A review of emissions reports from Google, Microsoft, Meta, and Apple by The Guardian revealed that actual emissions could be up to 662% higher than the companies’ official statements, The Guardian announced on Sep. 15.
The discrepancy is largely attributed to the difference between direct, “location-based” emissions at company facilities and the purchase of renewable energy credits, according to The Guardian. Meanwhile, the growing use of generative AI is further increasing the energy demands of big tech companies.
The Guardian reviewed emissions reports from 2020 to 2022.
Renewable energy certificates provide credit for indirect power purchases
In the renewable energy certification system, organizations purchase renewable-generated electricity to offset their spending on their energy consumption elsewhere.
Critics of this system say factoring credits, or “market-based” emissions, into emissions calculations obscures “location-based emissions,” the pollution created directly by company-owned infrastructure. Companies subtract the amount of sustainably produced electricity they buy from their emissions reports — even if that electricity is never used in the company’s facilities, The Guardian alleges.
The Guardian combined location-based emissions with reported market-based emissions, concluding that actual emissions could be 662% higher, or a difference of 7.62 times, compared with official reports.
Amid the discussion is a behind-the-scenes lobbying battle over the Greenhouse Gas Protocol, developed by an oversight body that allows market-based emissions to be factored into official calculations. Since these standards form the basis for how companies report their emissions, the inclusion or exclusion of market-based calculations can be controversial.
Meta also performs its calculations in accordance with the GRI Standards, an independent metric. Google and Microsoft are ahead of the curve in separating out credit-based metrics from their climate reporting, as seen in Microsoft’s 2024 sustainability report. Their respective 24/7 (Google) and 100/100/0 (Microsoft) goals remove carbon energy purchases from the equation.
Amazon, which also claims carbon neutrality, was too vast and complex for The Guardian to accurately assess how its reported emissions might differ from the actual emissions it creates. Amazon’s data centers don’t make up the bulk of its Scope 2 (in-house purchased electricity) emissions. Instead, e-commerce and warehouses impact its Scope 2 emissions highly.
When reached for comment, a Meta spokesperson pointed to the tech giant’s record of building electrical grids in the same locations as their data centers. The spokesperson also highlighted the company’s Sustainability Report, which lists both location-based emissions and market-based emissions, as well as its strategy of using long-term purchase agreements to support the development of sustainability electricity projects, along with its promotion of green tariffs.
SEE: Tech giants are aware of AI’s climate harms – but aren’t slowing down.
Recommendations for CISOs and CTOs
Emissions reports are a reminder that organizations should take into consideration the financial and environmental costs of resource-depleting technologies.
CISOs and CTOs should stay informed about the standards used to calculate emissions and their own company guidelines for making tech decisions that consider both energy use and environmental sustainability. An environmental policy can reassure customers that using your product or service doesn’t worsen human-caused climate change.
Use of generative AI, in particular, can increase emissions. A September 2024 report from Forrester advised companies to consider using smaller, more efficient AI models; employing AI only when it is truly needed; and leveraging AI to correlate sustainability practices and financial performance.
TechRepublic reached out to Google, Microsoft, Meta, and Apple for comment.