Reading Maturity From the Outside
The value of the maturity gradient is that it can be read from outside the fence. An analyst, a counterparty, a customer evaluating a colocation contract, or an investor assessing an operator does not need root access to the DCIM platform to place a facility on the gradient. They need the right questions, and the discipline to weigh the answers against what a mature operation would say.
The most revealing question is about available capacity, and specifically how it is derived. Ask an operator how much capacity remains and how they know. A stage-one operator answers from nameplate math: feed size minus installed nameplate load, a number that overstates reality. A mature operator answers from measured, coupled data: here is real power headroom at these positions, here is cooling deliverable there, here is where the three coincide and here is where they do not. The shape of the answer, whether it is a single paper number or a position-specific joint constraint, immediately reveals the stage. An operator who cannot distinguish paper capacity from usable capacity is telling you they do not model the coupling, which places them below stage three regardless of what their brochure claims.
The second question is about provisioning speed and its failure rate. How long from a decision to add load to that load running safely, and how often does a provisioning action cause an unexpected thermal or power problem. Mature operators provision fast and rarely surprise themselves, because the twin answered the risk question before the change was made. Immature operators either provision slowly, because every change requires cautious manual analysis and a nervous walk of the floor, or provision fast and get burned, discovering constraints by tripping them. Speed with a low surprise rate is the tell of a predictive facility model at work.
The third question concerns failure behavior. Ask how the facility detects an emerging physical problem, a fouling coil, a failing power supply, a developing hot spot, and how early. A stage-one shop finds problems when they become symptoms: something is hot, something tripped. A mature operation detects the divergence between modeled and measured state and catches the problem while it is still a drift, not yet a failure. The earliness of detection is a direct readout of whether a live model is running underneath, because only a model gives you an expected value to deviate from. An operator who describes catching problems early, by anomaly rather than by alarm, is describing a twin whether or not they use the word.
The fourth question probes automation and its limits, carefully. Ask what the system does on its own and what it defers to humans. The revealing answer is not maximum automation, it is calibrated automation: a mature operator can articulate precisely which loops are closed, within what envelopes, and why others are deliberately left open. That specificity signals a real closed-loop capability with real engineering judgment about its bounds. Vague claims of an autonomous facility, without the boundaries drawn, signal marketing rather than a running system. The presence of thoughtful limits is more credible than the absence of them.
Cross-cutting all four is a caution: do not let compute-side sophistication stand in for facility maturity. Operators will foreground their orchestration stack, their accelerator fleet, their software. Those are real, and they are on the other nervous system. The facility questions must be asked on their own terms, about power, cooling, and space, or the compute-side dazzle will paper over a physical plant run on folklore. The independence of the two maturities is exactly what makes the specific questioning necessary.
Faraday's read is that facility maturity correlates with the operational virtues that are hardest to fake and most predictive of how an operator behaves under stress: honesty about capacity, discipline in provisioning, early detection of trouble, and calibrated rather than reckless automation. These are not separable from the DCIM stage; they are what the stage measures. An operator who has climbed to stage three or four has, along the way, been forced to acquire exactly the habits that make infrastructure reliable. So the gradient is not merely a technical assessment. It is a character reference for the operation, legible to anyone who knows to ask about the building rather than the machines inside it, and to weigh the answers against the standard a mature plant would meet.