How a Fortune 50 Telecom Found Hidden Capacity and Deferred Millions in CapEx

A Fortune 50 U.S. telecom operator manages a nationwide portfolio of mission-critical facilities. Over 15+ years, the electrical distribution diagrams mapping power flow through each site, documents known as "one-lines," accumulated quietly without keeping pace with reality. Personnel turned over, systems were upgraded, loads were reconfigured, and new sites arrived through mergers and acquisitions, each carrying its own documentation standards and gaps. The one-lines weren't always updated. Slowly, steadily, the gap between what the diagrams showed and what actually existed inside the walls grew wider.

These documents were the foundation for every capacity decision the Critical Facilities Engineering team made: where to expand, what to consolidate, when to defer construction. But they were delivered in formats like .dwg, .pdf, and .png, which made automated extraction and analysis impossible. Every number had to be pulled manually. Every discrepancy required a site visit to resolve.

The consequences compounded:

The engineering team knew the one-lines were unreliable. But they had no alternative source of truth. Every expansion decision carried the same risk: either over-build and waste capital, or under-estimate and face costly outages and SLA violations. The margin of error was measured in millions of dollars per facility.

Verdigris deployed 1,400+ electrical panels across the operator's nationwide portfolio in under 10 months, with 28,000+ circuits validated and streaming real-time data to the platform. This was not monitoring for its own sake. The deployment was designed to replace the manual audit cycle that had been the only way to answer capacity questions.

The engineering team integrated Verdigris APIs into their corporate planning systems. For the first time, they had portfolio-wide PUE tracking built on measured load data, not estimates pulled from stale drawings. Capacity forecasting models could now use real power consumption patterns at the site and rack level, updated continuously, rather than snapshots taken during periodic audits that were outdated by the time the report was written.

The depth of integration went further than most monitoring deployments. The operator now includes Verdigris in the design specifications for new facilities and acquisitions. When the team evaluates a new site or plans an expansion, Verdigris instrumentation is part of the base infrastructure, specified alongside switchgear and generators. That is not something an engineering team does with a tool they consider optional.

The stale one-lines described a specific electrical distribution hierarchy at each facility: which panels feed which breakers, which breakers supply which loads, which racks connect to which DC power plants. The question was whether those documented relationships still reflected reality. Rather than audit each one by hand, Verdigris auto-discovered the actual hierarchy from measured data and compared it against the paper record.

Panel-to-breaker hierarchy. High-frequency power data captured at 8 kHz was used to correlate power consumption between panels and breakers. When a breaker draws power from a specific panel, their electrical signatures move together: correlated power patterns, matching load profiles, near-identical RMS behavior. At a pilot facility, Verdigris tested this against the four known hierarchy pairs in the building. All four were matched with greater than 0.98 correlation. Where the documented hierarchy listed a connection, the data confirmed it. Where the data showed a connection the documents did not list, the documents were wrong.

Rack-to-plant topology. The racks in a DC power plant share a common voltage bus. When a voltage fluctuation occurs, it propagates to every rack on that plant. By filtering noise from rack-level voltage measurements and clustering racks with correlated voltage shifts, Verdigris mapped which racks connect to which DC plants without relying on the existing plant assignment spreadsheets. Those spreadsheets only covered 88% of the racks in the facility. The remaining 12% were unaccounted for in any documentation. The measured data accounted for all of them.

Upstream hierarchy from BMS. Power summation relationships in BMS data were analyzed to discover which upstream feeds roll up to which downstream distribution points. This is where the data sources proved most valuable together: on one side of the electrical distribution, the BMS had a data quality gap (a missing measurement point), making it impossible to reconstruct the hierarchy from BMS alone. The high-frequency Verdigris data filled that gap, enabling full hierarchy reconstruction across both sides of the distribution.

At one facility, the auto-discovered hierarchy did not match the documented one-lines. The diagrams said the facility's capacity was fully utilized. The measured data said otherwise. The gap between what the paper described and what the electrical signatures revealed was 1 MW of real, usable headroom that had been invisible to every prior capacity assessment.

This was not a theoretical finding. It was a megawatt of power capacity that the engineering team could recover without constructing anything. Finding 1 MW of stranded capacity in a telecom data center typically offers $8 million to $14 million in deferred capital expenditure, because it avoids the immediate cost of building new, comparable capacity. In high-density or AI-optimized facilities, where build-out costs per MW are higher, this deferral value can exceed $15 million to $20 million.

That single facility validated the model. Applying the same methodology across the nationwide portfolio, the operator projects approximately $100 million in total CapEx deferral as stranded capacity is identified and recovered site by site.

The operational impact extends beyond the capital savings. Site capacity assessments that previously required multi-week physical investigations now take hours: the engineering team pulls measured hierarchy and load data directly from the platform instead of dispatching personnel to each facility. During power events, high-frequency waveform capture provides sub-cycle visibility for root cause analysis and management reporting, evidence that did not exist at any resolution before.

Verdigris has been instrumental in refining our capital planning processes, enabling us to make more informed and strategic investment decisions across our facilities.