Data Center Power Demand to Quadruple by 2035 - What Leaders Must Do Now

The projection that data center electricity demand could quadruple by 2035 reframes compute growth from an IT operational issue to a strategic energy and infrastructure problem. Rapid adoption of large models, edge deployments, and always-on services drives both peak and base-load power needs. Left unaddressed, this trajectory will increase operating costs, expose companies to grid instability and brownouts, and widen carbon footprints that attract regulatory and investor scrutiny.

For businesses that rely on cloud and data-center providers, the immediate impact is twofold: escalating energy cost exposure and potential capacity constraints in high-growth regions. Organizations with heavy AI workloads will face higher unit compute costs as providers pass on power, cooling, and carbon compliance expenses. Geographic concentration of new data centers may also create regional bottlenecks and compel customers to negotiate capacity or pay premiums for preferred locations.

Leaders should prioritize a three-part response: reduce consumption intensity, diversify supply, and embed energy risk into procurement decisions. Tactics include optimizing models and inference pipelines, migrating latency-tolerant workloads to lower-carbon regions or off-peak windows, and pushing vendors for energy-efficient instance types and transparent carbon accounting. Long-term levers include signing renewable PPAs, investing in on-site generation and storage for owned facilities, and collaborating with utilities on demand-response programs.

Finally, incorporate energy forecasting into tech and real-estate planning: model expected power requirements of roadmap projects, stress-test supplier SLAs against grid scenarios, and include sustainability KPIs in vendor contracts. Treating compute growth as an energy strategy issue enables firms to control costs, reduce regulatory risk, and turn sustainability into a competitive differentiator rather than an afterthought.