Tech giants such as Amazon, Microsoft, Meta, and Google are at the forefront of the artificial intelligence revolution. Their AI innovations power transformative technologies across industries, from advanced language models to sophisticated machine learning applications. However, these advancements come at an often-overlooked environmental cost.
The rapid expansion of AI requires massive computing power, driving the construction and operation of vast data centers. These facilities consume enormous amounts of electricity, significantly increasing the carbon footprint of companies leading the AI race. To counterbalance this, many of these firms rely on unbundled renewable energy certificates (RECs). While these credits create an appearance of sustainability, they often fail to represent actual emissions reductions, raising critical concerns about transparency in corporate environmental reporting.
An Amazon Web Services data center in Ashburn, Virginia.Photographer: Nathan Howard/Bloomberg
The Rise of AI and Emissions
Artificial intelligence has become a cornerstone of technological progress, but it also brings a steep energy cost. AI systems demand immense computational resources, from training large models to supporting real-time operations. This surge in demand has led to a sharp increase in emissions from data centers.
Microsoft, for example, has reported that its emissions are now 30% higher than in 2020, despite its ambitious goal to achieve carbon negativity. Similarly, Amazon and Meta have also seen emissions rise, attributing the increase to construction materials like steel and cement for new data centers rather than the energy-intensive nature of AI operations. While technically accurate, this narrative overlooks the growing strain AI places on energy resources.
Adding to the complexity, tech companies often market their AI services—such as Amazon’s AWS, Microsoft’s AI Copilot, and Meta’s Llama—as having minimal environmental impact. This messaging reassures consumers and businesses while obscuring the broader environmental consequences of adopting these technologies. Such narratives risk perpetuating misconceptions about the true cost of AI advancements.
Source: Company reports, Bloomberg
Note: RECs data for 2022
Unbundled RECs and Misleading Claims
Unbundled renewable energy certificates (RECs) are a mechanism allowing companies to offset emissions without directly using green energy. By purchasing these credits, companies can claim emissions reductions on paper, even if their electricity comes from fossil fuel sources. This practice has become widespread among tech firms, but its validity is increasingly questioned.
Amazon, for instance, relied on unbundled RECs for 52% of its renewable energy claims in 2022, while Microsoft used them for 51% and Meta for 18%. Critics argue that this approach misrepresents the environmental impact of these companies, creating a false narrative of sustainability. Studies suggest that unbundled RECs rarely lead to new renewable energy projects, undermining their effectiveness as a tool for meaningful emissions reductions.
Source: CDP, Bloomberg Analysis
Note: Data covers electricity consumption in 2022
The environmental impact becomes even clearer when emissions are recalculated without unbundled RECs. In such a scenario, Amazon’s emissions for 2022 would increase by 8.5 million metric tons—three times its reported figure. Similarly, Microsoft’s and Meta’s emissions would rise by 3.3 million and 740,000 metric tons, respectively. These discrepancies highlight the urgent need for more accurate and transparent carbon accounting methods.
The Need for Updated Carbon Accounting Standards
The Greenhouse Gas Protocol, established in 2001, serves as the foundation for corporate emissions reporting. While it has undergone minor updates, its allowance for unbundled RECs has come under increasing scrutiny. Experts argue that these rules fail to reflect actual greenhouse gas reductions, leading to inflated sustainability claims.
A growing body of evidence suggests that unbundled RECs do not incentivize the development of new renewable energy projects. Instead, they serve as a cost-effective way for companies to improve their environmental metrics without making substantial operational changes. Google recognized this issue years ago and phased out its use of unbundled RECs. Instead, the company focuses on direct renewable energy sourcing through long-term power-purchase agreements (PPAs), which ensure that operations are genuinely powered by clean energy.
These agreements not only offer a transparent and effective solution but also encourage the development of new renewable energy infrastructure. As renewable energy becomes more accessible and cost-effective, the reliance on unbundled RECs should diminish, paving the way for more accountable practices across the industry.
A Call for Transparency and Action
The growing demand for AI is driving unprecedented energy consumption and emissions. With data centers expanding rapidly to meet AI’s computational needs, the environmental toll will continue to rise unless the industry adopts more sustainable practices. Transparent reporting and direct renewable energy sourcing are critical steps toward addressing this challenge.
Tech companies must transition away from unbundled RECs and embrace methods that genuinely reduce emissions. By focusing on long-term renewable energy contracts and adhering to updated carbon accounting standards, they can align their practices with real sustainability goals. Furthermore, upcoming revisions to the Greenhouse Gas Protocol provide a unique opportunity to redefine how corporate emissions are measured and reported.
Ultimately, the tech industry has a responsibility to lead by example in addressing climate change. Transparent, accountable practices are not only necessary for environmental stewardship but also essential for maintaining trust among consumers and investors in an era where sustainability is a growing priority.