
AI is creating enormous economic value, but who will ultimately retain it? Looking at compute, power, capital requirements and payback reveals why the biggest AI beneficiaries may not necessarily capture the most value for shareholders.

The first AI winners supplied the infrastructure AI needed. History suggests the next winners may emerge elsewhere, as cheaper intelligence reshapes costs, productivity and competitive advantage.

AI is becoming cheaper and more efficient, yet demand for physical infrastructure keeps growing. As bottlenecks shift from chips to power, grids and equipment, a bigger question emerges: can infrastructure and productivity catch up before capital loses patience?

AI investment is moving beyond chips and models into the physical world. Data centers, power grids, transformers, electricity and execution capacity could become the defining bottlenecks of the next phase of the AI boom.