AI is being sold as a tool for tackling climate change, while its data centres and fossil-fuel applications may be adding to the problem. An Inside Climate News report argues that the gap between those promises and AI’s current climate costs deserves closer scrutiny.
Watch Desk analysis
What happened
The report centres on a paper published in a Nature journal in August. It says the paper found that emissions from powering AI data centres, together with AI’s use in optimising fossil-fuel production, outweighed savings from AI applications linked to clean energy. The report also notes that the International Energy Agency has said AI’s potential climate benefits remain on the horizon, while rising electricity demand is a more immediate concern.
The article sets those findings against upbeat claims made during Climate Week, including Nvidia’s account of AI being used in clean-energy projects. It also quotes independent data-centre researcher Jonathan Koomey warning that forecasts of the build-out are uncertain, and that customers could be left paying for power infrastructure built for projects that never materialise.
Why it matters
The climate case for AI cannot be settled by pointing to a handful of useful pilots, or by treating every forecast of future harm as certainty. The practical question is whether the emissions and energy costs of today’s AI uses are outweighed by real, measurable savings elsewhere. The report’s account of the paper says they are not, at least for the uses it examined.
There is a bill attached to speculative growth, too. If utilities build new power plants for data centres that fail to arrive, households and businesses may end up carrying the cost. That makes the pace and funding of data-centre construction a local infrastructure question, not just an argument between tech optimists and climate campaigners.
Our read
The useful test is evidence, not mood music. AI may help improve energy systems, but a promising application is not the same as a net climate benefit across the technology’s actual uses. The report gives readers a concrete reason to ask companies to account for both sides of that ledger, and to ask regulators who pays if the forecast demand evaporates.
What to watch
- Whether the paper’s findings are tested or challenged by further research.
- How quickly AI-related electricity demand grows, and which power sources meet it.
- Whether companies publish fuller accounts of AI’s climate costs alongside selected success stories.
- How regulators allocate the costs of power infrastructure built for data-centre demand. Source: Inside Climate News report, republished by Daily Kos
Discussion spark: Should data-centre developers pay the cost if the electricity infrastructure built for their projected demand is left underused, or should that risk be shared with other customers?
Sources and evidence
- AI’s meet-cute with climate denial – Daily Kos (10 October 2026, 18:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.