AI adoption is moving quickly. The returns are not.

PwC’s 2026 Global CEO Survey found that 56% of CEOs had seen neither increased revenue nor reduced costs from AI over the previous 12 months. Only 12% reported achieving both.

So why are some organisations getting more from AI than others?

The answer may have less to do with the AI itself and more to do with what businesses build around it.

What leading companies are doing differently

Stripe is a good example.

Its engineers use AI coding agents called Minions, which complete more than 1,000 code changes each week. But the AI isn’t left to operate unchecked. Its work passes through automated testing and human review before being approved.

Uber has taken a different approach. It built a central AI gateway that manages requests across different AI models and gives the business greater visibility over usage and cost.

DoorDash has built a similar system, with controls around authentication, usage limits and the cost of individual AI requests.

Different businesses. Different applications. But there is a common thread.

They haven’t simply given their teams access to AI and hoped for productivity gains. They have built processes around how it is used, measured and governed.

The lesson for leaders

Most businesses won’t need to build what Stripe, Uber or DoorDash have built.

But the principle applies whether you have 100 employees or 100,000.

If your team is using AI every day, can you answer:

Where is it saving time or money? Where is it improving outcomes? What is it costing? And who is responsible for making sure it delivers value?

If the answer isn’t clear, adding more AI tools is unlikely to fix the problem.

PwC found that organisations with stronger AI foundations were three times more likely to report meaningful financial returns.

That is the real takeaway.

AI isn’t the competitive advantage. Building an organisation that knows how to use it well is.

And increasingly, that comes back to leadership.