Is AI generating demand for carbon credits?

In a short few years, AI has become the major throughput in conversations about pretty much everything.

The speed of adoption has made it difficult to understand how negative AI’s contributions are to climate change. From the explosion in demand for chips enabling the boom and emissions associated with running computers to train AI models, to water consumption and biodiversity impacts of large data centers, it’s a complicated and constantly-evolving landscape. 

What’s easier to quantify is AI companies’ involvement in accounting for and offsetting their emissions. While carbon market activity is just a part of their overall sustainability approach, it provides a useful benchmark for understanding which companies are actively dealing with their emissions, and to what degree.

Methodology

Using (what else) AI to identify and categorize over 130 companies actively involved the fledgling tech sector, we scoured our data of over 30,000 buyers of carbon credits to identify VCM activity. Where relevant, we looked for multiple potential names to match to a company (e.g., xAI and SpaceX). Due to data limitations, we did not examine whether a company’s offsetting activity was specifically related to AI – so companies that have multiple business lines will be counted, even if their offsets were not for AI specifically. Companies may also offset but not report it publicly, meaning their data will not be picked up in the below analysis. We looked at all time offsetting activity, with the earliest credits being used back in 2008 by Google. The vast majority of retirements, however, have taken place since 2020, when the VCM experienced high growth.

The companies were categorized into four groups: 

• AI / Cloud End Users – hyperscalers, AI labs, and GPU compute

• Developers & Operators – data center companies and their investors

• Construction & Engineering – data center builders

• Infrastructure & Suppliers – microchip and hardware providers, power and cooling firms

This captured hyperscalers like Google, Meta, and Microsoft, AI labs like OpenAI and Anthropic, microchip developers like Nvidia, ASML, and TSMC, and construction companies like Turner and DPR. In total, 135 companies were identified globally.

Results

Of the companies we examined, 26% had any offsetting activity. This is a high ratio for a sector. Given the amount of negative attention that companies are facing around the environmental impact of AI, however, there is a clear opportunity for AI firms to leverage the VCM to support climate-positive initiatives. 

Microsoft leads the way, by far, in VCM activity. Of the 123m tons retired or purchased via offtake, Microsoft accounts for over 100m – over 80% of overall activity. For anyone watching carbon markets, that shouldn’t come as a surprise: the company has purchased millions of carbon credits over the last three years. 

With Microsoft, AI’s share of VCM activity was as high as 20%, in 2025; ex-Microsoft, the sector peaked at 3% of VCM activity in 2024 – and has been declining since then. 

The value of the credits retired or purchased via offtake by AI firms sits at $9.2b all time, again with a significant part of that coming from Microsoft. Between 2023-25, AI firms contributed an average of $3b to the VCM, with the majority of this coming from offtake deals signed by Microsoft. 

The full list of companies and their offsetting activity is at the bottom of this article.

Market Impact

The analysis shows there is a large opportunity for the AI sector to make better use of the VCM. How may that look in practice?

Given the recent news that Microsoft is slowing down its carbon procurement, the most obvious way for AI companies to get involved is to take over the baton and establish a leadership role in credit procurement. This isn’t something that will happen overnight: companies need to fully understand their environmental impact, teams need to be built to source and procure credits, and policy considerations must be studied. Filling the vacuum left by Microsoft’s pullback is a surefire way to bring VCM participants on board. And while the cost of engaging with the market in such a close way is high, it’s dwarfed by the investment, revenues, and valuations of the firms in the AI sector. 

Of course, very few companies will want to commit to the market in the same way that Microsoft has. Encouragingly, it’s easier than ever for companies to engage with the VCM in a way that ensures a high degree of integrity in a cost-effective way. Ratings agencies, market initiatives like ICVCM, and marketplaces that conduct due diligence prior to selling credits give buyers better insights into project quality; a push to promote portfolio approaches to carbon procurement means it’s easier to create cost-effective strategies without compromising on quality. The integrity of the credits will be especially important to AI companies whose public reputation is tarnished – the bad press from buying rotten carbon credits is a news story that no journalist would want to pass up. 

Another way for AI firms to engage with the VCM is to form a buyers’ club. Popularized by Frontier, which includes several of the companies in our analysis already, a club allows buyers to share information and due diligence projects jointly, lowering the cost of investment for all. An AI-led buyers’ club can be a way for companies to share resources in an attempt to counter the narratives around the sector’s environmental impact. For example, AI labs can join forces to identify projects located near their data centers and source credits directly from them. By focusing on specific geographies, the AI firms may address concerns of local communities who have been vocal in opposing data center buildout.

To date, AI companies (with the exception of Microsoft) have not meaningfully impacted demand for credits in the carbon market. The companies’ environmental impact is becoming an increasingly bigger problem for AI firms. With thousands of projects, varied price points, and improved integrity frameworks, the VCM can be a significant first step for AI in cleaning up its carbon problem.

Full list of companies with offsetting activity (all time):