In the boardrooms of America’s energy giants, a quiet revolution is developing to reshape how we generate electricity and who controls the digital economy. Talen Energy’s shares surged 20 per cent on Thursday after announcing $3.5 billion in acquisitions of gas-fired power plants, seemingly to meet “growing demand from data centres”. Yet this seemingly straightforward transaction reveals a significant transformation that energy companies are abandoning their traditional role as utilities and turning themselves into the infrastructure backbone of artificial intelligence.

Data centres consumed 4.4 per cent of total US electricity in 2023, a figure the Department of Energy projects will reach between 6.7 and 12 per cent by 2028, but these projections, dramatic as they appear, understate the true scope of the transformation. The White House warned this week that electricity prices could spike by 9 to 58 per cent by 2030 without massive new energy investment, requiring $1.4 trillion in additional infrastructure.

What we are witnessing is not merely increased demand for electricity, but the emergence of a new economic hierarchy where access to reliable power determines technological power. Consider recent developments, such as Google announced $25 billion in data centre investments across the PJM grid region, Blackstone committed $25 billion to Pennsylvania data centres and natural gas plants, and Wall Street firms are acquiring utility companies to “benefit from the rising demand for electricity from data centres”. This is not the gradual movement due to market forces, but a deliberate restructuring of the energy sector around AI infrastructure. The traditional model of utilities serving diverse customers is giving way to a new model where energy companies become dedicated servants of the digital economy. Talen’s acquisition of the Moxie Freedom Energy Center and Guernsey Power Station represents more than capacity expansion; it signals the start of a new category of energy provider optimised for the specific demands of artificial intelligence.

This has far-reaching geopolitical consequences, for example, US firms like Baker Hughes, Hunt Energy, and Argent LNG announced this week they will develop a “masterplan for Syria’s oil, gas and power sector” following the lifting of sanctions. The timing is hardly coincidental. As AI infrastructure becomes the foundation of economic and military power, controlling energy resources in strategic regions takes on new urgency. Devastated by years of conflict, Syria's energy sector presents an opportunity to build AI-optimised infrastructure from the ground up. Meanwhile, clean energy projects across the United States face an epidemic of cancellations disguised as “uncertainty.” For example, Fortescue suspended US green energy developments, while more than $15 billion in green energy projects have been abandoned in the first five months of 2025 alone. The official narrative blames policy uncertainty and regulatory changes, but the reality is more simple capital is flowing toward immediate AI infrastructure needs rather than long-term renewable projects.

This represents a fundamental misallocation of resources driven by short-term AI profits rather than long-term energy security. The irony is that as we race to build the infrastructure for artificial intelligence, we are abandoning the very technologies that could power it sustainably. Instead, we are doubling down on gas-fired plants that will lock in carbon emissions for decades while serving the computational demands of systems designed to solve humanity’s most significant challenges. The market has begun to price in this transformation, though incompletely. Constellation Energy and Vistra have rallied alongside Talen on AI power demand, while traditional utility stocks remain anchored to outdated assumptions about diversified customer bases and regulated returns. The companies that recognise this shift as soon as possible will capture disproportionate value as the energy sector bifurcates into AI infrastructure providers and legacy utilities serving increasingly marginalised traditional demand.

This shift will accelerate through 2025 and beyond, driven by the inevitable development of AI and the massive capital flows it steers. Energy companies that fail to pivot toward AI infrastructure will find themselves relegated to serving residential and small commercial customers with declining margins and limited growth prospects. Those that successfully transform themselves into AI infrastructure providers will capture extraordinary returns as digital demand continues its exponential growth.

The winners in this transformation will be the energy companies that move fastest to secure long-term contracts with major technology firms, acquire the most efficient generation assets, and position themselves in regions with favourable regulatory environments for data centre development. The losers will be traditional utilities clinging to outdated business models, renewable energy developers competing for increasingly scarce capital, and ultimately, consumers who will bear the cost of this infrastructure transformation through higher electricity prices.

Power has indeed become the new oil, but unlike the petroleum economy of the 20th century, this new energy model serves not human mobility and industrial production, but the computational powers of artificial intelligence. Whether this represents progress or a misallocation of society’s resources remains to be seen. What is certain is that the transformation is already underway, and its consequences will reshape not only the energy sector, but the fundamental relationship between technology and power.