Speridian Technologies Unveils FinOps Framework to Curb AI Spending

The firm, based in Albuquerque, aims to bridge the gap between engineering departments and finance teams, who frequently lack visibility into how AI consumption translates to the bottom line. According to Sourav Roy, vice president at Speridian, the rapid, exponential growth of token usage often leaves organizations with massive, opaque bills. The new framework targets four specific cost drivers: input and output token ratios, modality premiums, model tier taxes, and context window creep.

Speridian’s methodology focuses on three operational layers—design-time optimization, run-time optimization, and governance—to ensure that AI spend remains tied to measurable outcomes. The engagement process begins with a baseline assessment to identify waste, moves into technical implementation such as semantic caching and model routing, and concludes with the establishment of long-term governance policies. CEO Ali Hasan emphasizes that the initiative is built on the principle that organizations cannot improve what they fail to track, noting that government agencies and private enterprises alike require this structure to scale their initiatives without losing control of their budgets.