Ivan Nosov

A EN KPMG·Survey·

Global AI Pulse Q3 2026

A · Strong evidence A large, documented quarterly executive survey (2,131 senior leaders, 20 countries) from a Big Four firm with stated bases per figure, but all outcomes are self-reported and KPMG sells the remedies it recommends.

Summary based on the full text·Original: about 40 minutes to read

In short

The third 2026 edition of KPMG International's Global AI Pulse (September 2026), a survey of 2,131 senior leaders in 20 countries. The question has moved from orchestration (Q1) and value realization (Q2) to what running AI at scale requires. Its main finding: most firms now see what AI costs and say it creates value, but only 12% consistently judge that value against the cost. Regional maturity gaps have closed from 16 to 8 points, so KPMG argues the advantage shifts from being early to running AI well.

Key numbers

  • 12% vs 89%organizations that consistently assess AI value against cost vs organizations reporting measurable value from AI
  • 55% / 86%operate a formal AI harness layer, whole sample / organizations with established ROI
  • 23%include model routing in their AI management layer, the least common capability
  • US$210Maverage planned AI investment over the next 12 months, up from US$186M in Q1

As reported by the source.

Open PDF on assets.kpmg.com

English·PDF·opens assets.kpmg.com

Figure from the source

Exhibit 1

Cost controls by stage of AI journey

Dot chart. Share of respondents with each control in place, experimentation vs established ROI: AI cost monitoring dashboards 43% vs 77%, cost review at AI approval 50% vs 73%, usage or token budgets 31% vs 46%. Value assessed against cost consistently: 48% of established-ROI respondents.
Firms with established ROI run cost dashboards far more often (77% vs 43%).Source: KPMG International, Global AI Pulse Q3 2026, September 2026 (n=2,131).

What it reports

  • Scaling AI or beyond: Americas 64%, Asia Pacific 61%, EMEA 56% (Q1: 56%, 43%, 40%). Experimenters fell from 22% to 13%; established ROI rose from 8% to 10%.
  • Planned investment: global US$210M (+13%), EMEA US$203M (+29%), Americas US$191M (+7%), Asia Pacific US$246M (unchanged).
  • AI harness layer: 55% formal; 5% no formal approach. Among established-ROI firms 86% formal, 65% enterprise-wide; among experimenters 31% and 10%.
  • Layer contents: security and identity 43%, data access 43%, output monitoring 41%, cost tracking 37%, human review 34%, model routing 23% (17% of experimenters, 41% of established-ROI).
  • Accountability for AI-informed decisions at C-suite level or above: 53% (35% a named executive, 18% CEO or executive committee).
  • Significant employee adoption of agents: 34%, up from 25% in Q1; 48% in established-ROI firms vs 15% at proof of concept.
  • Cyber: 71% of established-ROI firms put cyber and data security in the AI budget vs 36% of experimenters; AI-assisted cyber defense 58% vs 8%.
  • Sovereignty: 21% enterprise-wide strategy, 51% formal criteria on most or selected decisions, 8% slowed deployment because of it.
  • Cost: 61% review costs at approval, 59% monitor in operation, 12% consistently assess value against cost (48% among established-ROI firms).

How they got it

Online survey fielded 23 July to 26 August 2026 among presidents, C-suite, SVPs, VPs and directors reporting to the C-suite (Americas n=670, Asia Pacific n=540, EMEA n=921). Companies needed at least US$50M revenue, or US$100M in ten larger markets. The maturity comparison sets experimenters (n=268) against firms reporting established ROI (n=208). Investment figures are weighted means adjusted for size and region.

Limits

  • "Measurable value" and "established ROI" are self-assessed; no financial verification.
  • Maturity-stage comparisons are correlations; firms reporting ROI may simply be larger or better run.
  • KPMG notes that quarter-on-quarter moves are small and should be read as direction, and the report closes with KPMG's own service offer.

Why it matters for companies

The useful number is the 12%: most firms can show AI spend but cannot link it to outcomes, which leaves budget decisions to opinion. The report gives a concrete checklist for the management layer: controls, routing work to the cheapest adequate model, a named accountable executive, and cost tied to outcomes. Firms relying on one model provider carry operational, legal and commercial risk in one place; the most mature group spreads it. A traditional company can set up the cost-to-value link before scaling further, since retrofitting it later is harder.

Summary based on the full text. Read the original for the full argument and data.Open PDF on assets.kpmg.com

Same week

All of Week 39