The Widening AI Value Gap: Build for the Future 2025 (BCG)
TL;DR
Boston Consulting Group’s 2025 global AI-maturity study, released on 30 September 2025. It surveyed 1,250 senior executives and sorted their companies into four groups by a maturity score across 41 capabilities.
- 5% are “future-built”: “at the forefront of AI innovation, systematically building cutting-edge AI capabilities across functions and consistently generating substantial value” (pp. 4, 19). In 2024 the share was 4% (p. 4).
- 35% are “scaling”, “13 percentage points more than in 2024” (p. 3). They “are scaling up their efforts and seeing some returns.”
- 60% are “laggards”: 46% “emerging” plus 14% “stagnating”. They are “not achieving material value at all, reporting minimal revenue and cost gains despite substantial investment” (pp. 3, 5).
- Future-built firms outperform laggards on 1.7x revenue growth, 3.6x three-year total shareholder return, 2.7x return on invested capital, 1.6x EBIT margin and 3.5x patents (Exhibit 1, p. 4). The TSR and EBIT data are external (Capital IQ, Orbis) (pp. 4, 20).
- The press-release headline is an expectation. “Double the Revenue Growth and 40% More Cost Savings” matches the report’s 2028 expected gains in the areas where AI is applied: 2.1x revenue increase and 1.4x cost reduction (Exhibit 2, p. 6; text, p. 5).
- Agents account for “17% of total AI value in 2025” and are expected to reach 29% by 2028. A third of future-built companies use agents, against 12% of scalers (pp. 3, 10).
Key data
| Group | 2024 | 2025 | Definition (p. 19) | Score band (p. 20) |
|---|---|---|---|---|
| Future-built | 4% | 5% | at the forefront; consistently generating substantial value | >75–100 |
| Scaling | 22% | 35% | AI strategy and advanced capabilities, scaling them, “starting to generate value” | >50–75 |
| Emerging | 49% | 46% | foundational capabilities, initial experimentation, “struggling to scale and generate value” | >25–50 |
| Stagnating | 25% | 14% | minimal or no AI action; “not generating value” | 0–25 |
Source for both years: Exhibit 1, p. 4 (n = 1,250 for 2025). “Laggards” = emerging + stagnating.
| Outcome, future-built vs laggards | Value | Page |
|---|---|---|
| Revenue growth | 1.7x | 4, 5 |
| Three-year TSR (June 2022–May 2025) | 3.6x | 4 |
| Return on invested capital | 2.7x | 4 |
| EBIT margin | 1.6x | 4, 5 |
| Realised 2024 revenue increase from AI, % of revenue | 6.2% vs 1.2% (5.3x) | 6 |
| Realised 2024 cost reduction from AI | 6.0% vs 2.0% (3.0x) | 6 |
| Expected 2028 revenue increase / cost reduction | 14.2% vs 6.8% (2.1x) / 13.6% vs 9.6% (1.4x) | 6 |
| Planned IT spend / share of IT budget to AI | +26% / up to +64% | 5 |
Method and definitions
- Sample (p. 20): “1,250 CxOs and senior executives who are AI decision makers across more than 25 sectors” in 68 countries and nine industries. Health care is 10% of respondents. Regions: Europe 40% (including Africa and South America), North America 37%, Asia-Pacific 23%. Most firms are large: 10% have under $1 billion in revenue and 42% have $1–9.9 billion. Respondents: 26% C-level, 26% executive (e.g. vice president), 42% function or division leaders, 6% other.
- Score. Respondents rated their company on “41 foundational capabilities, each measured along four clearly defined maturity stages” (p. 19). BCG weighted the capabilities with “random forest, logistic regression, and structural equation modeling” and sorted the weighted scores into four bands (p. 20).
- Self-report. “We drew insights on AI maturity and value from self-reported data … Results reflect the business area that the respondents know best—not always the full company. Unless explicitly stated as realized …, reported numbers reflect expected future impact … responses may be subject to perception bias” (p. 20).
