88% of companies use AI but only 6% profit: what separates them

88% of companies now use AI. Only 6% are making money from it.

That 82-point gap, revealed in McKinsey's 2025 State of AI survey of 1,993 executives across 105 countries, isn't a rounding error. It's the largest disconnect between technology adoption and financial return since the dot-com era. And if your company falls in the 82% still burning cash on AI pilots, the window to catch up is closing faster than you think.

The $285 billion wake-up call nobody saw coming

On February 3, 2026, Anthropic released Claude Cowork: a set of open-source plugins that let AI handle entire workflows in legal, finance, and sales departments. Not assist. Handle. The market's reaction was immediate. Thomson Reuters dropped 16%. RELX fell 14%, its worst day since 1988. The Goldman Sachs U.S. software basket plunged 6% in a single session, erasing $285 billion in market value.

Why did one product launch trigger a selloff of that magnitude? Because investors suddenly realized AI wasn't coming to augment SaaS products. It was coming to replace them. The companies selling seat-based licenses for contract review, compliance checks, and financial modeling watched their entire business model get threatened by a tool that costs a fraction of the price.

The paradox Goldman Sachs can't explain

Here's where it gets uncomfortable. Goldman Sachs found "no meaningful relationship" between AI adoption and productivity at the economy-wide level. Seventy percent of S&P 500 companies talk about AI on earnings calls. Only 1% have measured its impact on earnings.

But buried in that same report: the few companies that actually measured AI's impact on specific tasks reported a median productivity gain of 30%. The gains were real. They just concentrated in two areas (customer support and software development) and in a tiny fraction of organizations willing to track them.

This is the core pattern: AI doesn't fail because the technology doesn't work. It fails because companies bolt it onto broken processes and expect magic.

What the 6% do that the other 82% don't

McKinsey's data reveals three structural patterns separating the winners.

They redesign, not digitize. 55% of AI high performers fundamentally reworked their processes when deploying AI, nearly three times the rate of everyone else. They ask "How should this function work in an AI-native world?" instead of "Where can we automate a task?"

They commit real budgets. High performers are 5x more likely to allocate over 20% of their digital budgets to AI. The majority of companies treat AI as an experiment; the 6% treat it as infrastructure.

Their leadership actually uses AI. High-performing organizations are 3x more likely to have C-suite executives who don't just approve AI budgets but actively use the tools themselves. When leadership models the behavior, adoption cascades through the organization.

The uncomfortable truth about your AI strategy

If you're reading this and thinking "we have AI projects," ask yourself three questions. Have you redesigned a single workflow end-to-end around AI capabilities, or did you just plug a chatbot into an existing process? Is your CEO using AI tools weekly, or just talking about them in quarterly calls? Can you quantify AI's impact on a specific metric with a number, not an anecdote?

Goldman's data suggests fewer than 10% of companies can answer yes to all three. The McKinsey survey confirms that being in the 88% who "use AI" means almost nothing. The only number that matters is whether you're in the 6% who've rewired their organization around it.

The $285 billion selloff wasn't a market overreaction. It was a preview. Companies that keep treating AI as a feature instead of a foundation won't just miss returns. They'll watch their entire category get replaced by someone who understood the difference.