CarbonLayer measures what AI actually costs — power, water, carbon — not what the slide deck claims. Boundary-tagged metering for Scope 2/3 reporting that survives an audit.
AI companies report efficiency gains with no disclosed baseline, no measurement boundary, and no way to audit the claim. A number without a scope is marketing, not data.
CarbonLayer exists to close that gap. We meter what AI training and inference actually draw — power, water, carbon — with every reading tagged to its scope (1 vs. 2/3) and boundary (on-site vs. generation-side). Modeled projections stay separate from live metered numbers, always.
If a claim can't be checked against a stated baseline, it isn't a claim. It's a slide. We build metering that survives an audit — not one that survives a press cycle.
