The data center narrative around AI's carbon footprint was always a partial picture. Fast Company this week goes deeper, reporting that the more insidious climate cost of AI is in what it enables: new fossil-fuel-powered industrial processes, optimized extraction, accelerated demand generation. The direct emissions from running models are real. The enabled emissions are potentially far larger and almost entirely absent from corporate ESG disclosures.
Enabled Emissions and the Accounting Gap
The distinction between direct and enabled emissions maps onto a known problem in carbon accounting. Scope 3 emissions, those that happen upstream and downstream of a company's direct operations, are notoriously underreported and often excluded from net-zero pledges entirely. AI sits in a peculiar position: it can both produce emissions through compute and enable emissions through the decisions it optimizes for. A logistics AI that routes trucks more efficiently can reduce fuel use. A demand-generation AI that increases consumption increases the total carbon load of the goods being consumed. The arXiv paper on Evaluative AI by Yin, Miller, Potyka, Rago, and Toni argues for a framework where AI supports human decision-making through structured evaluation rather than direct recommendation. Applied to climate, this distinction matters enormously: an AI that presents the carbon consequences of a decision is categorically different from one that simply optimizes the decision.
CoreWeave, AI Infrastructure, and the Investor Reckoning
CoreWeave's earnings this week are a bellwether for the AI infrastructure investment thesis. The neocloud provider's forecast will tell investors whether AI compute demand is durable or whether the first wave of enthusiasm is already plateauing. But almost none of the investor analysis is incorporating climate liability into the valuation models. As regulatory pressure on enabled emissions grows, especially under European frameworks that are already pushing Anthropic toward C2PA compliance, the infrastructure layer will eventually face a carbon accounting reckoning. TurboFund's live investor signals track where AI infrastructure capital is flowing, but the climate liability variable is conspicuously absent from most deal memos. The gap between what AI enables and what it discloses is the next scandal in waiting.