How AI and the Trump administration are fuelling ‘quiet quitting’ on IT sustainability

That energy comes from datacentres, which are huge facilities filled with servers and computing equipment, that power these GenAI tools. While datacentres have existed for decades, the growth of AI has accelerated their construction and energy use to the point where scientists warn they could consume as much electricity as Japan by 2026.

AI is driving extraordinary innovation and investment, but it is also making corporate net-zero pledges harder, if not impossible, to achieve. Instead of abandoning their promises outright, companies are quietly shifting the goalposts, presenting their environmental strategies in softer language while pushing ahead with energy-intensive AI projects.

Google is a case in point. The company has pledged to achieve net-zero emissions by 2030 while simultaneously scaling up AI. Its 10th annual environmental report, published in June 2025, claimed a 12% reduction in datacentre energy emissions but acknowledged that overall GHG emissions had risen by 6.2%.

The non-profit advocacy group Kairos Fellowship, however, accused Google of reframing its commitments to obscure this reality. Its report, Google’s eco-failures, noted that between 2019 and 2024, net-zero targets were featured prominently in Google’s sustainability reports, always within the opening pages. 

By 2025, those references had been relegated to the appendix and rebranded as a “carbon reduction moonshot”. According to co-author Nicole Surgeman, this is how corporations move the goalposts without attracting attention: “I think they’re attempting to save face by subtly walking it back over a span of years, [and] they’re hoping that people will not notice that they used to have this goal by making it less prominent.”

Kairos also points out that Google’s emissions have risen by 121% since 2019. The organisation claims Google attempts to obfuscate this by buying carbon emission offsets, which is when companies compensate for their carbon footprint by funding projects that reduce or remove greenhouse gases elsewhere. 

Surgeman argues that this only obscures the truth: “What it doesn’t account for is that the real emissions that the’’re emitting are going up over the same period.” 

She and her colleagues linked this surge directly to AI, tracing it back to 2021 and the launch of LaMDA, the foundation for Google’s Gemini model, which drove rapid datacentre expansion. “The emissions are mostly from their datacentres and they’ve been building those at a very rapid pace to account for the power they need for their AI models,” she adds.

Google rejected the criticism, with a spokesperson insisting that its emissions are calculated under the widely used Greenhouse Gas Protocol and independently assured. 

But Franz Ressel, co-author and researcher on the report, says: “Google’s PR team can whip up evasive talking points…but it cannot hide from the truth: Google’s own third-party-assured data reveals that the corporation has utterly failed to decrease its emissions. Unfortunately, as we detail in our report, Google hides this data in the appendix of its environmental reports, cherry-picking PR-favourable data points to highlight instead.”

Computer Weekly reached out to Google for a response, but none was forthcoming at the time of publication.

The result, according to Kairos, is that Google will not be able to reach its net-zero goals by 2030 primarily due to its drive towards increasingly energy-hungry AI.

Yet despite this, it is claimed the company will not publicly admit this and is instead favouring an approach where the company quietly backs away from its sustainability aims and goals.