Major technology companies’ artificial intelligence ambitions are driving an unprecedented surge in electricity demand, forcing utilities across the United States to build new natural gas power plants to keep pace. From the Midwest to the Southeast, Microsoft, Meta, and Amazon’s AI data center projects are straining local power grids to such an extent that utilities are proposing billions of dollars in new fossil fuel infrastructure.
Microsoft’s $3.3 billion AI data center hub outside Milwaukee has prompted WEC Energy Group’s We Energies to propose a $2 billion plan for new natural gas generation, which the utility describes as “critical” to powering Microsoft’s AI operations. The project could require between 1 and 1.1 gigawatts of power, though current estimates may already be outdated.
Meta’s massive Louisiana data center campus, announced in December 2024 with a $10 billion investment, will have a capacity exceeding two gigawatts once complete — enough to power a major city. This has spurred Entergy’s Louisiana subsidiary to build new natural gas plants for the first time in 50 years, with a $3.2 billion infrastructure plan including 2 GW of gas generation.
Meanwhile, Amazon’s Mississippi data center plans have led Entergy to begin construction on a $1.2 billion, 754-megawatt natural gas plant — capacity that could theoretically power 385,000 homes.
Research by energy consultant Cathy Kunkel at the Institute for Energy Economics and Financial Analysis found that utilities in Virginia, Georgia, North Carolina, and South Carolina plan to build a combined 20 GW of new natural gas generation by 2040, primarily driven by data center demand. In 2024 alone, requests from natural gas plants to connect to the grid more than doubled from the previous year, according to BTU Analytics.
This rush to natural gas directly contradicts Big Tech’s public commitments to renewable energy and conflicts with climate scientists’ recommendations for reducing emissions. While companies like Microsoft, Meta, and Amazon have announced investments in solar, wind, geothermal, and nuclear power projects, these initiatives face significant challenges. Many renewable projects are smaller in scale, depend on unproven technologies, or face steep regulatory hurdles — like Microsoft’s deal to restart Three Mile Island’s nuclear reactor.
The fundamental problem is timing: AI data centers need massive amounts of reliable, 24/7 power immediately, while renewable energy infrastructure takes years to develop. “Natural gas is relatively the quickest way to get substantial expansion of grid capacity so a data center can get up and running fast,” explained Philip Krein, research professor at the University of Illinois. Experts warn that utilities may be overbuilding infrastructure, leaving all customers — not just Big Tech — paying the bill for decades to come.
Key Quotes
The only concrete plans I’m seeing are natural gas plants
Cathy Kunkel, energy consultant at the Institute for Energy Economics and Financial Analysis, highlighted the disconnect between Big Tech’s renewable energy promises and the reality of what utilities are actually building to meet AI data center demand.
There’s a big time scale mismatch between the rates at which data center folks want to develop and the rates at which the electric power grid can develop
Philip Krein, research professor of electrical and computer engineering at the University of Illinois, explained why natural gas is becoming the default solution despite environmental concerns — AI companies need power faster than renewable infrastructure can be built.
Everybody wants to move really fast, and they’re not willing to wait. I hope they don’t get ahead of themselves
Professor Krein warned about the risks of rushing to build natural gas infrastructure without fully considering the long-term environmental and financial implications for all utility customers, not just the tech companies driving demand.
It is really concerning for the ratepayers of Mississippi, especially residential ratepayers, that a utility has effectively carte blanche to build whatever they see fit
Logan Burke, director of the Alliance for Affordable Energy, criticized Mississippi’s decision to waive approval processes for power infrastructure connected to Amazon’s data centers, leaving ordinary customers vulnerable to paying for potentially unnecessary infrastructure.
Our Take
This story reveals a fundamental tension at the heart of the AI boom: the technology’s transformative potential is colliding with physical infrastructure limitations and climate realities. The speed at which Microsoft, Meta, and Amazon are deploying AI capabilities has outpaced the grid’s ability to provide clean power at scale. What’s particularly concerning is the structural incentive problem — utilities profit from building infrastructure regardless of whether it’s ultimately needed, while ratepayers assume the risk.
The renewable energy commitments from Big Tech increasingly appear more aspirational than operational. Power purchase agreements for future solar and wind farms don’t help when you need gigawatts of reliable power today. This gap between promise and practice could become a significant reputational and regulatory challenge as public scrutiny of AI’s environmental impact intensifies. The industry may need to fundamentally reconsider either the pace of AI deployment or accept that current AI architectures are incompatible with near-term climate goals.
Why This Matters
This development represents a critical inflection point for both the AI industry and climate policy. As artificial intelligence becomes central to Big Tech’s business strategies, the technology’s enormous energy requirements are forcing difficult tradeoffs between innovation speed and environmental commitments. The scale is staggering — single AI data center campuses now consume as much electricity as entire cities.
The implications extend far beyond Big Tech’s balance sheets. Investor-owned utilities profit from building new infrastructure, creating incentives to potentially overestimate demand and overbuild capacity. If AI growth doesn’t materialize as projected, or if data centers don’t consume all the power utilities are planning for, ordinary ratepayers across multiple states will bear the financial burden through higher electricity bills for decades.
This also exposes the gap between corporate sustainability rhetoric and operational reality. While Microsoft, Meta, and Amazon tout renewable energy investments and carbon-neutral goals, their immediate needs are being met with fossil fuel infrastructure that will operate for 30-40 years. This could lock in emissions and undermine broader climate goals, raising questions about whether AI’s benefits justify its environmental costs and whether current regulatory frameworks adequately protect consumers from utility overinvestment.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- How AI Can Help Make Buildings More Energy Efficient
- Wall Street Asks Big Tech: Will AI Ever Make Money?
- The AI Hype Cycle: Reality Check and Future Expectations
- Meta Plans Massive Underwater Cable to Connect India and Southeast Asia
Source: https://www.businessinsider.com/utilities-ai-natural-gas-power-microsoft-meta-amazon-2025-2