Amazon Is Funding the Biggest Gas Power Plant in the US — While Pledging to Fight Climate Change

Amazon has a climate pledge. They signed it. They promote it. They put it on their website. And they're using it to market themselves as a responsible tech giant.

They're also backing what could become the single largest source of climate pollution in the United States.

The Story

Amazon Web Services is supporting a massive natural gas power plant in Mississippi — one that could become the top source of US climate pollution. According to Ars Technica, the plant is being built to power Amazon's data centers, which are increasingly dedicated to AI workloads.

Let that sink in. The company that created "The Climate Pledge" — a commitment to reach net-zero carbon by 2040 — is directly funding fossil fuel infrastructure that would make the climate crisis worse. Not indirectly, through a supply chain they can't control. Directly, by backing a specific power plant project.

The AI Connection

This isn't just about Amazon being hypocritical (though it is). It's about the AI industry's dirty secret: AI is incredibly energy-intensive, and that energy has to come from somewhere.

Training a single large language model can consume as much electricity as 100 US homes use in a year. Running inference at scale — which is what Amazon's data centers do — requires constant, massive power draw. As AI adoption accelerates, the energy demands are growing exponentially.

The tech industry has been selling a narrative that AI will help solve climate change — optimize power grids, improve renewable energy forecasting, accelerate materials science for better batteries. And some of that is true. But the immediate reality is that AI deployment is driving massive new demand for electricity, and that electricity is coming from fossil fuels.

The Scale of the Problem

The Mississippi gas plant isn't a small facility. If built as planned, it would be one of the largest natural gas power plants in the United States. Natural gas is cleaner than coal, but it's still a fossil fuel — it still emits CO2 when burned, and the extraction and transport of gas leaks methane, which is 80 times more potent as a greenhouse gas than CO2 over a 20-year period.

The math is simple: if Amazon's AI data centers are powered by fossil fuels, then every AI query, every model inference, every API call contributes to climate change. The carbon footprint of AI isn't abstract — it's being measured in power plant emissions.

The Greenwashing Problem

Amazon's Climate Pledge commits to reaching net-zero carbon by 2040. The key word is "net-zero" — not "zero." Net-zero means Amazon can continue emitting carbon as long as they offset it through carbon credits, reforestation, or other mitigation strategies.

The problem with net-zero accounting is that it's highly susceptible to greenwashing. Carbon offsets are notoriously unreliable — many reforestation projects fail, carbon credits are often double-counted, and the verification systems are weak. A company can technically meet a "net-zero" pledge while continuing to increase its actual emissions, as long as it buys enough offsets on paper.

When Amazon backs a massive gas plant while simultaneously pledging net-zero, they're betting that future offsets will somehow cancel out current emissions. Given the scale of the plant they're supporting, that bet looks reckless.

The Broader Pattern

Amazon isn't alone in this. The entire tech industry is grappling with the tension between AI ambitions and climate commitments:

• Microsoft: Pledged to be carbon-negative by 2030, but AI data center growth is causing emissions to rise, not fall.
• Google: Committed to 24/7 carbon-free energy by 2030, but their emissions increased 13% in 2024, driven largely by AI infrastructure.
• Meta: Net-zero by 2030, but building massive AI training clusters that draw enormous power.

Every major tech company has a climate pledge. Every major tech company is also building more data centers, consuming more electricity, and — in many cases — increasing their carbon emissions. The pledges and the actions are moving in opposite directions.

What This Means for AI Developers

If you build AI systems, you should care about this. Not because you personally are responsible for Amazon's power plant choices, but because the carbon footprint of AI is becoming a real issue that affects the industry:

Regulatory risk: The EU's AI Act includes provisions for environmental impact assessment. Other jurisdictions are likely to follow. AI companies may soon need to report and justify the energy consumption of their systems.

Customer pressure: Enterprise customers are increasingly asking about the carbon footprint of their AI providers. If you're building an AI product, your customers may eventually require carbon disclosures.

Reputational risk: The "AI is green" narrative is starting to crack. As stories like Amazon's gas plant get more attention, the public is becoming aware that AI has a real environmental cost.

Practical action: If you're running AI models, consider local inference. I run models on a Raspberry Pi that draws 5-12 watts. That's roughly $1/month in electricity and a trivial carbon footprint. Compare that to running the same workload on a cloud API powered by a gas plant in Mississippi. The local option isn't always feasible, but when it is, it's dramatically cleaner.

The Path Forward

The tech industry needs to reconcile its climate pledges with its AI ambitions. This means:

1. Transparency: Companies should report the actual energy consumption and carbon emissions of their AI infrastructure, not just net-zero projections.
2. Renewable energy procurement: If AI requires massive power, that power should come from renewables — not new fossil fuel plants. Amazon has the resources to build solar and wind at scale. They're choosing not to.
3. Efficiency standards: AI models should be evaluated not just on performance but on energy efficiency. A model that's 5% more accurate but uses 10x more energy isn't necessarily the better choice.
4. Policy action: Governments should require tech companies to account for the full lifecycle carbon emissions of their AI infrastructure, including the power plants that feed it.

Amazon's decision to back a massive gas plant while claiming climate leadership is a test case. If the tech industry can get away with this level of greenwashing, the climate pledges are meaningless. If customers, regulators, and developers push back, there's still a chance to align AI growth with climate responsibility.

I'm not optimistic. But I'm not giving up either. The first step is being honest about what's happening: AI is driving new fossil fuel infrastructure, and the companies doing it are lying about their climate commitments. That's the reality. What we do about it is the question.