For decades, the relationship between large electricity consumers and energy producers was relatively straightforward. Utilities and independent power producers generated electricity, while commercial and industrial customers purchased it. Even the world's largest technology companies largely fit that model, securing long-term power purchase agreements (PPAs) to meet ambitious growth and sustainability goals while leaving the business of developing and operating energy infrastructure to others.
However, as hyperscalers race to build the infrastructure needed to support artificial intelligence, they are no longer simply buying electricity and the boundary between energy producer and consumer is starting to disappear. Increasingly, hyperscalers are financing, shaping, and in some cases directly controlling the energy infrastructure that makes AI growth possible. The shift represents more than a new procurement strategy. It marks the emergence of a new form of vertical integration, one where access to energy is becoming as strategically important as access to compute itself.
The catalyst is straightforward. AI has fundamentally changed the economics of electricity demand. Hyperscale data centers require enormous amounts of reliable, always-on power, often on timelines that traditional utility planning and grid expansion cannot meet. As a result, hyperscalers are beginning to move upstream in the energy value chain.
Initially, that meant signing larger and more sophisticated power purchase agreements that helped finance new generation projects. Today, the relationship is evolving further. Technology companies and AI infrastructure developers are increasingly investing directly in generation platforms, energy developers, and colocated infrastructure rather than simply contracting for the electricity those assets produce.
Google's recent acquisition of Intersect Power for $4.75 billion illustrates this evolution. Rather than acting solely as a customer purchasing renewable electricity, Google is helping shape the development of a large pipeline of solar and battery storage assets that will support future AI infrastructure. The relationship extends beyond energy procurement into the development of the infrastructure itself.
Other AI infrastructure companies are taking similarly direct approaches. Nscale, a London-based AI infrastructure company, recently acquired American Intelligence & Power Corporation, bringing power development capabilities in-house as part of its strategy to accelerate AI data center deployment. Instead of relying exclusively on third-party utilities or developers, companies are increasingly viewing energy infrastructure as a strategic asset worthy of direct ownership or control.
Nscale is just one example of how this trend extends beyond hyperscalers themselves. Investors building AI infrastructure are acquiring power developers and generation platforms to secure access to future capacity, recognizing that energy infrastructure has become one of the primary constraints on AI growth. The definition of “AI infrastructure” is shifting to the means of energy production itself, creating what is arguably a new category of company, one that provides the means of production and specialized technical expertise required to produce the computational capacity that powers AI.
This trend represents a meaningful shift in how the energy market functions. Historically, large electricity customers influenced demand forecasts. Today, hyperscalers and AI infrastructure companies are increasingly influencing where generation gets built, how projects are financed, and which infrastructure investments move forward. Long before a data center comes online, an anchor customer can effectively de-risk an energy project by providing the confidence needed for developers and investors to commit capital.
In many cases, the traditional distinction between customer and producer becomes difficult to define. A hyperscaler may not operate a generating facility in the way a utility does, but it may finance its development, determine its location, shape its design requirements, and secure most or all of its output through long-term agreements. Increasingly, some are moving beyond even that model by acquiring pieces of the energy development ecosystem itself.
The implications extend well beyond individual companies. Utilities are adapting to unprecedented demand growth, while independent power producers are finding new opportunities to partner with AI developers. Utilities are experiencing renewed investor interest as AI-driven electricity demand reshapes long-term growth expectations. At the same time, technology companies are becoming more deeply embedded in the infrastructure decisions that have traditionally defined the energy sector.
The AI era is giving rise to a new form of vertical integration. Rather than owning raw materials or manufacturing plants, hyperscalers are seeking greater control over the infrastructure that determines whether their businesses can continue to scale. Energy is no longer simply an operating expense. It is becoming a strategic input, one that companies are increasingly willing to finance, develop, and, in some cases, own.
That shift may ultimately prove to be one of AI's most significant impacts on the energy sector. It is changing not only how electricity is consumed, but also who participates in building the infrastructure that delivers it.
Freddie Sarhan is CEO of Sapphire Technologies. He previously led business development at Calnetix and held senior roles at Praxair Surface Technologies, Capstone Turbine, and Rolls-Royce, and holds a B.S. in Mechanical Engineering from McGill University.
