Context and chronology

ServiceNow’s CEO, Bill McDermott, has signalled that broad deployments of autonomous software agents could materially reduce demand for early-career hires, projecting unemployment for recent graduates could move into the mid-30s if firms hasten replacement of high-volume, repetitive tasks. That warning sits alongside concrete corporate moves: Block announced roughly 4,000 role cuts (about half its workforce) and Atlassian trimmed around 10% of staff amid a roughly 54% year‑to‑date share decline — visible signals that companies are reweighting labor vs. automation as they chase margins and cash flow.

Independent labor measures show the market is already strained: the New York Fed reports recent‑graduate unemployment near 5.7% while recent‑graduate underemployment is about 42.5%, the highest in more than a decade — a baseline that makes prospective jumps more consequential for new cohorts. Vendor-level claims are also stark: ServiceNow says it has eliminated roughly 90% of prior human-driven customer-service workflows through software agent replacements, illustrating the volume of routinized tasks on the chopping block.

Complementary research paints both grimmer and more structural scenarios. A Citrini Research modelling exercise maps how rapid, economy‑wide adoption of autonomous agents could produce macro feedback loops: substitution of contractors and third‑party services with in‑house agents would lower payroll but also erode spending that sustains other firms, producing outcomes such as a near‑term doubling of unemployment in its central scenarios and a greater‑than‑33% drop in aggregate equity value within roughly 24 months in severe runs. That modelling is directional and disputed on timing and magnitude, but it highlights a pathway by which firm‑level optimization cascades into demand shocks.

Technical and market constraints complicate a neat, immediate substitution story. Experts warn that high‑quality data, integration work, governance, SLAs, legal liabilities and customer trust limit how fast unsupervised agents can fully replace human roles; semi‑supervised deployments — however — can still scale substitution fast enough to significantly affect entry-level hiring over quarters rather than decades. Supply‑side concentration further amplifies risk: industry estimates place global AI infrastructure spending near $1.5 trillion in 2025, channelled through a small set of hyperscalers, which raises vendor lock‑in and speeds adoption among firms that can afford the stacks while leaving smaller employers lagging.

Policy and institutional responses are central to whether displacement becomes a short, sharp shock or a longer transitional challenge. Thought leaders including AI executives have proposed pairing demand‑side supports (retraining, apprenticeships, targeted wage or stipend pilots) with supply‑side remedies (public investment in portable, auditable infrastructure, competition policy to limit lock‑in, and standards for interoperability). Without coordinated measures, universities, staffing firms and public training programs face a near‑term squeeze: fewer paid internships and entry roles reduce on‑ramps into work, depressing practical experience for graduates and potentially lowering the value proposition of a degree.

For corporate leaders and investors the immediate task is diagnostic: map where internal automation will hollow out external revenues, stress‑test revenue against contractor spending declines, and build transition supports to preserve talent pipelines. For campus recruiters and higher education, the tactical moves are to expand paid, assessed work experiences and apprenticeship‑style partnerships. For policymakers the options include targeted fiscal measures in disruption zones, funding for rapid retraining, and rules to limit infrastructural concentration that accelerates displacement.

In short, McDermott’s warning is a high‑profile signal embedded in a broader set of analyses that range from cautious to alarmed: while the exact path and magnitude are contested, the combined corporate signals, empirical labor indicators, and systemic scenario work create a plausible downside risk that demands coordinated private and public responses to avoid persistent scarring for early‑career cohorts.