Skip to content

95-5-genai-divide.md

Latest commit

 

History

History
40 lines (28 loc) · 2.38 KB

95-5-genai-divide.md

File metadata and controls

40 lines (28 loc) · 2.38 KB
Raw
title 95/5 GenAI Divide
source-id mit-nanda-genai-divide-2025
wiki-page (02_References/enterprise-ai/wiki/frameworks/scope-execution-grid.md — parent source)
last-synced 2026-06-07

95/5 GenAI Divide

What this reference contains

MIT NANDA 2025 finding: 95% of enterprise GenAI pilots return zero measurable P&L impact. 5% capture nearly all the value. The 5% don't have better models — they make different structural choices about what to build, how to scope it, and what to measure.

Key claims (with citations)

  • 95% zero P&L impact. The vast majority of enterprise GenAI deployments return no measurable financial impact. Source: MIT NANDA 2025.
  • 5% capture disproportionate value. The outlier cohort focuses on narrow-scope, simple-execution use cases (see Scope × Execution Grid). Source: MIT NANDA 2025.
  • Buy vs. Build 2:1 advantage. Firms that buy AI capability rather than build it internally are 2× more likely to capture value. Source: MIT NANDA 2025.
  • Narrow + simple wins. Voice AI for call summarization, document automation for contracts, code generation — the fast wins land in 6–12 months. Full end-to-end orchestration of complex business processes is where most failures live. Source: MIT NANDA 2025, Scope × Execution Grid.
  • Top-quartile GenAI startups reach $1.2M annualized revenue within 6–12 months by staying narrow. Source: MIT NANDA 2025.

Scope × Execution Grid

Narrow Scope Broad Scope
Simple Execution Fast wins ✓ Partial pilots
Complex Execution Early pilots Fails ✗

Start upper-left. Expand outward — either broader scope or more complex execution, not both at once.

When to consult

  • For skill general-idea-diagnostic: the 95/5 divide is the framing for why rigorous Q1–Q4 gating matters before committing.
  • For skill general-roi-gate: cite the 95/5 divide as the empirical basis for why formal gating is necessary, not bureaucratic.
  • For skill process-pilot-design: the Scope × Execution Grid is the primary design tool for sizing the pilot correctly.
  • For skill tech-buy-vs-build: the 2:1 advantage is the empirical anchor for the BUY recommendation.

Source

MIT NANDA GenAI Divide Study 2025. Analyzed enterprise GenAI deployment outcomes across industries; identified the structural patterns separating value-capturing from non-value-capturing deployments.