By: James Vance – SeaPRwire – The biggest risk in the AI race is no longer model performance. It is the electricity bill hiding behind it. Many executives spent years worrying about cloud costs. Now they are discovering that power availability and energy efficiency may become even tougher constraints. According to a survey of 300 senior executives from companies generating at least $1 billion in annual revenue, every respondent expects energy measurement and management to become a core business KPI within the next two years. That is a remarkable shift. Energy is moving from the facilities department into the boardroom.
The numbers explain why. AI workloads are consuming power at a pace few organizations anticipated. The survey found that 68% of executives have already experienced energy cost increases of at least 10% during the past year because of AI and data-intensive operations. Nearly all respondents expect costs to continue rising over the next 12 to 18 months, while only 22% believe their organizations are highly prepared. Meanwhile, U.S. data centers consumed about 4% of national electricity in 2024, a figure projected to reach 12% by 2028. A modern 100-megawatt data center can consume as much electricity as roughly 80,000 American households. Some newly planned facilities are targeting gigawatt-scale capacity. Against this backdrop, traditional metrics such as Power Usage Effectiveness, or PUE, no longer provide enough visibility. Enterprises increasingly need workload-level insight into where energy is consumed, why it is consumed, and how infrastructure decisions influence long-term operating costs.
This is where energy intelligence begins to resemble the rise of FinOps a decade ago. Cloud spending once appeared manageable until organizations realized they lacked visibility and accountability. Energy is following the same path. Infrastructure choices now determine future efficiency. Storage architecture offers a clear example. Flash-based storage systems consume less power, last significantly longer than traditional hard disk drives, and can store substantially more data within the same physical footprint. According to examples cited in the report, Virgin Media O2 reduced storage energy consumption by 98% after migrating to all-flash infrastructure. British Telecom achieved reductions exceeding 90%, while THG Ingenuity lowered data center power consumption by 80% without disrupting operations. These results highlight a broader lesson. The largest efficiency gains often occur before optimization begins, at the stage when technology decisions are made.
The organizations that treat energy intelligence as a strategic discipline will gain more than lower utility bills. They will free capital for AI expansion, reduce operational risk, and create greater flexibility when energy markets tighten. The survey already shows that 74% of leaders are optimizing existing infrastructure and 69% are partnering with energy-efficient cloud and storage providers. The next phase of AI competition may not be decided by who deploys the largest models. It may be decided by who understands the cost of every watt behind them.
Author bio: James Vance, a senior technology columnist covering enterprise AI, cloud infrastructure, data center economics, and the long-term business impact of emerging technologies.