The Spend Is Real. The Returns Aren't.
Executives surveyed by KPMG in early 2026 planned to invest an average of $186 million in AI over the next 12 months. That number is going up. And yet, BCG's 2025 research found only 5% of companies were achieving AI value at scale, while 60% saw no material value at all. Not 50%. Not 30%. Five percent. MIT's GenAI Divide report found that about 95% of corporate generative AI pilots delivered no meaningful financial return. That's a staggering gap between investment and return. And it's not because the technology doesn't work. The technology works. The problem is almost always the same thing: the people using the technology weren't prepared to use it well.
What the 5% Are Doing Differently
When you look at the organizations in that top 5% — the ones actually getting measurable returns from AI — a few things consistently separate them from everyone else. They treat training as infrastructure, not a perk. They don't buy AI licenses and hope adoption follows. They invest in structured, hands-on learning for the people who are supposed to use these tools every day. Not a lunch-and-learn. Not a vendor webinar. Real training, delivered by people who actually use the tools professionally. They identify champions early. The most effective enterprise AI rollouts we've seen start with a small group of motivated people who get trained deeply — then become internal advocates who pull the rest of the organization forward. You don't need everyone fluent on day one. You need ten people who are genuinely good. They measure what actually matters. Most organizations track AI adoption by license utilization. The 5% track business outcomes — deal cycle time, content production speed, code review throughput, call prep time. When you measure the right things, you can see what's working and fix what isn't. They don't confuse access with capability. Giving someone a Claude or Copilot license is not the same as giving them the ability to use it well. Proficiency takes practice. It takes real exercises on real workflows. It takes feedback. BCG found that 79% of employees with more than five hours of AI training became regular users, versus 67% of those with less. That gap doesn't close on its own.
The Training Problem Is Worse Than It Looks
Here's what makes this particularly frustrating: most companies think they're solving it. DataCamp's 2026 State of Data & AI Literacy Report found that 82% of enterprise leaders say their organization provides some form of AI training. Yet 59% of those same leaders still report a significant AI skills gap. Training is happening — it's just not working. The reason is almost always delivery. Generic e-learning modules. Passive webinars. Content that has nothing to do with the actual tools your team uses or the actual work they do. That kind of training doesn't change behavior. It checks a box. What changes behavior is hands-on practice with the specific tools your team uses, guided by someone who uses those tools to deliver real work for real clients. That's a different experience entirely.
What This Means for Your Organization
If you're in that 95% — and statistically, you probably are — the question isn't whether to invest more in AI. You're already investing. The question is whether the people on your team are actually equipped to use what you've paid for. We work with enterprise teams to build the training programs that answer that question — built around your tools, your stack, your workflows. Not off-the-shelf content delivered to a passive audience. Custom programs that actually change how your team works. If you want to talk through what that looks like for your organization, the conversation costs nothing. For a practical guide to measuring whether your AI training is actually working, see our post on corporate AI training ROI.
Ready to build something that actually delivers ROI? Start a conversation →