Is it ethical to build energy-intensive AI systems during a climate crisis?

A Genuine Ethical Tension, Not a Simple Answer

Whether it’s ethical to build increasingly energy-intensive AI systems while the world faces a climate crisis is a question that doesn’t have a single settled answer among ethicists, technologists, or environmental advocates. It sits at the intersection of two things that are each individually well established but hard to weigh against one another: first, that training and operating today’s largest AI models requires substantial and growing amounts of energy and computing infrastructure; and second, that there are genuine, if not fully proven at scale, arguments that AI could contribute positively to climate-related goals in certain applications.

Reasonable people who take climate change seriously can and do land in different places on this question, and that disagreement reflects real underlying uncertainty rather than one side simply being uninformed.

The Case for Concern

Critics of continued expansion in energy-intensive AI point to a straightforward argument: the world is already struggling to meet climate targets, and adding a rapidly growing source of energy demand — much of which is still met, at least in part, through non-renewable sources depending on the region and utility — works directly against those goals. This concern is heightened by the pace of AI infrastructure buildout in recent years, including large new data center projects, which some environmental advocates argue represents a significant and under-scrutinized addition to global energy demand at a moment when reducing consumption, or transitioning it entirely to renewable sources, is critical.

There’s also a fairness dimension to this critique: many of AI’s benefits accrue disproportionately to companies and users in wealthier regions, while some of the environmental costs, particularly for facilities located near strained energy grids or water-scarce regions, are borne more locally and don’t always align with where the benefits are captured.

The Case for a More Complicated Picture

Others argue the picture is more complicated than a straightforward energy cost. Proponents point to specific applications where AI has shown promise in climate-relevant work, such as optimizing energy grid efficiency, accelerating materials science research relevant to renewable energy technology, or improving climate modeling. Some also argue that comparing AI’s energy footprint in isolation, without accounting for potential efficiency gains AI might enable elsewhere in the economy, understates the fuller picture. This argument is contested, however, since these potential benefits are not uniformly proven or realized at the scale needed to offset AI’s own growing energy demand, and critics note that speculative future benefits shouldn’t be used to justify certain present-day costs.

Bottom Line

There is no expert consensus on whether building energy-intensive AI systems is ethical during a climate crisis — it involves a genuine and actively debated tension between AI’s real and growing energy costs and its contested, not-yet-proven-at-scale potential climate benefits, and how any individual weighs that tradeoff depends significantly on values and judgments about uncertain future outcomes.

Go deeper

• How Does AI Help Manufacturers Reduce Energy Consumption?
• Could AI Itself Help Design More Energy-Efficient Computing Systems?
• Can AI help insurers price climate related risk more accurately?