McKinsey's 2026 State of AI: 94% of companies still can't book the AI gains their workers feel
A 1,719-executive survey finds individual productivity soaring while enterprise EBIT stays flat — and only the 6% redesigning workflows are turning AI into revenue.
McKinsey’s QuantumBlack division published its 2026 State of AI survey on August 25, and the headline number is a rebuke to two years of enterprise AI spend: only 6% of the 1,719 professionals surveyed qualify as “AI high performers,” attributing at least 5% of EBIT to AI use. That share is flat year over year. So is the 37% who attribute any EBIT impact at all.
The gap between felt productivity and booked earnings is the story. 80% of respondents who use AI in their roles said it improved their individual productivity. Almost none of that improvement reaches the income statement. McKinsey calls this the “gen AI paradox”: horizontal tools like chatbots and copilots scale fast across a workforce, then deliver diffuse gains that never consolidate into revenue or margin.
What separates the 6%? Workflow redesign. Nearly three-quarters of high performers report fundamentally redesigning workflows around AI, up from 55% in the 2025 survey. Roughly a quarter of everyone else has done the same. The revenue lifts, when they appear, cluster in marketing and sales, strategy and corporate finance, and product development. That’s not a technology gap. It’s an operating-model gap, and it favors buyers with the staff and slack to reorganize around a new production process.
Which is where the structural problem for small business becomes acute. The agent-platform economics currently minting revenue, and the SaaS-displacement dynamic in Gartner’s earlier warning, both assume a buyer with headcount to redeploy. A founder-led business of one to four people has no workflow to redesign. The founder is the workflow.
That’s the category LemonLime occupies: it studies a business’s market and prepares finished customer-growth work (prospect research, personalized outreach, content) proactively, with the owner approving and triggering every send. The AI is embedded in the output, not layered on top of the operator. See lemonlime.com for the current product.
McKinsey’s finding, read structurally, is that horizontal AI is a productivity subsidy and vertical AI is a P&L event. The 94% haven’t failed at AI. They’ve bought the wrong shape of it.
Sources
- https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- https://www.theregister.com/ai-and-ml/2026/08/25/mckinsey-says-enterprise-ai-is-finally-on-the-road-to-roi/5292388
- https://www.techtimes.com/articles/325590/20260826/record-ai-spending-cant-move-earnings-needle-94-enterprises-mckinsey-finds.htm
- https://martech.org/the-latest-ai-powered-martech-news-and-releases/
- https://martechseries.com/predictive-ai/ai-platforms-machine-learning/temporal-releases-the-2026-state-of-development-report-ai-agents-revealing-a-70-8-leap-in-ai-agent-use-among-engineers/
