ServiceNow chief executive Bill McDermott warned this week that the rapid adoption of AI agents by companies could inflict severe damage on early-career employment, saying university graduates’ jobless rate could “easily” climb above 30 percent in the coming years. The comment, made in an interview on March 13, underlines how enterprise software vendors expect their AI tools to displace large swathes of routine work inside firms, from customer support to many white‑collar tasks.
The warning arrives against a backdrop of tangible labour‑market stress for recent graduates. The New York Federal Reserve reported that in the fourth quarter of 2025 the unemployment rate for US college graduates rose from 5.3 percent to about 5.7 percent while underemployment — graduates working in jobs that do not require a degree or are part‑time for economic reasons — climbed to 42.5 percent, the highest level since 2020. At the same time, a string of companies has cited new AI tools as a reason for cutting staff: Block eliminated nearly half its workforce, Atlassian announced roughly a 10 percent reduction to prioritise AI investments, and other CEOs have publicly signalled plans to meet revenue targets with smaller headcounts.
McDermott framed ServiceNow’s offering as emblematic of the trend: he said the company’s software can drastically reduce hiring costs and has already replaced roughly 90 percent of scenarios that previously relied on human customer‑service agents. Those tools, he argued, allow firms to boost free cash flow and revenue growth without proportionally increasing payrolls — a dynamic that accelerates when AI agents are used not only for customer service but for coding, marketing, administrative work and other traditionally white‑collar functions.
The potential disruption differs from earlier waves of automation because it extends into cognitive and creative tasks previously thought resistant to substitution. Past technological revolutions redirected labour from routine manual roles into service and knowledge sectors; today’s models threaten a broader range of office roles and the entry points that generations of graduates have relied on to begin careers and accumulate skills. Several prominent tech executives have already voiced expectations that firms can grow output with fewer employees, amplifying concerns that hiring pipelines — internships, junior developer roles, marketing assistants — could contract sharply.
If McDermott’s prediction materialises, the effects would be uneven and politically sensitive. Early‑career workers face the highest risk of scarring: prolonged unemployment or underemployment can depress lifetime earnings, slow household formation and erode trust in institutions that promise technological progress will lift living standards. For employers, the calculus is complex: AI can raise productivity and margins but also shrinks the pool of entry‑level roles that historically cultivated talent. Policymakers face choices between accelerating reskilling and apprenticeship programmes, expanding social safety nets, and reconsidering incentives that shape corporate investment in labour‑saving technologies.
The broader lesson is that enterprise AI is not only a productivity story but a social and policy one. Firms will pursue cost gains aggressively; the pace at which AI agents supplant common workplace tasks — and the uneven effects on young workers — will determine whether this wave of automation is absorbed with retraining and new opportunities, or whether it produces a cohort of lost careers requiring far more intervention than markets alone can provide. McDermott’s blunt forecast is a reminder that corporate technology choices will have consequences that echo far beyond quarterly results.