In the AI era, CEOs rely on a single metric to determine their future headcount

Tim Walsh is familiar with the metric that is subtly transforming the workforce perspective of corporate America. This key measure is not revenue per employee, the traditional benchmark for headcount choices for years. It also is not productivity. It is what Walsh, the chair and CEO of KPMG U.S., terms labor cost margin. Grasping this concept provides more insight into the real economic direction of AI than most discussions currently happening in boardrooms.

“For every one of my engagements,” Walsh stated, the central question becomes “what is my mix of labor? What’s my mix of technology? And what’s the overall cost of delivering that engagement?”

He explained that he anticipates the “labor cost in the mix” to decrease, while the technology costs for the same engagement will rise. “And ultimately, I will be capable of handling significantly more volume through my business in previously impossible ways.”

This rationale—reduced labor expense per work unit, increased total output, resulting in net growth—is the underlying calculation for almost every major AI investment decision in U.S. corporations today. According to the 2026 KPMG U.S. CEO Outlook Pulse Survey released Tuesday, the speed at which leaders are adopting this model is increasing much more rapidly than the public discourse about AI and jobs has recognized. Conducting business in a true economic boom is “dizzying,” he added.

They think the hype is real — just not yet

The survey, which interviewed 100 CEOs from major U.S. firms, revealed that 77% concurred that generative AI was overhyped in the past year. However, they also believed its genuine disruptive power over the next five to ten years is likely under-hyped. This nuance often gets lost in what Walsh referred to as the “noise” of the wider debate, which swings between Silicon Valley’s triumphant claims and apocalyptic forecasts of mass joblessness. The CEOs polled by KPMG mostly dismissed both extremes. Instead, they depict a more structurally profound and less predictable shift: a gradual, then abrupt, restructuring of work processes and the entities that perform them.

“There is no question that every tier of the labor force will experience disruption,” Walsh said. “But anyone claiming to know the exact outcome or its final form is not being honest, as it remains uncertain currently.”

The survey data reflects this uncertainty alongside the substantial investments being made regardless. Almost 80% of CEOs reported dedicating a minimum of 5% of their total capital budgets to AI, with 41% allocating at least 10%. Thirty-five percent are investing between 11% and 20% of their entire capital budget in the technology.

To put this in perspective, this investment level is comparable to what companies allocated to cloud infrastructure during the peak of the cloud transition—a shift that required a decade to completely transform the economy.

The jobs that are ‘a scary place to be right now’

The emerging employment landscape points to a purposeful, though uncertain, transformation. Fifty-five percent of CEOs stated that AI will cause them to increase hiring within the next year. Walsh noted that his own headcount at KPMG has not decreased, but the profile of new hires has fundamentally shifted.

“We are recruiting technologists at an unprecedented rate,” he stated. “We are hiring individuals we call orchestrators, people who manage massive sections of our workflow to ensure completeness, accuracy, and correct outcomes.” KPMG also indicated a need to hire for roles such as AI agent adoption strategists (tasked with aligning AI agents with strategy, design, and workforce planning, and promoting worker adoption), AI agent orchestration engineers (linking agents, tools, and workflows, and setting autonomy and safety parameters for agents), and AI agent operations managers (overseeing the daily performance, incidents, and modifications of agents).

This outlines the new structure of white-collar work coming into view: not widespread job loss, but layering. The most vulnerable positions, according to Walsh, are clear. “You can identify jobs involving repetitive tasks, where people perform the same actions daily. That is a precarious position to be in currently.”

However, he contended that the majority of knowledge workers do not fall into this category. For these white-collar professionals, work is not “just one thing.” “It involves building relationships. It involves developing business. It involves making judgments about which tasks to perform … Not all of that can be neatly packaged into an automated solution.”

Nevertheless, two-thirds of the CEOs surveyed confessed they have not yet formally redefined job roles or career trajectories to incorporate AI, a notable admission considering the current level of investment. The survey also found that 31% of CEOs expressed their primary concern regarding AI’s effect on leadership development is fewer chances for junior employees to develop judgment via hands-on experience. Simply put, the fear is that businesses might be cultivating a generation of managers who have never needed to solve problems independently.

The pressure to keep up

The metric Walsh monitors—labor cost margin—is fundamentally the numerical representation of these changes. It quantifies the replacement of labor with technology, the increase in capacity without a corresponding rise in staff numbers, and the productivity improvements every CEO is pressured to achieve. He confirmed this pressure is tangible, with every chief executive being closely watched and expected to improve their labor cost margin.

“It is stressful if you are not investing, if you are falling behind,” Walsh said. “Because lagging carries the risk of losing market share.”

This competitive drive—to automate more quickly than competitors, to discover efficiency gains before investors require them, to upskill employees for roles that are not fully defined—is the underlying reality of the current AI era captured by the survey. Sixty percent of CEOs identified the speed of AI innovation and its risk management as the most significant factor influencing their organization’s success over the next three years. Not trade tariffs. Not interest rates. Not global politics.

“It is dizzying,” Walsh admitted, adding that he considers CEOs in the mid-2020s to be highly resilient.

The machines are not seizing control. But the leaders of America’s biggest companies are calmly and systematically reassessing the precise number of people they require, and the resulting figure is markedly different from their original calculation.