Sub-system
Power usage effectiveness compares total facility energy to the energy reaching the equipment. It drove a decade of genuine improvement and is now close to its floor in well-built facilities — which means the useful questions have moved to what the equipment does with the power once it has it.
In one sentence
Power usage effectiveness is the ratio of a facility's total energy consumption to the energy consumed by its computing equipment; a value of 1.0 would mean no energy spent on cooling, distribution losses or anything else.
The metric was useful because it was simple and because facilities were genuinely wasteful — values near two were common, meaning as much energy went to cooling and losses as to computing. Containment, higher operating temperatures, better economisation and more efficient conversion brought well-run facilities close to the practical floor.
Its limitation is now the more interesting fact. A facility can improve the ratio by making its computing equipment less efficient, since that raises the denominator. It says nothing about whether the work being done is useful, and it does not count water or carbon at all — which is why operators are judged on a set of measures rather than one.
In a modern facility, cooling dominates the non-computing energy, followed by conversion losses in the power chain and then lighting and ancillary systems. That is why liquid cooling matters beyond density: removing heat with pumps and warm water rather than with compressors and fans removes most of the remaining overhead.
Evaporative cooling saves electricity by consuming water, which the energy ratio does not capture at all. Water usage effectiveness and carbon usage effectiveness were introduced to make that trade visible, and in water-stressed regions the water figure is now the one that determines whether a site can be permitted.
The measure operators increasingly care about is work delivered per unit of energy — tokens generated, requests served, training throughput. That depends on chip efficiency, on serving software, and on utilisation, none of which the facility metric sees. It is harder to define and it is the number that actually matters.
Technology dependencies are solved by engineering; supply dependencies are solved by building something, which takes years.
Cooling is the largest component of facility overhead, so the cooling choice largely determines the efficiency result.
CoolingEach stage in the distribution chain loses energy; the design of that chain sets the floor on overhead.
Distribution and backupComparable figures require agreement on where energy is measured and over what period; otherwise the numbers are not comparable at all.
Whether heat leaves by evaporating water or through dry coolers sets the water figure and part of the energy one, so the two measures cannot be improved independently.
Heat rejection and waterOther pages in this map that name Efficiency and metrics as something they cannot do without.
What each company supplies at this step, and — where a public figure exists — its share of this specific market — with what that share measures, the period it covers and who published it. Some rows also show the company’s own reported revenue for the segment covering this step, which is a different thing: it says how much this business matters to that company, not how much of the market it holds. Not a ranking and not a recommendation.
Supplies the infrastructure-management software most operators actually read their efficiency numbers out of.
Supplies the monitoring layer bundled with the power and cooling equipment it also sells.
Supplies the building automation and metering the facility-side measurements come from.
Supplies competing building controls and the instrumentation behind the same readings.
Owns the infrastructure-management line that reports capacity and efficiency down to the rack.
Whether water usage becomes the binding permitting constraint in more regions.
Whether the industry converges on a credible measure of useful work per unit of energy.
Whether reported efficiency figures move from design values toward audited operating data.
Each page explains one technology in plain language, states what it depends on, and names companies by what they supply at that step. Company roles are described qualitatively and deliberately carry no market shares, revenue figures or rankings — those change faster than an explainer can, and a stale number is worse than none. Ticker links point at company pages on this site and are provided for reference only.
Nothing here is investment advice, a recommendation, or a forecast. A company named on a page about a technology is not thereby a good investment, and the chokepoints described are structural facts about supply chains rather than predictions about prices. Technology moves; where a page describes something as unresolved or in development, that was true when it was written.