Hub 01 Power Systems

Why a data center is a power business that happens to house computers: IT MW vs utility MW, the scale of modern facilities (a 100 MW campus draws like a small city; the frontier is gigawatt-scale AI campuses), and why power availability — not land, not capital — sets site feasibility, schedule, and cost. Teaches the binding-constraint logic behind that claim: land, capital, equipment, and labor are substitutable, grid power is not. Establishes the vocabulary (capacity, load, MW) and PUE literacy (common values, how they changed, what counts as efficient), plus the mental model every later lesson assumes: follow the megawatts and you find the project. Builds on: nothing. The first lesson in the curriculum.

Traces power from the utility feed through substation, MV switchgear, step-down transformers, LV distribution, UPS (static and rotary/DRUPS architectures), and busway/PDU to the rack — every conversion point and what each package does. This defines the buying taxonomy: each link is a separate package with its own vendors, lead times, and specification, and the order of the chain is the order in which buys must be released. Builds on: 01.10 — the vocabulary and the constraint logic; now the physical chain.

N, N+1, 2N, and 2N+1 against the Uptime Institute Tier I-IV framework — what "concurrently maintainable" vs "fault tolerant" demands of the electrical design. Redundancy is the single biggest multiplier on equipment quantities (the Tier III to Tier IV jump roughly doubles MEP spend per MW), so every redundancy decision is a procurement scope decision made years before the PO. Builds on: 01.20 — redundancy multiplies the equipment quantities the chain defines.

The current market: large transformers at 140-160+ weeks, MV switchgear at 44-104 weeks, generators and UPS on 30-110 week horizons, and a vendor base concentrated in ABB, Schneider Electric, Vertiv, and Eaton. How hyperscalers reserving transformers by the dozen concentrates factory capacity, why suppliers now prioritize "credible" projects, and how a buyer without scale secures production slots in a seller's market. Builds on: 01.20 (the packages and vendors), 01.30 (why redundancy choices intensify slot competition).

Why grid connection — not capital or construction — is now the gating item: multi-year interconnection queues, the years between agreement and energization, and regional chokepoints. Power must be treated as a procured input with its own lead time; mitigations (BESS as bridge-to-grid, behind-the-meter generation, early power analysis during site selection) belong in the buying strategy. Builds on: 01.20 (what connects), 01.40 (lead-time thinking, applied to the grid).

How to buy technology that isn't standard yet. Battery storage moving from backup accessory to core infrastructure: qualifying BESS integrators and new-chemistry suppliers, pilot structures, peak-shaving economics, BESS configured as UPS alternatives. Plus what's coming — 800V DC distribution, solid-state transformers — and the discipline for evaluating emerging line items before they become standard packages: vendor viability, avoiding stranded assets on dead-end platforms, keeping a second source alive. Builds on: 01.20 (storage's place in the chain), 01.50 (bridge-to-grid and load-smoothing at the interconnection point).

Prefabricated power skids as a procurement strategy: trading factory lead time and site labor risk for earlier vendor commitment and tighter specifications. When modular beats stick-built — speed, repeatability across campuses, reduced on-site commissioning risk — and what it demands of the buyer in submittals and Factory Acceptance Testing (witnessed testing at the vendor's plant before shipment; taught fully in 03.60). Builds on: 01.20 (the packages being prefabricated), 01.40 (the lead-time pressure that makes modular attractive).

When the grid can't deliver on time, generating on site stops being a backup plan and becomes the project: gas turbines and reciprocating engines as primary power today, fuel cells and SMRs on the horizon. When on-site generation pencils out against waiting for interconnection, what changes for the buyer (air permits, fuel contracts, the O&M burden, the utility demoted to backup), and how it interacts with the interconnection strategy from 01.50. Builds on: 01.10 (the constraint it answers), 01.50 (the grid alternative it replaces).

Key sources: Wood Mackenzie transformer/equipment lead-time research; Terrapin CG 2026 critical equipment lead-time survey; Data Center Knowledge; POWER Magazine.