Electrical scope for AI and cloud hyperscale campuses across Texas. Operator-specific design topologies, paralleled generator plants, 2N critical distribution, and Level 5 IST commissioning.
Hyperscale data centers aren’t bigger versions of enterprise data centers — they’re a different category of project entirely. A single hyperscale campus typically lands between 100MW and 1GW+ of IT capacity across multiple buildings, with each individual data hall running 30MW to 100MW of critical load. Power density per rack has climbed from 5–10kW air-cooled five years ago to 50–100kW+ liquid-cooled in current AI-focused builds. The redundancy topologies, generator plant sizing, and commissioning protocols all operate at scales that enterprise contractors don’t encounter.
We deliver hyperscale electrical scope under EPC and GC contracts for the operators driving Texas’ hyperscale build-out: AWS, Microsoft Azure, Google Cloud, Meta, and the colocation operators (Equinix, Digital Realty, QTS, Stack) that serve hyperscale customers. Each operator has its own internal design standards, commissioning protocols, and contractor qualification processes. We build to operator design intent, not generic Uptime tier classifications.
Customer-owned substation construction taking utility service at 35kV, 69kV, or 138kV class transmission depending on campus load. Coordination with Oncor, CPS Energy, LCRA, and ERCOT for transmission planning, primary service, and parallel-operation agreements.
Diesel and gas standby generator plants from 30MW through 200MW+ aggregate across multiple data halls. 2–3MW units paralleled through ASCO 7000, Russelectric, or operator-specified switchgear. NFPA 110 Level 1 EPSS with 24/7 fuel system management and automated load-bank testing infrastructure.
Metal-clad MV switchgear per ANSI C37.20.2 distributed across the campus. SEL or Schweitzer relay protection (SEL-351, SEL-487, SEL-787 platforms) with operator-customized settings. Main-tie-main and ring-bus configurations for concurrent maintainability through equipment failures.
Static UPS plants (Eaton, Vertiv, ABB) in 2N or 2(N+1) configuration scaled to per-hall load. Battery rooms with thermal management, fire suppression, and BMS monitoring. STS integration and critical bus distribution to PDUs and RPPs.
PDU and RPP installations supporting 30–100kW+ per rack for AI training workloads. Overhead busway distribution (typically 800A–1600A per zone), branch circuit monitoring, intelligent rack PDU support, and coordination with the liquid-cooling CDU power plant.
Factory witness testing, Level 1–4 site commissioning, and Level 5 Integrated Systems Testing. Pull-the-plug scenarios across utility, generator, and UPS, load-bank stress testing per operator protocols, and full failure-mode validation before customer-load energization.
Hyperscale operators don’t use generic Uptime Institute tier classifications. Each has internal design topology standards that exceed Tier IV on some axes and trade off on others, with operator-specific requirements for:
We build to operator design intent. Our preconstruction team studies the operator’s issued design package and produces field execution that matches it — not what a generic Tier III spec would suggest.
Texas hyperscale build-out is concentrated in three corridors: DFW (the established #1 US data center market), the San Antonio–Austin axis (driven by Microsoft, Meta, and Google site selections over the past three years), and the Permian Basin (emerging as an AI-build-out corridor due to power availability and cooling resources). We have project capability across all three.
Typically through the EPC or GC selected by the operator. Hyperscale operators procure construction through their selected GC, and electrical sub work flows under that GC. We engage with the operator’s design and commissioning teams during preconstruction even when contracted under the GC.
50kW+ rack densities require electrical infrastructure that wouldn’t support current air-cooled enterprise loads. We build to current AI rack power specifications including NVIDIA reference designs and operator-specific deployments. Branch circuit ampacity, busway sizing, and rack PDU selection all shift in this density range.
For campus loads exceeding the largest available distribution-class service, hyperscale projects take utility service at transmission voltages (138kV or higher). We work with the utility transmission planning group through ERCOT generation interconnection processes and coordinate the customer-side substation construction.
Send us your design package, site plan, and target energization. We’ll engage during preconstruction with utility coordination and equipment lead-time discipline.