Microsoft has been working towards matching all of its hourly electricity use with renewable energy purchases within the next five years.

That goal, set in 2020, stood out at the time because it went beyond annual offsets and focused on real-time energy use. Now, people familiar with internal discussions say the company is weighing whether to delay or drop it.

The discussion comes as Microsoft spends heavily on infrastructure to support artificial intelligence. Like Amazon and Alphabet, it is building large data centres to run services such as Copilot and its Azure cloud platform.

These facilities require vast amounts of electricity, and demand is rising faster than earlier projections suggested.

Some of the newer sites under development are expected to draw several gigawatts of power. To put that in context, one gigawatt can supply roughly 750,000 homes in the United States. Meeting that level of demand with renewable sources alone is proving difficult, especially within tight timelines.

Energy supply is now a practical concern. Renewable projects take years to plan and build. By contrast, natural gas plants and nuclear facilities can be brought online more quickly, and companies are turning to them to avoid delays.

Microsoft has already taken steps in that direction. In 2024, it reached an agreement with Constellation Energy to help restart a unit at the Three Mile Island nuclear plant in Pennsylvania. The deal shows how the company is balancing its climate targets with immediate energy needs.

The pressure is not limited to Microsoft. Across the sector, emissions have continued to rise despite public commitments to cut them. That gap is drawing attention from regulators and investors, who are watching closely as companies expand their AI operations.

For now, Microsoft has not commented publicly on the reported review. What is clear is that the scale of AI is changing earlier assumptions. The company set one of the most ambitious clean energy goals in the industry. Whether it can still meet it on time is now uncertain.

This brings Meta’s total clean energy capacity from Invenergy to 1,800 MW, nearly doubling its earlier figure of 1,000 MW. It’s a direct response to the high energy demands driven by the company’s AI-focused infrastructure.

The new deals cover four major renewable projects scheduled to come online between 2027 and 2028: Yellow Wood Solar (300 MW) and Pleasant Prairie Solar (140 MW) in Ohio, Decoy Solar (155 MW) in Arkansas, and Seaway Wind (196 MW) in Texas. 

The electricity from these sites will feed into their respective regional power grids, but Meta will retain the renewable energy credits, effectively allowing the company to account for the clean power in its carbon reporting.

While financial terms remain undisclosed, the scale and scope of the deal involve Meta committing long-term to a diversified clean energy mix to meet the immense demands of AI processing.

Ted Romaine, executive vice president of Origination at Invenergy, said the partnership goes beyond another supply deal. “Winning the AI race requires reliable, cleaner, affordable energy and energy infrastructure — today and in the future. We’re grateful for our continued relationship with Meta and look forward to future partnerships as we work to strengthen American energy independence and economic prosperity.”

Meta recently inked a 20-year agreement with Constellation Energy for 1.1 gigawatts of nuclear energy from the Clinton Clean Energy Centre in Illinois. It also joined forces with geothermal developer XGS Energy to build a 150 MW water-independent geothermal facility in New Mexico.

Urvi Parekh, Meta’s Head of Global Energy, stressed the urgency behind these moves. “We’re laser-focused on advancing our AI ambitions—and to do that, we need clean, reliable energy. We’re grateful for Invenergy’s longtime partnership that helps us support our energy needs and implement our clean energy goals, and look forward to continued collaboration.”

At the heart of this strategy is a transition toward long-term, stable, and zero-carbon energy sources that can reliably support Meta’s future AI workload. 

The company’s model now relies on both immediate and forward-looking procurement, blending traditional renewables with more experimental and baseload options like nuclear and geothermal.

These infrastructure investments are expected to yield economic benefits. Beyond environmental impact, the projects are projected to create jobs and strengthen local economies in the host states, while also supporting national efforts towards energy independence.

Invenergy, headquartered in Chicago, is currently the largest privately owned clean energy developer in the United States. Its role in powering Meta’s massive data infrastructure reveals how private-sector energy partnerships are becoming indispensable to the tech industry’s future.