Microsoft Reevaluates 100/100/0 Clean Energy Target Amid AI Expansion

Microsoft’s clean energy target under pressure from AI data centres

Microsoft is reportedly reconsidering its “100/100/0” clean energy commitment due to the massive power requirements of AI data centres. The company’s total energy use has surged 168% since its 2020 baseline, significantly outpacing its 71% revenue growth.

Why This Matters

The technical reality of “hourly matching”—requiring zero-carbon power at the exact time and location of consumption—is colliding with the physical scale of AI infrastructure, which requires roughly one gigawatt of new capacity every three months. While annual renewable matching is achievable through credits, the 100/100/0 model requires grid-level synchronization that is increasingly difficult as total emissions rise.

This tension is forcing a pivot toward high-density power sources like nuclear restarts and on-site natural gas. With Microsoft expecting to spend $190 billion on data centres by the end of 2024, the financial and operational pressure to maintain 24/7 uptime for AI services like Azure and Copilot is currently outweighing the feasibility of strictly hourly green energy constraints.

Key Insights

• Microsoft’s 100/100/0 target requires 100% zero-carbon electricity matching on an hourly basis within the same grid region, 2021.
• Scope 1, 2, and 3 emissions increased 23.4% from the 2020 baseline due to AI and cloud expansion, Microsoft Sustainability Report 2025.
• US data centre power demand is projected to rise from 34.7 GW in 2024 to 106 GW by 2035, BloombergNEF.
• Behind-the-meter power projects, such as gas-fired generation co-located with data centres, reached 30GW in Texas announcements between 2024 and 2026, Reuters.
• Microsoft signed a power agreement with Constellation Energy to restart a unit at the Three Mile Island nuclear plant to support AI infrastructure, 2024.

Practical Applications

• Use case: Microsoft and We Energies developing 1.2GW of solar and battery projects in Wisconsin to provide carbon-free energy by 2028. Pitfall: Relying solely on intermittent renewables for 24/7 AI loads leads to grid instability or the need for fossil fuel backups.
• Use case: Implementation of behind-the-meter gas-fired generation co-located with large-scale data centres in Texas to bypass grid constraints. Pitfall: Using on-site fossil fuels contradicts net-zero targets and increases Scope 1 emissions.
• Use case: Utilizing nuclear power restarts (e.g., Three Mile Island) to provide carbon-free baseload energy for high-density AI clusters. Pitfall: High regulatory hurdles and long lead times for nuclear infrastructure may delay capacity expansion relative to AI demand.

References:

• https://www.cloudcomputing-news.net/news/microsoft-clean-energy-target-ai-data-centres/