---
id: "claim-widening-performance-gap"
type: "claim"
source_timestamps: ["¶3"]
tags: ["performance-metrics", "competitive-dynamics"]
related: ["concept-compounding-ai-capabilities", "quote-widening-gap", "question-laggard-catchup-viability"]
confidence: "high"
testable: true
speakers: ["Bruce Lawler", "Vijay D'Silva", "Vivek Arora"]
source_url: "https://hbr.org/2025/01/what-companies-succeeding-with-ai-do-differently"
source_title: "What Companies Succeeding with AI Do Differently"
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-cl-89-companies-succeeding-with-ai"
sourceUrl: "https://hbr.org/2025/01/what-companies-succeeding-with-ai-do-differently"
sourceTitle: "What Companies Succeeding with AI Do Differently"
---
# The AI performance gap between leaders and laggards is widening

**Claim:** The performance advantage of AI leaders (the **top 25% of respondents**) over the bottom half of companies has increased significantly.

- **2021:** leaders saw performance levels **2.7x** that of the bottom half.
- **2023:** this multiplier increased to **3.8x**.

The authors attribute the widening to the [[concept-compounding-ai-capabilities|compounding effect]] of building differentiated AI capabilities over time. See the verbatim figure in [[quote-widening-gap]] and the strategic follow-on in [[question-laggard-catchup-viability]].

**Confidence: high.** The specific 2.7x → 3.8x figures are directly supported by secondary reporting of the MIT–McKinsey study. The *interpretive* label "compounding" is a reasonable synthesis rather than a directly measured metric. Testable via longitudinal tracking of individual firms.


## Related across articles
- [[claim-95-percent-failure]]
- [[claim-marginal-business-impact]]
- [[concept-compounding-ai-capabilities]]