A short lesson sequence. Work the notes in numbered order; each builds on the last.
Part III: The Function. The processes that move a buy from need to payment, and the systems that run them.
Assumes: Part II complete. Processes formalize the craft — the cycle names what sourcing, costing, and contracting taught.
The S2P cycle
- Spend analysis: what we buy, from whom, at what price. The baseline for every sourcing decision.
- Sourcing: identifying, evaluating, and selecting vendors (RFI, then RFP or RFQ, then award).
- Contracting: negotiating terms, CLM, DOA approvals.
- Requisitioning and ordering: the PO, approvals, releases against contracts.
- Receiving: confirming delivery, inspection, acceptance. Critical for LLE, where delivery and installation are split under OFCI.
- Invoicing and payment: three-way match (PO, receipt, invoice agree), then pay.
What breaks in data centers
- The OFCI split breaks the textbook cycle: the owner sources and buys, the contractor receives and installs. Receiving, acceptance, and warranty handoffs need explicit design or they fall between the parties.
- LLE submittal cycles (specification, shop drawings, approval, PO) sit on the critical path. A generic S2P process that treats a transformer like office supplies will miss the schedule.
- Terms: Glossary (P2P, CLM, vendor master, DOA).
Learning path
09.10 The Source-to-Pay Cycle End to End
The six stages and where value leaks at the handoffs between them. Establishes the vocabulary (S2P vs. P2P vs. e-procurement) and the discipline every later lesson assumes. Builds on: Hubs 05 and 07 — the cycle names what the Craft taught.
09.20 Procurement Data and Analytics
The data foundation the cycle stands on: spend analytics as the first investment (09.10's spend analysis stage needs data to analyze), supplier-master hygiene, and dashboard discipline around spend under management, maverick spend, cycle time, and savings realization. Includes the savings-measurement trap: without baseline integrity, reported savings are fiction. Builds on: 09.10 — the spend-analysis stage that needs data.
09.30 Where Generic S2P Breaks on Long-Lead Capital Equipment
The three failure modes that make textbook procurement dangerous on data center builds: OFCI splits stranding accountability between owner-buyer and contractor-installer, submittal cycles sitting on the critical path, and factory-slot economics where a quoting error costs a place in the manufacturing queue. Current reality: large power transformers at 128-160+ week lead times (Wood Mackenzie), high-horsepower gensets reportedly sold out through 2028, developers buying production slots before finalizing sites. Builds on: 09.10 (the cycle), 03.10 (the OFCI split), 03.50 (submittal cycles), 01.40 (factory slots).
09.40 The Systems Stack: P2P, CLM, and Supplier Master
Procurement systems architecture — which system owns the vendor record, which owns the contract, how the layers integrate — and the five-stage maturity model from manual/reactive to AI-native. The key judgment: diagnosing which stage an organization is actually in, since most transformation failures come from expecting the wrong stage to solve the problem. Builds on: 09.10 (the processes), 09.20 (the data they hold).
09.50 E-Sourcing and Automation That Pays
The automation hierarchy by ROI: touchless invoice processing and PO-first discipline first, e-RFx and catalog buying second, everything else after. Covers the unglamorous prerequisites (clean vendor master, budget validation at requisition) that determine whether automation delivers or just accelerates chaos. Builds on: 09.40 (the systems), 09.20 (the prerequisites).
09.60 AI in Procurement: Signal vs. Hype
Where AI demonstrably works — compressing the sourcing sequence from spend analysis to bid evaluation, contract redlining and deviation flagging, supplier risk monitoring, tail-spend management — versus where judgment stays human. Grounded in 2025-2026 evidence: 10-25% early-adopter productivity gains (Hackett Group), 12-29% savings in AI-assisted sourcing events (McKinsey). The governing principle: autonomous preparation, not autonomous procurement — humans keep selection, negotiation, and accountability. Builds on: 09.50 (the automation hierarchy), 09.10 (the sequence it compresses).
Key sources: Suplari on the procurement tech stack (2026); The Hackett Group on agentic AI and procurement; pv magazine USA on transformer supply constraints; IECI on supply chain bottlenecks for contractors.