UBS says the majority of enterprise companies it's talked to recently are 'throttling AI spend' (www.businessinsider.com)

UBS analysts have reported that approximately 60% of enterprise IT executives are now implementing spending restrictions on AI, reflecting a growing concern over escalating costs related to AI token usage. These discussions, which began in early June, reveal that companies are recognizing the need for "guardrails" as AI expenses rise, with many executives citing difficulty in justifying the return on investment (ROI) from their AI initiatives. While some enterprises are significantly throttling their AI spending, others still prioritize innovation or have not yet hit their budget limits. The analysts suggest these cutbacks may pose challenges for AI model developers like OpenAI and Anthropic, who are more exposed to cost reductions, while open-sourced and Chinese models gain traction as cost-effective alternatives. Despite the spending slowdowns, UBS views this trend as a typical phase of optimization rather than an alarm for the industry. The analysts emphasize that the focus is shifting from mere deployment to efficient utilization of AI tools, indicating that organizations may prioritize optimization as a continuous engineering practice rather than a response to financial strain. This shift could pave the way for next-generation models with enhanced cost efficiency, further reshaping the AI landscape.