For years, the conversation around technology’s environmental impact has been dominated by the hum of cooling fans and the massive electricity consumption of data centers. While it is true that training large-scale AI models requires a staggering amount of power, a recent study published in npj Climate Action suggests we have been looking at the wrong culprit. Former Microsoft sustainability experts Will and Holly Alpine have turned their attention toward what they call “enabled emissions”—the pollution generated not by the servers themselves, but by the ways AI empowers the fossil fuel industry to extract, refine, and distribute oil and gas more efficiently. By treating these two sectors as separate entities, the tech world is inadvertently overlooking a massive blind spot that could derail global climate goals.
The Alpines, who resigned from Microsoft earlier this year due to the company’s deepening ties with oil giants, argue that the partnership between Big Tech and fossil fuel corporations is a self-reinforcing cycle. When tech companies provide the AI tools that help energy firms locate deposits or optimize drilling operations, they aren’t just selling software; they are lowering the cost and risk of fossil fuel production. This, in turn, keeps oil prices lower and demand higher, cementing our global reliance on hydrocarbons. The researchers highlight that these industries are two sides of the same coin, yet tech giants continue to treat their collaboration as a purely operational or financial business decision, conveniently ignoring the environmental footprint that follows the implementation of their code.
The sheer scale of the emissions generated by this AI-fueled efficiency is sobering. Using complex economic modeling to forecast how productivity gains in the energy sector translate into broader environmental harm, the study estimates that AI’s role in optimizing fossil fuels could drive a global increase in energy-related emissions by 1.2 to 4.8 percent. To put that in perspective, the lower end of that range is equivalent to the entire annual greenhouse gas output of Mexico, while the high end reaches the staggering levels of Russia, one of the world’s largest emitters. The researchers found that this “enabling” effect far outweighs the emissions produced by the data centers themselves, and it even eclipses the potential climate benefits gained from using AI to speed up the development of renewable technologies like solar or wind.
A core frustration for the Alpines is the narrow way “sustainability” is defined within the corporate world. Current reporting standards focus almost exclusively on “operational emissions”—the electricity a company buys to run its own headquarters or data centers. However, this creates a false sense of progress. By focusing on greening their internal operations, tech companies are essentially cleaning their own house while simultaneously handing the fossil fuel industry a megaphone to increase their own destructive output. By failing to account for “enabled emissions,” the tech industry is essentially washing its hands of the damage caused by the very tools it sells to the oil and gas sector, ignoring the fact that these tools are becoming a primary engine for continued fossil fuel dependency.
The symbiotic nature of this relationship is becoming increasingly literal, as seen in the recent deal between Microsoft and Chevron. Rather than simply being customers, these industries are now building the infrastructure of the future together, with Chevron planning a large-scale gas plant in Texas specifically to power Microsoft’s data centers. Crucially, the arrangement also allows Chevron to leverage that same computing power for its own internal AI needs, creating a closed loop of fossil-fuel-powered digital development. This blurring of lines between the energy sector and the tech sector proves the Alpines’ point: the “digital transformation” of the oil industry is not a side project; it is a fundamental reconfiguration of how we produce and consume energy, often at the expense of the climate.
Ultimately, this research serves as a wake-up call that the AI revolution cannot be decoupled from the climate crisis. If tech companies want to be true leaders in sustainability, they must look beyond their own electricity bills and critically evaluate who they are empowering. The promise of AI to solve complex global problems is often touted as the industry’s greatest legacy, but if that same technology is simultaneously making it cheaper and easier to extract more oil and gas, its net impact may be profoundly negative. Moving forward, the tech industry will need to reckon with the moral weight of its “enabled emissions,” or risk becoming the primary architects of the very climate instability they claim to be working against.
