The productivity paradox has a name now, and McKinsey just said the quiet part out loud.
The Summary
- McKinsey's North America chair says AI is making individual workers more productive, but companies aren't seeing enterprise-level gains — except in software development, where output has measurably increased
- The disconnect: companies handed out ChatGPT and Claude access without redesigning workflows or identifying systemic inefficiencies
- Kutcher calls AI a potential "magic bullet" for US debt reduction if it can push GDP growth from 2.2% to 4%, though energy constraints could slash projected data center capacity by 25-30%
- The implication: giving workers AI tools without rethinking how work flows through an organization just makes people faster at doing the wrong things
The Signal
Eric Kutcher told reporters Wednesday what most companies don't want to hear: their AI investments are boosting individual output without moving the needle on organizational productivity. Workers are getting more done. Companies aren't getting more valuable work out of the system. That's not a tool problem. It's a systems problem.
The pattern is familiar to anyone who lived through previous waves of enterprise software adoption. Companies buy the shiny thing, distribute it to employees, declare victory, and wonder why the ROI looks thin. Early AI deployments followed the same script: give people access to Claude and ChatGPT, maybe tell them to use it, then wait for productivity to compound. It didn't.
"You have people that are getting more productive as individuals, but they're just changing some of the work that they're doing."
Software development is the exception. Developers are generating more code and shipping features faster because AI fits neatly into an existing workflow with clear inputs, outputs, and quality metrics. The work is structured. The feedback loops are tight. The tooling integrates with how developers already operate. Contrast that with most knowledge work, where the scope of tasks, the definition of quality, and the path from effort to business outcome remain fuzzy.
Kutcher's comments land in the context of a broader anxiety about national productivity. He called AI the only viable path to lifting US GDP growth from 2.2% to 4%, framing it as the solution to a debt burden that has topped $40 trillion. That's a lot of weight for technology to carry, especially when energy constraints could cut projected data center capacity by 25-30% just as demand is spiking.
The gap between individual and organizational productivity reveals something important about how companies think about AI:
- They're treating it as a feature add, not a workflow redesign
- They're optimizing tasks, not outcomes
- They're measuring activity, not value creation
Until companies map inefficiencies across entire systems and redesign how work flows through the organization, AI will make people faster at generating reports no one reads, drafting emails that create more work, and attending meetings that could have been resolved asynchronously. The technology works. The question is whether companies are willing to redesign themselves to take advantage of it.
The Implication
If you're leading a team, ask what enterprise-level productivity would actually look like. Not "can my people use AI tools," but "what work should we stop doing, what processes should we collapse, and what outputs should we measure differently." The companies that figure this out won't just move faster. They'll move in better directions.
For workers, the message is simpler: your productivity gains are real, even if your company hasn't figured out how to capture them yet. The skills you're building by integrating AI into your workflow will matter when you work somewhere that has redesigned itself to take advantage of them. Or when you build that place yourself.