AI factories as flexible loads for power systems BOX 6 Business case AI demand is shaped through design and governance. Companies embed efficiency into systems, optimize data and model size, and track resource use, while policies and education promote digital sobriety, aligning AI growth with trust and sustainability goals. Smarter demand management also enhances economic competitiveness and energy security, ensuring AI expansion supports system reliability and sustainable growth.Challenge AI data centres create rising electricity demand and operate as inflexible loads. Without adaptive management, this growth risks grid strain and slower decarbonization. The goal is to make AI infrastructure smarter in how and when it consumes power. Solution –Platform enables data centres to optimize power use based on grid conditions –Predictive models schedule workloads and balance energy demand in real time –Secure application programming interfaces (APIs) and governance frameworks ensure compliant, responsible operations Impact Near-term impacts realized or anticipated (less than one year) –40% peak demand reduction –1,080 kilowatt-hours (kWh) savings –0.68 tons of carbon dioxide emissions avoided per megawatt-hour (tCO2/MWh) Further impacts realized or anticipated (more than one year) –Improved system uptime –Additional revenues from flexibility services –Expanded peak demand reduction and carbon emissions avoided Reviewing the levers in action Usage-based pricing models: Aligns compute intensity with energy availability, driving efficiency incentives Consumer dashboards: Real-time visibility into workload timing and energy use Model selection guidance: Predictive models optimize scheduling and power use Digital sobriety campaigns: Internal efficiency focus, not public-facing* Regulatory nudges: Follows National Institute of Standards and Technology (NIST)/ International Organization for Standardization (ISO) standards, not policy-driven* *See Table 4 for relevant “shape demand wisely” use case examples. Source: AI Energy Impact public use case database, Infosys; Emerald AI. From Paradox to Progress: A Net-Positive AI Energy Framework 21

From Paradox to Progress A Net Positive AI Energy Framework 2025