Where Did the Money Actually Go?

Ask any executive whether AI savings funded layoffs or people, and you get the same calm answer: reinvestment, not reduction. Ask their training budget the same question, and it tells a different story.

Picture the manager who sat through the town hall in January. Leadership announced the AI platform, the productivity numbers, the reassurance that nobody's job was the target. Six months later, her team's training line lost real money while the software line grew. Nobody announced that part out loud.

Start with the reassuring number, because it is real. EY's fourth-wave AI Pulse Survey polled 500 US senior decision-makers last fall. Only 17% of organizations seeing AI productivity gains used those gains to cut headcount. Far more redirected the money. 47% reinvested in AI capabilities. 39% put money into R&D. 38% put money into upskilling and reskilling their own people. On paper, this looks like the opposite of the layoff story. Companies choosing growth over cuts.

Now look at what happened to the training budgets themselves. A Careerminds study of 600 US HR leaders, published in July 2026, tracked the actual dollars rather than a survey opinion. The average training budget fell by $126,000 this year. The average AI or automation budget, at the same companies, rose by $88,000. Nearly half of the leaders cutting training were raising AI spend at the same time. More than half, 53.7%, said their company would accept slower employee skill development in exchange for greater AI investment.

Both numbers are true at once. Companies say they are reinvesting productivity gains in people. Companies are also trading real skill-building time for AI line items, one budget cycle at a time. The gap between what shows up in a survey answer and what shows up in a budget line is where a leader's actual priorities live. Not the ones stated in the town hall.

I call this the say/do gap. The same gap shows up when a company claims skills-based hiring and still screens resumes by pedigree. Here, it sits between the reinvestment a company announces and the training line it actually funds. It costs more than most boards expect. A second Careerminds study covered 600 HR professionals who had run AI-related layoffs. 75% of those organizations spent more on rehiring than the layoffs ever saved. Nearly a third lost critical skills and expertise outright. More than half rehired over half of the roles they had eliminated, within six months. 90% said, looking back, they would have made the decision differently. The savings on the spreadsheet and the savings that actually landed were two different numbers. The gap between them was people, found again or replaced at a premium.

Factories learned this lesson with electricity, and it took them decades. Economist Paul David studied early 20th-century manufacturing plants that bought electric motors and kept running the floor exactly as they had under steam power. New engines, wired to old layouts, for years. The equipment changed first. The productivity gains waited for the work itself to change, and the work only changed once someone invested in the people running it, not just the machine. Every AI budget line making the same trade today, capability up, training down, is choosing the wiring over the floor.

The wider question sits underneath both surveys. When AI takes over a task, the skill built around doing that task by hand starts to lose value, whether or not anyone updates it. That is the relevance cliff, showing up inside one job description, not a whole occupation. A company can answer it with severance, or with a training budget that grows to meet the moment. Right now, in a meaningful share of companies, the training line is the one getting cut while everything else scales. Someone with a name attached to that budget made a choice. It happened on purpose, quarter after quarter, long before anyone announced a layoff.

Multiply that choice across a few thousand companies and it stops being a line item. It becomes the shape of a decade. Whether the person whose task just got automated gets six months of real retraining, or a severance letter and a LinkedIn post. Families feel that difference long after the board has moved to next quarter's numbers.

Reinvestment on a slide is not the same as relearning in a paycheck.

If you sit anywhere near a budget this quarter, here is the useful question. Did your training and development line grow, hold flat, or shrink in the same year your AI budget grew? Pull both numbers. Put them side by side before the next planning meeting. If you don't hold the budget yourself, ask your manager which number moved for your team. That question costs nothing, and it tells you more than any survey will.

I'm easy to find if you want to compare notes on what you find.

Sources: EY, "AI-driven productivity is fueling reinvestment over workforce reductions," AI Pulse Survey (fourth wave), December 9, 2025 (500 US senior decision-makers, fielded September 19–October 16, 2025) · Careerminds/Innovative Human Capital, "Companies Slash Employee Training Budgets by $126K as AI Budgets Soar," July 28, 2026 (600 US HR leaders) · Careerminds, "AI-Led Layoffs: What HR Leaders Wish They Knew Before Making Job Cuts," February 2026 (600 HR professionals) · Paul A. David, "The Dynamo and the Computer," American Economic Review, 1990.