Bain Capital is telling CEOs they're deploying AI like they're installing new software, and that's exactly why most AI projects are going nowhere.
The Summary
- Bain Capital Managing Partner David Gross says executives are treating AI as a tech rollout instead of a business model rethink
- The real failure is strategic, not technical: companies are automating old processes instead of reimagining workflows from scratch
- This explains why billions in AI spending are producing marginal gains while startups with clean slates are eating established players' lunch
The Signal
The pattern Gross is calling out is everywhere. A Fortune 500 company spends eight figures on an AI implementation that shaves 15% off a process that shouldn't exist in the first place. Meanwhile, a three-person startup builds an agent that eliminates the entire category of work.
This isn't about technology maturity. The tools work. What doesn't work is the frame. Established companies are asking "how do we use AI to do what we already do, but faster?" The right question is "if we were building this business today with AI-native assumptions, what would we build?"
The difference is profound. One path gets you incremental efficiency. The other gets you a completely different cost structure, a completely different product, and completely different competitive positioning. Gross is essentially describing the innovator's dilemma playing out in real time, except the innovation is happening inside the same technological generation.
What makes this observation valuable coming from Bain is the vantage point. Private equity sees the actual numbers. They know what works and what's theater. When a PE managing partner says the strategy is wrong, not the technology, that's data talking, not hype.
The Implication
If you're leading AI implementation at an established company, the forcing function should be: would you build this process at all if you were starting from zero today? If the answer is no, your AI project is polishing the wrong thing. The companies winning in the agent economy aren't automating legacy workflows. They're building products and services that couldn't exist without agents as the foundation. That requires killing sacred cows, not optimizing them.
Source: Bloomberg Tech