BlockBeats News, June 15th - In its latest research report, UBS stated that enterprise AI adoption is facing new friction due to the rapid increase in Token and computing power costs. However, this issue is more a result of a surge in usage rather than price inflation, overall risk, or market overvaluation. The report pointed out that with the deployment of high-intensity tools such as AI encoding agents, enterprise Token consumption far exceeds expectations. This phenomenon has frequently appeared in investor discussions and has raised concerns about the potential slowdown in the adoption speed of AI technology in enterprises.

UBS found through interviews with approximately 13 enterprise IT executives that about 60% of the surveyed organizations have considered AI Token and computing power costs as a substantial issue. Particularly, after transitioning from simple chatbots to autonomous operational agents, costs have shifted from fixed SaaS expenses to variable consumption costs, significantly reducing budget predictability.

Most enterprises have already implemented or plan to introduce mitigating measures, including Token pooling, model degradation usage, waste reminders, and restrictions on heavy users, to eliminate obvious waste rather than completely curb adoption. Some executives have explicitly stated their reluctance to significantly restrict employees' use of AI, emphasizing, "Our goal is to get employees to start using AI." Therefore, they choose to optimize other budgets by cutting external IT services, integrating cloud expenses, etc., to balance the rising AI costs. The report emphasizes that almost all surveyed enterprises mentioned that AI adoption rates are accelerating, especially among developer teams. This indicates that the cost increase is mainly driven by usage growth rather than unit cost inflation.

UBS believes that this situation is part of normal cost management behavior for enterprises and is not a signal of hindered AI adoption. Even companies like Uber, which have used up their annual AI budget in one quarter, still maintain high Token limits and fully promote AI applications while offsetting costs by improving engineer efficiency.

UBS further analyzed that AI model providers and hyperscale cloud service providers are accelerating efforts to improve Token efficiency, which may limit recent price increases and have an impact on the share distribution among cloud service providers. Google Cloud and AWS are gaining advantages in cost control through in-house chips and models.

At the same time, enterprises' aversion to usage-based pricing models may increase, potentially leading to additional pressure on the non-AI software spending environment. Combining insights from a previous survey of 140 enterprise AI companies, the report pointed out that the "unclear investment return" is still the biggest adoption barrier, while "lack of budget" has not yet entered the top five. However, as the Token cost issue becomes more prominent, this dynamic is becoming a key factor for enterprises to pragmatically optimize AI deployments.