Global – – October 15, 2026 -- Only 17% of health and life sciences leaders report a centralized approach to internal data governance, according to 2026 Voice of Our Clients (VOC) research, even as organizations move generative AI from pilots into production.
Data governance gaps widen across partner ecosystems
The VOC research found that 54% of the health and life sciences business and IT leaders interviewed have a holistic data strategy across their internal enterprise. Across external ecosystems of partners and suppliers, the share with a holistic strategy falls to 32%, and the share with centralized data governance falls to 14%.
Enterprise AI depends on trusted data and governance that extends beyond individual experiments, so these gaps limit how far AI programs can scale.
Token consumption makes GenAI costs harder to forecast than traditional IT
GenAI consumption varies with how models are used, which models are selected and how agents and applications are designed. Tokens, the units many GenAI models process, are one component. Individually their cost may appear small, but at enterprise scale leaders need to know where consumption occurs and whether it creates enough value.
Costs can also escape scrutiny during development and testing. Without governance, organizations struggle to allocate spending to the right solutions and products and to assign accountability.
Cloud FinOps offers the template for AI cost control
Cloud adoption followed a similar path. Many organizations saw costs grow rapidly without the visibility to manage them, and cloud FinOps emerged in response. The guidance is to set financial and governance disciplines for AI before adoption outpaces cost control, and to build them in before pilots start.
Architecture matters as well. Cloud designs that do not support cost management, limited use of autoscaling and automatic shutdown, and on-premises practices applied to cloud environments can weaken control.
Three actions target ownership, outcomes and continuous review
The recommended steps are to define ownership and cost allocation for each AI resource and use case before scaling, to measure consumption alongside business outcomes, and to review costs, architecture and consumption patterns on an ongoing basis. A lower AI bill is not necessarily a better result if it reduces the application's value.
Rather than creating new governance layers, organizations can embed AI decision rights in existing project management offices and enterprise architecture boards.