Getty Images/iStockphoto
Navigating energy management strategies in AI data centers
Data centers face increasing energy demands driven by AI applications. This shift prompts a focus on energy management strategies, sustainability and innovative approaches.
Data center energy use conditions have changed, but the fundamentals of measuring energy consumption and costs have not. As data center energy consumption continues to surge, driven primarily by the rapid rise of AI applications, industry stakeholders are grappling with the escalating costs and the urgent need for energy management.
In January 2026, the U.S. Congress reported that U.S. data centers, excluding cryptocurrency, consumed about 176 terawatt-hours (TWh) in 2023, representing 4.4% of the country's total electricity use. Projections suggest that this consumption could double or triple by 2028, reaching up to 12%.
This increase is largely driven by the surge in AI applications requiring significant computing power. The Federal Reserve estimated that corporate adoption of AI rose by 68% between January and September 2025.
Data centers, especially hyperscalers, face rising electricity costs due to increasing demand. They also encounter challenges such as delays in grid connections, infrastructure costs and fluctuating wholesale prices. The high power density of AI workloads adds to cooling and operational expenses, prompting operators to explore on-site energy generation and load management solutions to mitigate rising costs and reliability risks. This article examines energy management strategies that data center managers should consider.
Developing new data center energy management strategies
The what-if modeling for different data center energy scenarios helps prepare data center managers for cost management and short-term events. However, there are also longer-term trends in energy costs and consumption that should be part of data center energy and cost management planning.
Transition to sustainable and renewable energy
Recent U.S. policy has reverted to an emphasis on fossil fuels as primary energy sources. However, renewable energy and sustainability will be the path to the long-term future of data centers.
Many data centers have committed to running on 100% renewable energy, with one significant caveat: that this 100% commitment doesn't necessarily mean a 100% conversion to only renewable energy sources in the data center itself. To arrive at the 100% renewable energy figure, data centers are factoring in the renewable energy they generate and sell back to a central power grid for renewable energy credits (RECs). This can be counted in the data center's renewable energy calculus and factored in as a profit that offsets data center energy costs.
Invest in on-site and near-site renewable energy options
On-site or near-site solar and wind energy is being used by more data centers, including larger sites that can invest in their own self-sufficient, renewable energy microgrids.
The microgrid option is being considered as a way for data centers to control their own energy destiny and uptime in the event of a major infrastructure disruption, such as a weather event or cyberattack, that disrupts service from the central energy grid. Large AI data centers being built by Amazon, Microsoft, Meta and others are exploring microgrids that generate their own energy. This approach can help alleviate community concerns about increased energy costs and demand.
Enter into power purchase agreements (PPAs)
Most data center admins manage energy costs by measuring IT equipment draw, PUE, DCiE, Water Usage Effectiveness and Computer Power Efficiency.
"After they do this, data center managers must incorporate scenarios for fuel prices, demand surges and infrastructure tariffs," said Joe Nyangon, VP of Solutions at Energy Exemplar, a B2B SaaS platform for energy modeling and simulation. "To do this, they often blend vendor data, EIA (environmental impact assessment) projections and site-specific contracts to create multi-year estimates amid volatility."
Thanks to the formalization of energy management metrics, even smaller data centers can assess their current and projected energy consumption. This enables them to negotiate PPAs with energy providers that guarantee that a site will use a certain amount of energy per year in exchange for discounted pricing.
Data center managers and facility managers must collaborate
The energy management metrics that data centers use, in combination with the energy consumption ratings that energy-conscious IT equipment vendors now provide, have enabled data center managers to see how much energy a server, storage bay or even a chip uses. However, not as much time has been spent on the facilities side of the data center.
For instance, is the data center using the most energy-efficient cooling? Are data center HVAC units positioned near the greatest data center "hot spots" to ensure that cooling is applied where it is most needed? Is the roofing over the data center such that outside heat from the sun is reflected away from the data center so that less cooling is needed? All are areas where data center managers lack the expertise that facilities managers have. The two must work together.
Final thoughts on energy management
Energy consumption is a major driver of data center cost management and investment. To plan for it, data center managers are largely on their own, as there are currently no legally binding energy standards that apply explicitly to the operation of data centers in the private sector.
"What data center managers can do is invest in advanced cooling technologies like liquid or immersion systems, high-efficiency power supplies (like 80 PLUS Titanium), and adopt hot/cold aisle containment, server virtualization, and workload orchestration around renewables," said Nyangon. They can also conduct regular energy audits, work hand in hand with their facilities managers and keep the C-level apprised.
Mary E. Shacklett is president of Transworld Data, a technology analytics, market research and consulting firm.
