Due Diligence for the AI Buildout
A seventeen-essay framework for AI infrastructure due diligence. Each essay maps one workstream from scoping through post-close validation, with question banks, worked examples, and cross-references to the AI Power Chain vendor screen. Reference material for PE sponsors, corporate development teams, and target-side operators preparing for capital events.
01The essays
-
01The scoping decisionWhat DD covers before the data room opens
-
02Team, timeline, and the anatomy of a data roomFive internal DD roles, sequencing, and reading the seller's room
-
03Market and demand-side DDTesting whether demand is what the CIM says it is. Bottom-up sizing
-
04Customer, pricing, and competitive DDWhat the target captures from the demand base. Concentration + pricing power
-
05Product architecture and technology DDProduct-spec fit, roadmap evidence tiers, differentiation audit
-
06Manufacturing, IP, and supply chain DDCapacity vs demand, yield, IP audit, tier 1/2 supply chain risk
-
07Quality of earnings and working capital DDAdd-back discipline, dynamic working capital, QoE walk to normalised EBITDA
-
08Capex, growth economics, projection stress-testingCapex per revenue $, unit economics, three downside scenarios
-
09Organisational, key personnel, culture, systems DDWhether people + culture + systems can execute the DD thesis
-
10Legal and contract DDCorporate structure, change-of-control, non-competes, IP, litigation
-
11Regulatory, CFIUS, environmental, cyber DDApprovals that gate closing, liabilities that follow, cyber that walks the R&W
-
12Deal structure, R&W, escrow architectureHow DD findings translate into the transaction paper
-
13Financing structure, covenants, closing riskHow the deal gets paid for; covenants; what stops closing after signing
-
14Post-close validation + the DD improvement loop100-day validation, six-month scorecard, feeding lessons back into DD
-
15The Question Banks250+ questions organised by workstream + AI Power Chain layer
-
16Seller-side prep: the DD workbookThe framework inverted for target-side founders + CEOs + sell-side teams
The AI Power Chain vendor screen
Named vendors across every layer, scored on moat, execution risk, and chokepoint. Cross-referenced from every essay in this series.
Vendor screen →WS15 and WS16 rebuilt as operating layers, not documents
WS15 · The Question Banks is now positioned as the interrogation layer of the DD framework, not a checklist. Includes the Killer 42 — three questions per workstream that most often kill or reprice AI-infrastructure deals, each with evidence, respondent, falsifier, and downstream-workstream metadata. Question tiers (Core / Conditional / Deep), respondent guidance, and a worked question-routing chain (WS05 → WS06 → WS08 → WS13) show how one finding routes into the next workstream.
WS16 · Seller-side prep is now the self-diligence and buyer-simulation system, not a data-room checklist. Includes the CIM Claim Register (every material CIM claim reconciled to primary evidence with expected buyer challenge pre-answered), the five-part remediation taxonomy (Eliminate / Evidence / Explain / Disclose / Allocate), P / D / M evidence grading, and the DD-ready test that determines whether the seller is actually ready.
The four-piece stack now reads: WS01–14 (what to understand) → WS15 (what to ask) → WS16 (what the seller should already have proven) → DD Evidence Matrix (can we prove it).