The Scarcest Resource in the AI Revolution Is Electricity

The scale of this buildout is unprecedented. McKinsey projects AI-related infrastructure spending will exceed $5 trillion by 2030, while JLL estimates developers need 100 gigawatts of new data-center capacity. This demand has turned energy into a zero-sum game, forcing tech giants to scramble for dedicated generation assets. Microsoft is moving to restart a nuclear reactor at Three Mile Island, while Google, Amazon, and Meta are signing long-term power arrangements to bypass grid delays that can stretch for years.

For some, the strategy was to secure power before the AI surge began. Bitzero (NASDAQ: AIBZ) repurposed its early focus on Bitcoin mining to lock in low-cost electricity and land in Norway, Finland, and North Dakota. By controlling over a gigawatt of potential capacity, the company has positioned itself as a landlord for enterprise AI. This shift was validated on May 5, when the firm signed a 15-year, $2.6 billion lease with OneQode Networks for its Namsskogan campus. As Kevin O’Leary, a Bitzero backer, notes, half of the announced data centers may never materialize simply because the grid cannot support them. In this environment, the companies winning the AI race are not those with the best models, but those with the most stable power supply.