A short lesson sequence. Work the notes in numbered order; each builds on the last.
Part II: The Craft. How to know what something should cost, whether the vendor can deliver it, and how to run the buying process.
Assumes: Part I complete. You can't buy what you can't describe — the packages, lead times, and vendor landscape come from the Domain.
Cost assessment
- Should-cost modeling: build an independent estimate from materials, labor, factory capacity, and reasonable margin, then test bids against it. The procurement matrix (cost ranges per package) is the starting dataset.
- Bid leveling: normalize competing bids to identical scope, terms, and exclusions before comparing. Bids that aren't leveled aren't comparable.
- TCO vs price: the cheapest bid can be the most expensive contract. Taught in 06.20.
- Market timing: LLE pricing moves with factory backlog. A 140+ week transformer market is a seller's market; leverage sits with whoever holds production slots.
Capability assessment
- Technical: can they build it to spec? Reference installs, factory audits, FAT history.
- Capacity: do they hold production slots when we need them? For LLE, backlog and allocation matter more than list price.
- Financial: will they survive a 100+ week manufacturing window? This matters most for single-source packages.
- Delivery: logistics capability — staging, rigging, and placement of multi-ton equipment.
Learning path
05.10 The Long-Lead Equipment Buying Playbook
Why procurement sets the date on a data center build: hundred-week lead times mean the order date for the longest item fixes the energization date. Ordering at land close on partial design, specifying by configuration so multiple manufacturers can bid, and confirming factory slots in writing — a PO date alone proves little when fabrication starts a quarter later. Builds on: 01.20 (the packages), 01.40 (the lead times), 03.50 (the critical path).
05.20 Should-Cost Modeling for Capital Equipment
Building an independent cost model for a transformer, UPS line, or chiller package from materials, factory capacity, labor, overhead, and reasonable margin — then using the gap against supplier quotes as the negotiation agenda. Adapting teardown-style should-cost methods to low-volume, engineer-to-order electrical equipment where factory backlog, not unit economics, drives price. Builds on: 05.10 (what you're costing), 01.20 (package anatomy).
05.30 Bid Leveling for Equipment Packages
Comparing bids never written to the same scope: set the scope baseline before bids arrive, normalize line by line (exclusions added back at carried cost, double-counts removed, commercial terms equalized), rank only on adjusted numbers. The data center traps — OFCI-split packages with delivery and installation in different contracts, submittal-cycle costs buried in qualifications. Builds on: 05.20 (the cost model as normalization baseline), 03.10 (the OFCI split).
05.40 Vendor Capability Assessment: Technical, Financial, Capacity
The three-lens qualification for a factory that will hold your schedule for two years: technical fit (reference installs, factory audits, FAT history), financial durability (will they survive a 100+ week manufacturing window), real capacity (backlog, slot allocation, not brochure throughput). How to read a vendor's order book position, and why certification regimes are becoming qualification shortcuts. Builds on: 05.10 (why capacity and durability matter), 01.40 (the backlog market).
05.50 Strategic Supplier Partnerships and Capacity Reservation
Beyond transactional buying: master purchase agreements, framework agreements, reserved production slots matched to construction need dates. Standardization and SKU catalogs as the commercial enabler, the multi-sourcing trade-off (resilience in procurement vs. complexity in operations and spares), and how hyperscaler bulk-buying reshapes partnership for everyone else. Builds on: 05.40 (partner with the qualified), 04.10 (how hyperscaler behavior shapes the market).
05.60 Negotiating in a Seller's Market
What leverage actually exists when factories are full: slot timing, escalation and indexation clauses, tariff exposure and country-of-origin planning, tender structures with realistic validity periods, deposits as cheap insurance against delay costs. Adapting tenders to volatile markets instead of demanding fixed prices no one will honor. Builds on: 05.20 (the cost model is the negotiation agenda), 05.50 (the market structure), 01.40 (the seller's market).
05.70 Expediting: Chasing the Factory
The discipline between PO and delivery: production monitoring, milestone verification, FAT coordination, and the escalation path when a factory slips. Why expediting is a dedicated role for LLE — a PO without follow-up is a wish. What good expediting reports contain and how they feed the construction schedule. Builds on: 05.10 (the PO), 05.40 (factory knowledge), 03.50 (the schedule it feeds).
05.80 Logistics: Getting It to the Site
The last mile for multi-ton equipment: Incoterms (EXW/FOB/DDP) and where risk transfers, rigging and heavy-haul planning, staging when the site isn't ready, receiving discipline — open-and-inspect on delivery, because concealed damage found at installation becomes nobody's liability. Logistics choices connect back to the critical path. Builds on: 05.70 (execution follows the factory chase), 03.50 (the critical path it serves).
Key sources: Global Data Center Hub on equipment procurement strategy; Data Center Knowledge on tariffs and supply shortages; IECI on supply chain bottlenecks; Power Magazine on the transformer market.