Executive Summary
- EY’s AI Pulse Survey (n=500 SVP+ leaders, Oct 2025) finds 96% of AI-investing organizations report productivity gains — but 65% cannot directly tie those gains to AI adoption. The attribution problem is now the central barrier to continued investment.
- The 2025 Work Reimagined Survey (n=15,000 employees + 1,500 employers, 29 countries) reveals 88% of employees use AI at work — yet only 5% use it in advanced, work-transforming ways. The gap between access and capability is vast.
- EY’s Responsible AI Survey (n=975 C-suite leaders, 21 countries) reports 99% of organizations have suffered financial losses from AI-related risks, averaging $4.4M per company. Responsible AI governance correlates with 65% higher likelihood of cost savings.
- CEO headcount-reduction expectations dropped from 46% to 24% in twelve months. Leaders are reinvesting AI-driven productivity gains into R&D, cybersecurity, and employee reskilling — not layoffs.
- Only 12% of employees receive sufficient AI training. Those who do (81+ hours annually) report 14 hours per week in productivity gains — nearly double the median of 8 hours. Training investment is the single largest lever most organizations have not pulled.
The Productivity Paradox: Everyone Gains, Nobody Can Prove It
EY’s fourth US AI Pulse Survey, fielded September–October 2025 among 500 senior leaders across 10 industries, presents an unusual finding: near-universal productivity improvement paired with near-universal inability to measure it.
Ninety-six percent of AI-investing organizations report some productivity gains. Fifty-seven percent call those gains “significant.” Yet 65% of leaders admit they struggle to directly attribute productivity improvements to AI. Sixty-three percent say senior leaders do not always credit gains to AI even when the connection is clear.
This is the attribution gap — and it matters because budget decisions require evidence. Organizations investing $10M or more in AI see significant gains at a 71% rate, compared to 52% for those spending less. The investment threshold correlates with outcomes, but without reliable attribution, CFOs cannot build the business case for crossing that threshold.
The budget trajectory tells its own story: 27% of organizations currently allocate more than 25% of their IT budget to AI. Fifty-two percent expect to reach that level next year. The money is moving even without clean measurement — a signal of executive conviction, but also of risk if results disappoint.
The 5% Problem: Access Without Capability
The EY 2025 Work Reimagined Survey — 15,000 employees and 1,500 employers across 29 countries and 19 sectors — quantifies a pattern familiar from the BCG and McKinsey data: adoption is widespread, but depth is shallow.
Eighty-eight percent of employees use AI at work. Five percent use it in advanced ways that transform how they work. The remaining 83% are doing search and summarization — useful, but not the productivity step-change organizations are investing for.
The training data explains the gap. Only 12% of employees receive sufficient AI training to capture full productivity benefits. Employees receiving 81 or more hours of annual AI training report 14 hours per week in productivity gains — versus 8 hours at the median. That is a 75% improvement from training alone, measured in hours returned to the business each week.
One counterintuitive finding: highly trained AI users are 55% more likely to leave their organization. The skills are portable, and the labor market rewards them. Companies that invest in training without investing in retention will fund their competitors’ AI capabilities.
Shadow AI remains pervasive. Between 23% and 58% of employees across sectors use unauthorized AI solutions — a range that tracks with the findings in Deloitte’s 2026 survey (60% employee access, 30% governance readiness).
The Cost of Getting Governance Wrong: $4.4 Million
EY’s Responsible AI Survey (975 C-suite leaders, organizations over $1B revenue, 21 countries, August–September 2025) puts a dollar figure on AI risk that has been missing from the corpus.
Ninety-nine percent of organizations report financial losses from AI-related risks. The average loss is $4.4 million. Sixty-four percent suffered losses exceeding $1 million. The most common risk categories: regulatory non-compliance (57%), negative sustainability impacts (55%), and biased outputs (53%).
The governance knowledge gap is striking: C-suite leaders can correctly identify only 12% of appropriate AI controls. Chief risk officers perform at 11%. The people responsible for AI governance do not know what the controls should be.
Organizations that implement real-time AI monitoring see 65% higher likelihood of cost savings and 34% higher likelihood of revenue improvement compared to those without monitoring. Responsible AI is not a compliance cost — it is a financial performance differentiator.
Sixty-six percent of companies allow citizen AI development. Sixty percent of those have formal policies. Fifty percent lack high visibility into what citizen developers build. The implication: half of the organizations enabling grassroots AI innovation cannot see what is being created or assess its risk.
CEO Posture: From Headcount Cuts to Reinvestment
The EY CEO Outlook 2026 (1,200 CEOs, 21 countries, November–December 2025) tracks a rapid shift in how CEOs frame AI’s workforce impact.
In January 2025, 46% of CEOs expected AI investments to reduce headcount. By December 2025, that figure dropped to 24%. The shift happened in under twelve months.
Where is the productivity dividend going? The reinvestment priority ranking from the AI Pulse Survey: expanding existing AI capabilities (47%), developing new AI capabilities (42%), strengthening cybersecurity (41%), R&D (39%), and employee upskilling (38%). Headcount reduction ranks far lower.
Among CEOs, 54% identify generative AI as the most transformative technology. Thirty-seven percent are focusing on agentic AI — “from AI as assistant to AI as operator.” This tracks with the McKinsey agentic positioning (37% planning agentic deployment) and the BCG AI Radar finding that CEOs are moving past the experimentation phase.
