Climate

Digital infrastructure has a physical footprint — electricity, water, land, minerals, and the emissions embodied in manufacturing hardware that is replaced on a short cycle. It is nonetheless treated as immaterial in most of the decisions that authorise it, in a way that would be unthinkable for a comparable investment in physical plant. A new facility is approved on cost and capability; its demand on a specific grid, at specific hours, in a specific water catchment, is frequently not part of the assessment at all.

Two problems compound. The first is measurement: the dominant reported figures are efficiency ratios, which have genuinely improved and which say almost nothing about total impact, because efficiency gains in computing have historically funded expansion rather than reduction. An operator can report an audited year-on-year improvement while its absolute consumption rises. Both statements are true; usually only one is published. The second is that the systems used to measure, report and verify emissions across the wider economy are themselves digital infrastructure — so the credibility of climate policy rests on the integrity of exactly the kind of system we study.

We work on honest accounting for the material cost of computation: absolute consumption reported alongside every efficiency figure, disclosed reporting boundaries rather than assumed ones, water reported as consumption weighted by local scarcity, and location-based emissions published alongside market-based ones rather than the more flattering figure being selected. And we work on the integrity of measurement and verification systems, because a climate commitment is only as sound as the infrastructure that reports against it.