$400 billion spent on corporate training. 74% of companies still can't keep up with skills demand.

Josh Bersin's latest research puts a hard number on a problem L&D leaders already feel: more content isn't closing the gap. The delivery model is.

Source: New Research: How AI Transforms $400 Billion Of Corporate Learning — Josh Bersin, February 11, 2026 

Here is a number worth sitting with: the global corporate training industry spends $400 billion a year.

Here is a second number: 74% of companies report they are not keeping up with skills demand.

Both figures come from the same research — Josh Bersin's February 2026 study on AI and corporate learning. They sit side by side. And together, they describe a system that is working exactly as designed, and still failing.

"L&D is being rebuilt from content delivery to skill-graph orchestration."

— Josh Bersin, joshbersin.com, February 2026

Not a budget problem

The obvious read is that companies need to spend more. That reading is wrong.

The organizations at the front of this data set aren't outspending the ones at the back. The gap isn't in budget. It's in model.

Corporate training was built for a world where skills were stable, change was gradual, and the core problem was knowledge transfer. Get the information to the right people, confirm they received it, and move on. That model — content delivery, completion tracking, course catalog maintenance — was designed for a different era of work.

That era ended. AI didn't end it; AI accelerated a transition that was already underway. Skills that took years to develop now shift in months. Capabilities that differentiated teams last year are baseline expectations this year. And the gap between what employees know and what the organization needs them to know keeps widening — not because people aren't learning, but because the delivery model can't keep pace.

What skills demand actually looks like now

Bersin's conclusion is worth quoting directly: L&D is being rebuilt from content delivery to skill-graph orchestration.

That phrase is doing real work. It describes a shift from 'do employees have access to the right content?' to 'are employees actually changing how they work?'

The two questions have entirely different answers. Most organizations can answer the first one easily: here is the catalog, here are the completions, here is the compliance rate. The second question is harder. It requires visibility into behavior, not just completion. It requires reinforcement inside real workflow contexts, not just a course made available. It requires a system designed for sustained behavior change, not point-in-time delivery.

That is the gap Bersin is describing. Not a content gap. A model gap.

The treadmill dynamic

The $400 billion / 74% pairing describes something specific: organizations spending more to stay in the same place.

If you build a content library and 74% of your organization still can't keep up with skills demand, adding more content doesn't solve it. If you run more workshops and completion rates rise but behavior change doesn't follow, more workshops won't solve it either.

The model is the constraint. Not the investment level.

The L&D teams that are closing the gap aren't the ones with the most content. They're the ones that have stopped treating learning as an event and started treating behavior change as infrastructure.

What the shift looks like in practice

Behavior-change infrastructure differs from a content delivery program in a few specific ways.

It meets people where work happens — inside the tools they already use, at the moment a decision occurs, not in a scheduled session disconnected from the actual workflow. It reinforces over time rather than delivering once and moving on. It measures behavior change and business outcomes, not completion rates. And it personalizes to role, readiness, and observed behavior — not a single learning path built for an entire org.

That is the shift Bersin describes. It is also the shift BrainStorm was built to operationalize, through the ADOPT™ framework: Awareness, Desire, Orientation, Participation, and Transformation. Each stage maps to a specific kind of intervention. Together, they describe how people move from access to habit.

The distinction matters because access and habit are not the same outcome. Organizations that measure access will always have the $400 billion problem. Organizations that build for habit will not.

The $400 billion question

The number is not the problem. The model is.

Every L&D team facing pressure to prove ROI on learning investment should ask one question before adding content or scheduling another workshop: is our current model capable of producing behavior change, or is it capable of producing completion data?

If the honest answer is the second one, that is where to start.