The 30 percent software cut we cannot source, and the renewal-book squeeze we can

The number on the card was 30 percent, and the material behind this brief does not support it. What UBS does put a figure on is narrower and more useful to anyone holding a renewal calendar: more than half of its enterprise customer conversations now include anecdotes of containing non-AI software spend, as AI infrastructure investment crowds out core software budgets ^[1]. That claim arrived attached to a downgrade. UBS moved ServiceNow to neutral from buy and cut its price target to $100 from $170 ^[2], a 41 percent reduction ^[3], having held it as the bank's only buy-rated application software stock precisely because it looked better positioned than peers for the AI era ^[4]. Analyst Karl Keirstead wrote that confidence in that view had weakened despite a material year-to-date de-rating to 15x 2026 free cash flow ^[5]. The mechanism is buy-versus-build at the edges, not replacement at the core. UBS reports no client desire to replace ServiceNow as a system of record, but some interest in using AI to custom-build workflow automation apps or to handle service management tickets in a more agentic way ^[6]. Customer support was named the function most exposed to AI-driven headcount reductions, which puts seat growth at risk in a CSM segment worth roughly 10 percent of revenues ^[7], and UBS heard no consistent buy-in for ServiceNow at the agent orchestration layer ^[8]. It trimmed its cRPO growth estimate to exit 2026 at 16 percent constant currency, down from 20 percent ^[9], a fifth off the prior figure ^[10]. Renewal books are where this lands. UBS's earlier checks with 12 Fortune 500 CIO and CTO contacts ranked spend rationalization first among the causes of software weakness, driven by over-buying, enterprise license agreement fatigue and price increases customers called excessive, and suggested the timing may be a response to aggressive vendor selling at fourth-quarter renewals ^[11]. The comparison UBS draws is to cloud: enterprises are doing to SaaS what took them four to five quarters to optimise out of AWS and Azure ^[12]. Its screen is shelfware, hard pressing on pricing and bundling, and premium price per seat, on the view that cheap seats get rationalised last ^[13]. What would move the figure is pricing architecture. Panelists at UBS's venture capital summit expect surviving vendors to abandon seat-based economics for usage, workflow or outcome-based models ^[14].