This is hypocrisy and a governance failure. Most organizations still treat sustainability as a reporting function and AI as a strategic imperative. When priorities collide, AI wins—quietly, automatically, and repeatedly—because the incentives are aligned that way. Business units get rewarded for growth and speed, not for the long-term externalities of energy use, water consumption, and grid strain.
Even worse, the definitions are slippery. “Renewable-powered” can mean offsets. “Carbon-neutral” can mean accounting boundaries that exclude parts of the supply chain. “Efficient” can mean per-transaction improvements while total transactions explode. Meanwhile, the physical reality remains: More AI usage generally means more data center demand. More data center demand typically means more energy use, regardless of how compelling the sustainability narrative sounds.
AI value and carbon realities
First, enterprises should treat carbon as a primary architectural constraint, not just a retrospective report. They need to set explicit emissions or energy budgets at the product and platform levels, similar to budgets for latency, availability, and cost. If a new AI feature demands five times the compute, the decision shouldn’t be simply to ship and celebrate. Instead, organizations should consider whether they are willing to fund and publicly accept the operational and environmental costs. The old adage, “Don’t do anything you don’t want to read about in the news,” applies here as well, because, rest assured, the word will eventually get out about how much that feature costs in terms of sustainability.
