Eighty-Eight Percent of Enterprises Use AI. Only Six Percent Profit From It. The Gap Has One Explanation.

Eighty-eight percent of companies now run AI in at least one business function. Thirty-nine percent can trace any measurable financial impact to it at the enterprise level. Just six percent meet McKinsey's threshold for high performer — generating five or more percent of EBIT (operating profit, before interest and taxes) directly attributable to AI. The gap is not a technology problem.

McKinsey's State of Organizations 2026 asked thousands of business leaders what high-performing companies do differently. The answer comes back to one structural choice: they rebuilt workflows from the ground up. High performers are three times more likely than their peers to fundamentally redesign business processes around AI, rather than adopt AI tools without changing the underlying process. Among that group, 55 percent rework workflows entirely when deploying AI. Not incrementally. From scratch.

The distinction seems obvious until you're inside an organization justifying a multi-million dollar AI budget to a CFO watching quarterly margins. A legal team that adopts an AI contract review tool saves roughly two hours per lawyer per week. That's real. It shows up in satisfaction surveys. What it doesn't do is change revenue, deal close rates, or litigation outcomes — the numbers on a P&L. Changing those requires redesigning how legal work is commissioned, routed, reviewed, and billed. That's a different product.

For founders and investors, the 6% figure is a useful lens on market size. It means 94% of the enterprise has not yet rebuilt its core workflows around AI. Every legacy software product with AI bolted on — a CRM with a chatbot, a helpdesk with an AI summary button — is a target for a vertical AI company that rebuilds the same workflow with AI as the foundational design principle. Vertical AI companies, those rebuilding entire business processes for a specific industry or function, are already generating the revenue that makes the thesis concrete: Sierra reached $150 million in annual recurring revenue in eight quarters. Cognition's Devin reached $492 million in annualized revenue in twelve months. Both rebuilt from zero.

In Brazil and across Latin America, the workflow redesign opportunity carries a structural amplifier: many high-value business processes were never digitized in the first place. Payroll managed in spreadsheets, freight brokerage over WhatsApp, insurance adjustments by phone, SMB accounting in paper records. A vertical AI company targeting these segments doesn't compete with a SaaS product users are reluctant to abandon. It competes with a blank page. The first company to rebuild Brazilian freight brokerage, insurance adjustment, or SMB payroll with AI as the core design assumption isn't adding a feature. It's creating the category.

There's a window, and it doesn't stay open. As enterprise AI investment grows, the companies currently at 94% will either redesign their own workflows or acquire the companies that did it first. Enterprises that reach workflow-native AI in the next 18 to 24 months build institutional knowledge that's very hard to displace. The application layer is open right now. Open doesn't mean permanent.

What McKinsey's data doesn't tell you is how long the 94% will wait before they act. The organizational pull to keep layering is real — redesign is disruptive and layering is comfortable. That tension is where the next generation of enterprise AI companies builds its moat.