The FinOps Foundation today made available an update to a FinOps Open Cost and Usage Specification (FOCUS) for normalizing the billing of cloud services that adds a provider-agnostic method to recognize and reconcile costs.

Announced at the FinOps X conference, version 1.4 of FOCUS also adds additional data integrity tooling that enables the specification to be incorporated into a system of record.

Rob Martin, a FinOps Fellow at the FinOps Foundation, said those capabilities also lay the groundwork for a forthcoming FOCUS 1.5 update that will add provider list pricing and native token tracking capabilities to enable organizations to provide more visibility into the cost of artificial intelligence (AI) services.

Finally, the FinOps Foundation also today added a Technology Value certification that enables IT professionals to prove they have the skills and expertise needed to manage spend across multiple technology categories, along with an AI Value certification that shows how that expertise has been extended to managing the consumption of tokens.

The FinOps Foundation plans to work closely with the Tokenomics Foundation, another arm of the Linux Foundation that has been set up to specifically define a set of best practices for cost-effectively consuming tokens, says Martin.

Longer term, there will soon be a plethora of AI agents that have been trained using the data exposed by the FOCUS specification to automatically generate recommendations to optimize cloud costs. In fact, Amazon Web Services (AWS) earlier this week previewed an AI agent that has been trained to make those recommendations using data collected from multiple management tools the cloud service provider makes available. There is already an existing Model Context Protocol (MCP) server that makes it possible for an AI agent to query FinOps data, noted Martin.

Eventually, there will be a system of agents that IT teams will be able to invoke to automate FinOps workflows, with some AI agents optimized for specific cloud services while others orchestrate tasks across multiple cloud service providers, said Martin.

Each IT organization will need to determine to what degree to automate FinOps workflows, but the more cost data that can be shared with application developers and IT teams as cloud services are provisioned, the more likely they are to make better financial decisions.

Additionally, there is a massive opportunity to improve utilization rates of servers that in many cases sit idle as other workloads are deployed on additional servers. That situation can be especially egregious in environments where AI workloads have been deployed on servers configured with expensive graphics processing units (GPUs) where utilization rates never rise above the single digits.

Hopefully, the rise of the FOCUS specification coupled with AI agents will provide IT teams with better pricing leverage as they negotiate contracts with cloud service providers. In the meantime, however, there’s no substitute for instilling a little old-fashioned financial discipline within IT teams that all too often tend to view the cloud as a set of free infinite resources rather than fully appreciating how much of the IT budget they actually do consume.