While AI has been quickly gaining daily active users, it only represented about a quarter of all data center workloads in 2025, with training driving most of the demand. However, a significant shift is anticipated in 2027, when inference workloads could overtake training as the dominant AI requirement.

While an AI model represents a one-time or periodic investment, once the model is created, inference generates ongoing revenue through actual application usage. Looking forward, every AI model deployment creates sustained inference demand that grows with user adoption. This growth, however, depends on the emergence and rapid adoption of inference applications that don't yet exist at scale.

Inference demand requires geographical distribution to reduce latency and serve users effectively. This will drive regional deployments and embedded systems at the edge.

Data center operators are expected to increase behind-the-meter power arrangements and explore colocated battery storage as the average wait time for a grid connection in primary data center markets exceeds four years.

Natural gas is projected to play a major role in alleviating grid constraints in the U.S., both for temporary bridge power and increasingly for permanent on-site power generation. This can be evidenced by surging global turbine orders. However, it is worth noting that some of the largest data center tenants are averse to natural gas solutions as they are not viewed as sustainable.

Natural gas as a solution is less prominent in EMEA and APAC. In these regions, renewables such as solar and wind are seeing increased utilization. In EMEA for instance, projects combining renewables and private wire transmission can reduce the cost of power for tenants by 40% compared to the grid.

Due to utility interconnection delays, some data center operators are moving beyond PPAs to directly fund their own energy generation. Additionally, a number of markets have implemented ‘bring your own power’ mandates (Ireland, Texas, et al.), which is fueling this trend.