---
type: "synthesis"
sources: ["spine"]
tags: ["pilot-purgatory", "scaling", "failure-rates", "execution"]
id: "cd-pilot-purgatory-to-scale"
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-seg-spine"
sourceUrl: "(unified vault: 9 sources)"
sourceTitle: "HBR — Strategic Spine — value thesis & how much to bet"
---
The corpus opens, again and again, on the same alarming backdrop — most AI value never scales — and then diverges on the diagnosis, which is itself instructive.

**The shared statistics.** A047 and A061 both cite BCG's finding that ~60% of AI investors get no material value and only ~5% capture value at scale; A047 adds McKinsey (88% adopt, only 39% see EBIT impact, usually <5%) and Deloitte (2–4 year ROI). A055 cites its own (unverified) figure that 42% of companies abandoned most AI initiatives in 2025. A098 adds the 'hope for the best' pathology ([[quote-hope-for-the-best]]).

**Four diagnoses of the same failure:**
- A061: uncoordinated pilots ([[claim-piecemeal-drain]]) — cure is portfolio discipline.
- A047: mis-measurement and tactical over-clustering ([[claim-tactical-spending-cluster]]) — cure is type-specific metrics.
- A055: ambition/reality misalignment ([[claim-misalignment-causes-failure]]) — cure is the matrix.
- A098: stopping at individual gains ([[claim-individual-gains-insufficient]]) — cure is climbing the pyramid.

**The convergent verdict** is captured in A055's [[quote-scaling-vs-pilots]]: the next decade is won by companies that scale, not those that pilot most. Notably, several of these 'failure' statistics are flagged in enrichment as unverified or source-specific (A055's percentages, A047's Deloitte figure) even as the *directional* claim is robust — a distinction a downstream agent must preserve. Scaling requires solving the human constraint ([[cd-absorptive-capacity-binding-constraint]]) and abandoning the wrong lens ([[cd-roi-is-the-wrong-lens]]).