Companies launch 300 AI projects, complete only 3

What happened

Business leaders and technologists gathered at Fortune's Brainstorm Tech conference acknowledged a widespread problem—teams across companies are starting numerous AI pilots and use cases, but many are never completed or serve no clear strategic purpose. Executives described situations where employees identify 300 potential AI applications when only about three would actually move the business needle.

Why it matters

Even though AI adoption is meant to create efficiency, the proliferation of incomplete or unfocused projects can slow companies down rather than speed them up. Each pilot requires a dedicated team, KPIs, and engineering resources, making it harder to identify which AI initiatives actually deliver business value and allowing poorly performing projects to persist. The real challenge, panelists noted, is that business leaders often focus too heavily on the technical aspects of AI rather than whether projects align with overall strategy.

What to watch

Leaders stressed that successful AI transformation must be business-led and driven from the top—involving the board, CEO, and entire executive team—rather than driven solely by technical teams. One executive observed that completing a single AI project fully and iterating on it tightly delivers far more value than advancing multiple projects to partial completion, since prototyping, documentation, and code generation have become so inexpensive that it is easy to start work without finishing it.