Nvidia's rapid rise as the dominant chipmaker powering artificial intelligence has come with a staggering environmental cost: supply chain emissions have surged 725% since 2020, according to a new report by eight climate and corporate accountability groups. The findings challenge the tech industry's claims that AI will ultimately reduce global emissions, raising urgent questions about whether the AI boom can coexist with climate goals.

The report, released by organizations including Greenpeace International, Friends of the Earth, and Stand.earth, estimates that operating Nvidia chips sold since 2022 could have generated between four and 21 million tonnes of carbon dioxide equivalent in 2025 alone, depending on the electricity grid powering them. To put that in perspective, the report compares these emissions to those from burning coal sold by Russian state-owned company Russian Coal.

Nvidia's officially reported supply chain emissions, measured under what the industry calls Scope 3 (indirect emissions beyond direct operations), tell the story of explosive growth. In 2020, these emissions totaled about 1.3 million tonnes of carbon dioxide equivalent. By 2026, they had climbed to 10.7 million tonnes, according to figures compiled from Nvidia's sustainability reports.

Why Are Nvidia's Emissions Growing So Fast?

The surge isn't primarily from running the chips in data centers, though that matters. Instead, the biggest driver is what happens before the chips even ship: extracting raw materials, manufacturing components, and assembling the processors themselves. Producing advanced semiconductors requires enormous amounts of energy and water, and Nvidia's manufacturing largely takes place through suppliers in East Asia.

The timing is crucial. Nvidia's chips have become central to the generative AI boom, which exploded after late 2022. As demand for AI computing power has skyrocketed, so has the need for more chips, more manufacturing capacity, and more supply chain activity. This acceleration explains the dramatic 725% increase in just six years.

What Does Nvidia Say About This?

Nvidia has publicly argued that AI will ultimately reduce emissions across the economy. The company points to organizations including the International Energy Agency, World Economic Forum, and Boston Consulting Group as evidence that AI could cut emissions in other sectors through greater energy efficiency. However, the report's authors argue that these efficiency gains may not offset the overall impact when demand for computing power is rising rapidly.

"The very rapidly growing consensus is that AI is most likely to lead to net emissions reductions, especially if it's deployed broadly," said Josh Parker, Nvidia's head of sustainability.

Josh Parker, Head of Sustainability at Nvidia

But critics call this narrative "greenwashing." Ketan Joshi, the lead author of the report and an Oslo-based climate and energy analyst, stated that Nvidia's claims about AI's climate benefits are being used to mask growing pollution.

"The biggest AI profiteer also has one of the world's most obscured and opaque climate footprints. NVIDIA's claims that AI has a 'net climate benefit' are a greenwashing tactic used by the industry to mask ballooning pollution," said Ketan Joshi.

Ketan Joshi, Lead Author and Climate and Energy Analyst

What Specific Issues Does the Report Raise?

The report identifies several problems with Nvidia's current approach to climate accountability and the broader AI industry narrative:

• Lack of Transparency: Nvidia does not currently disclose emissions from the use of its sold products, making it impossible for customers or regulators to fully understand the climate impact of their purchases.
• Demand Response Limitations: Nvidia has promoted "demand response," where data centers temporarily reduce electricity use during peak demand periods. However, the report cites Princeton University research showing this flexibility may cover less than 1% of the year in scenarios with high data center growth, doing little to offset year-round increases in electricity demand.
• Efficiency Paradox: While AI may improve efficiency in specific applications, the overall effect is unclear when total computing demand is rising rapidly. More efficient chips may simply enable more AI deployment, not less total energy use.

How to Understand Nvidia's Climate Footprint

To grasp the scale of the problem, consider these key factors:

• Supply Chain Dominance: The vast majority of Nvidia's emissions come from upstream activities like mining silicon, manufacturing components, and assembling processors, not from running the chips in data centers.
• Scope 3 Reporting Gap: Nvidia reports supply chain emissions under Scope 3, which is less regulated and less transparent than direct operational emissions. The company does not yet report emissions from the actual use of its chips once they're installed in data centers.
• Cumulative Impact: The report's estimates are based on cumulative chip sales, assumed utilization rates, and different electricity grid emissions factors, meaning the true impact depends heavily on how and where the chips are powered.

The report calls for Nvidia to increase transparency by disclosing emissions from chip use, providing stronger evidence for its climate benefit claims, accounting for the wider emissions impact of data center growth, and setting clearer targets to cut supply chain emissions through additional renewable energy investment.

What Does This Mean for the Broader AI Climate Debate?

Nvidia's emissions surge arrives at a critical moment. World leaders are gathering this week in New York for the United Nations General Assembly and Climate Week, where the environmental footprint of AI and data centers is a major topic of discussion. The fear is that the data center boom will drive a surge in fossil fuel consumption, even though data centers can theoretically be powered by renewable energy.

Some experts remain optimistic. MIT Vice President for Energy and Climate Evelyn Wang predicted that data centers will eventually become water neutral within five years and stop adding to warming within a decade. However, others warn that the current trajectory is unsustainable without major changes to how AI infrastructure is built and powered.

The Nvidia report underscores a fundamental tension in the AI era: the technology may offer climate solutions, but the infrastructure required to power it is currently adding significantly to the problem. Whether the industry can resolve this paradox before the climate window closes remains one of the defining questions of the next decade.