- Fieldwork dates are not given.
- “Maturity”, in BCG’s words, is “the ability to create value at scale” (p. 8).
Scope and reliability
- Consultancy instrument. BCG sells AI transformation work. The report closes with BCG’s own “Future-Built Playbook” (p. 18). It is not peer-reviewed.
- Mostly self-reported, partly expected. The value figures are respondent estimates for their own business area. Several are forecasts for 2028. The financial multiples come from external data, but BCG notes that “overall company performance has multiple drivers beyond AI” (p. 20).
- Comparability with 2024. BCG prints 2024 and 2025 shares side by side on the same scale (p. 4). The 2024 report was Where’s the Value in AI? (p. 3). BCG’s October 2024 press release for it said “only 26% of companies have developed the necessary set of capabilities to move beyond proofs of concept and generate tangible value” (found by web search, not ingested). That 26% equals the 2024 scaling (22%) plus future-built (4%) shares printed here.
- One sector table only. The industry exhibit (p. 9, n ≈ 900) covers six industries; health care has no row.
- File date. The PDF on BCG’s publications host is named “October-2025” and was created on 16 October 2025. The cover says September 2025.
Dynamic-capabilities reading
digital-transforming/improving-digital-maturity: the instrument is a maturity score. Future-built firms upskill “more than 50% of the internal workforce” (p. 4) and plan 3x more FTEs for AI upskilling (p. 12).digital-seizing/balancing-digital-portfolios: the report is about moving from pilots to value at scale. Laggards “experiment too widely, spreading their resources over scores of complex workflows” (p. 6).digital-transforming/redesigning-internal-structures: future-built firms co-own AI between business and IT and are “three times as likely as others to appoint a chief AI officer” (p. 12).
How it connects
- enterprise-ai-adoption: a value-at-scale figure (5%) and a middle group (35%) that is growing fast.
- ai-maturity-measurement-comparison: a strict-definition instrument with a published 2024 → 2025 series. Its successor, the Applied AI Index 2026, reports 7.5% future-built and 41% scaling.
- Accenture’s front-runners guide: a similar capability-cluster design with an 8% top group.
- Lukic and Goydan on The So What from BCG: a co-author citing this report’s outperformance figures.
- Beauchene on The So What from BCG: a co-author on few-process transformation against tool rollout, from a different BCG survey.
- Dutt et al. (Bain and OpenAI) and micro-productivity-trap: workflow reinvention against isolated use cases.
- HBS AI Institute on change resilience: quotes a BCG survey figure of 26% of organisations achieving value; the 2024 row of Exhibit 1 shows where it comes from.
- MIT NANDA’s GenAI Divide: another 2025 figure of about 5%, on a different instrument and denominator.
What was actually ingested
The full 24-page report PDF from media-publications.bcg.com, converted with pdftotext -layout. Exhibits 1 and the respondent profile (pp. 4, 20) were checked against page renders. bcg.com refuses plain curl. The press release and publication page were read with WebFetch for the headline, date and quoted authors only.
Linked entities and concepts
- Boston Consulting Group: publisher.
- enterprise-ai-adoption, micro-productivity-trap: concepts.
- ai-maturity-measurement-comparison: thread.
Dangling (deferred; note that thirteen of these fifteen authors also sign the 2026 Applied AI Index, so the second-source rule applies at merge): Jessica Apotheker, Vinciane Beauchene (guest on 2026-10-20-beauchene-bcg-so-what-ai-strategy-more-than-rollout), Nicolas de Bellefonds (BCG global AI leader), Patrick Forth, Marc Roman Franke, Michael Grebe, Nina Kataeva, Santeri Kirvelä, Djon Kleine, Romain de Laubier, Vladimir Lukic (guest on 2026-08-25-lukic-goydan-bcg-so-what-ai-costs-a-fortune), Amanda Luther, Mary Martin, Jeff Walters, Christoph Schweizer (BCG CEO).