Industry-Specific Depth: Banking and Wealth Management
EY-Parthenon’s sector-specific surveys add granularity absent from the broad surveys.
In banking, 77% of institutions have actively launched or soft-launched GenAI applications — up from 10% in 2023. The velocity is notable: a 7.7x increase in live deployments in two years.
In wealth and asset management, 95% of firms have scaled AI adoption to multiple use cases, and 78% are already exploring agentic AI. Financial services continues to lead enterprise AI adoption by deployment depth, not just pilot count.
Key Data Points
| Finding | Stat | Source | Date | Sample |
|---|---|---|---|---|
| AI productivity gains reported | 96% | EY AI Pulse Survey W4 | Oct 2025 | n=500 SVP+ |
| Cannot attribute gains to AI | 65% | EY AI Pulse Survey W4 | Oct 2025 | n=500 |
| Employees using AI at work | 88% | EY Work Reimagined | Aug 2025 | n=15,000 |
| Advanced AI users | 5% | EY Work Reimagined | Aug 2025 | n=15,000 |
| Sufficient AI training received | 12% | EY Work Reimagined | Aug 2025 | n=15,000 |
| Productivity gain (81+ hrs training) | 14 hrs/week | EY Work Reimagined | Aug 2025 | n=15,000 |
| Financial losses from AI risks | 99% of orgs | EY Responsible AI Survey | Sep 2025 | n=975 C-suite |
| Average AI-risk financial loss | $4.4M | EY Responsible AI Survey | Sep 2025 | n=975 |
| CEO headcount-cut expectations | 46% → 24% | EY CEO Outlook | Dec 2025 | n=1,200 |
| Banks with live GenAI apps | 77% | EY-Parthenon Banking | 2025 | sector survey |
| Shadow AI usage range | 23%–58% | EY Work Reimagined | Aug 2025 | n=15,000 |
| C-suite AI control accuracy | 12% | EY Responsible AI Survey | Sep 2025 | n=975 |
What This Means for Your Organization
Three findings from the EY corpus deserve attention from any organization past the pilot phase.
First, the attribution gap is not a measurement problem — it is a management problem. If 65% of senior leaders cannot tie AI gains to AI, the gains may be real but they are invisible to the budget process. Organizations that cannot measure AI’s contribution will underinvest in the next cycle, ceding ground to competitors who can. The fix is not better dashboards — it is defining AI-attributable metrics before deployment, not after.
Second, the training deficit is the single highest-ROI intervention available. Only 12% of employees receive adequate training. Those who do gain 14 hours per week — nearly double the median. At a fully loaded cost of $80/hour for a mid-market professional, that is $58,000 per employee per year in recaptured time. The training investment required to reach that level is a fraction of that return. If your organization has deployed AI tools without a structured training program, the gap between current performance and achievable performance is wider than most leaders assume. If that question is worth exploring for your team, I would welcome the conversation — brandon@brandonsneider.com.
Third, the responsible AI data reframes governance from cost center to profit driver. Organizations with real-time AI monitoring see 65% higher cost savings. Those without governance averaged $4.4M in AI-related losses. The business case for responsible AI is no longer abstract.
Sources
- EY US AI Pulse Survey Wave 4 (Dec 2025, n=500 SVP+). https://www.ey.com/en_us/insights/emerging-technologies/pulse-ai-survey — Credibility: MEDIUM-HIGH. Large-firm survey of senior leaders; self-reported productivity gains lack independent verification, but attribution-gap finding is methodologically sound and internally consistent across four survey waves.
- EY 2025 Work Reimagined Survey (Nov 2025, n=15,000 employees + 1,500 employers, 29 countries). https://www.ey.com/en_gl/newsroom/2025/11/ey-survey-reveals-companies-are-missing-out-on-up-to-40-percent-of-ai-productivity-gains-due-to-gaps-in-talent-strategy — Credibility: HIGH. Largest employee-side AI usage survey in the corpus; 15,000 respondents across 19 sectors provides meaningful cross-industry validity.
- EY Responsible AI Survey (Oct 2025, n=975 C-suite leaders, 21 countries). https://www.ey.com/en_gl/newsroom/2025/10/ey-survey-companies-advancing-responsible-ai-governance-linked-to-better-business-outcomes — Credibility: MEDIUM-HIGH. C-suite sample with $1B+ revenue filter means findings skew large-enterprise; the $4.4M loss figure is self-reported and may include opportunity costs. Financial-outcome correlation with governance is directional, not causal.
- EY CEO Outlook 2026 (Jan 2026, n=1,200 CEOs, 21 countries). https://www.ey.com/en_gl/newsroom/2026/01/ceos-double-down-on-ai-transformation-and-m-and-a-to-drive-growth-amid-uncertainty-in-the-global-economy — Credibility: MEDIUM-HIGH. Standard CEO survey methodology; sample skews large-enterprise.
- EY-Parthenon AI in Banking Survey (2025). https://www.ey.com/en_us/insights/banking-capital-markets/ai-in-banking-ey-parthenon-genai-survey-insights — Credibility: MEDIUM. Sector-specific survey; sample size not published.
- EY GenAI in Wealth & Asset Management Survey (2025). https://www.ey.com/en_us/insights/wealth-asset-management/gen-ai-in-wealth-asset-management-survey — Credibility: MEDIUM. Sector-specific; methodology details limited.
All sources are Tier 1 (Q4 2025) or Tier 2 (Q3 2025). No temporal caveats required.
Brandon Sneider | brandon@brandonsneider.com April 2026