permits in several counties until power infrastructure catches up.
Two implications follow. First, power certainty is now the single most valuable commodity in commercial real estate. Second, the largest fortunes in technology are increasingly being created not by those building the applications, but by those controlling the underlying infrastructure. This layer is measured in kilowatt-hours.
Several regions are particularly well positioned to benefit, including the Gulf, the Nordics, India and Canada, each combining surplus energy, available land and sovereign ambition. Investors who recognise this shift early may well define the next investment cycle.
#AIInfrastructure #DataCenters #EnergyTransition #NuclearPower #AIMSummit
