For over a decade, enterprises have relied on Financial Operations (FinOps) practices to bring discipline to cloud computing expenses. By combining finance, engineering, and business teams, FinOps has helped organizations optimize their cloud spend through reserved instances, rightsizing, and usage monitoring. However, the explosive growth of generative artificial intelligence (GenAI) is now upending established cost management frameworks and introducing new complexities that demand a paradigm shift.

According to the FinOps Foundation's 2026 State of FinOps report, an overwhelming 98% of global FinOps practitioners are now tasked with managing AI-related expenditures, a dramatic increase from just 31% in 2024. This surge reflects the rapid adoption of AI across industries, from customer service chatbots to code generation tools. As a result, AI cost management has become the most sought-after skill set for technology finance teams this year, surpassing traditional cloud optimization expertise.

The Rise of Tokenomics in Enterprise Budgeting

AI services like ChatGPT and Google Gemini use tokens as their primary billing unit. A token represents a chunk of text—roughly four characters in English—and every query consumes a specific number of tokens. This granular pricing model is forcing organizations to think about AI usage in entirely new ways. Matt Pinter, Asia-Pacific field chief technology officer at Apptio (an IBM company), notes that tokens are becoming a de facto corporate currency. Some companies now allocate a monthly token budget to developers, allowing them to use AI tools for coding, code reviews, and documentation within defined limits.

This approach, often called tokenomics, encourages engineers to optimize their queries and reduce unnecessary token consumption. Instead of generating lengthy outputs, developers learn to craft precise prompts that deliver the required results with minimal token usage. Pinter observes a cultural shift where engineers begin to ask, "How can I get my job done with the fewest tokens?" This mindset aligns with the growing trend of shifting left in FinOps—addressing cost optimization early in the software development lifecycle, before workloads reach production. By integrating cost awareness into the development process, organizations can prevent expensive AI usage patterns from propagating downstream.

The Hidden Costs of Homegrown AI Solutions

While off-the-shelf AI services offer convenience, many enterprises are building custom models tailored to their specific needs. These homegrown AI solutions require significant investments in graphics processing units (GPUs), whether deployed in on-premises data centers or rented from cloud providers. GPUs are currently in high demand, leading to long lead times and premium pricing. Beyond hardware procurement, organizations must account for power consumption, cooling, and network infrastructure to support large-scale AI workloads.

Pinter emphasizes that the total cost of ownership for AI extends far beyond compute instances. In on-premises environments, electricity costs can become a major line item, especially as AI training runs can last weeks or months. This has led to a growing intersection between FinOps and GreenOps—the practice of managing cloud costs alongside environmental sustainability. New climate regulations in the Asia-Pacific region now require companies to measure and report carbon emissions, making energy-efficient AI operations a priority. According to the FinOps Foundation, nearly half of FinOps teams are actively managing physical data center costs to capture the full footprint of AI computing demands. These teams also collaborate with environmental, social, and governance (ESG) departments on initiatives to reduce emissions.

Struggling to Quantify Return on Investment

Despite billions of dollars poured into generative AI, many organizations struggle to articulate its return on investment (ROI). Pinter notes that customers often lack a clear end state when embarking on AI projects. Without defined success metrics, it becomes difficult to justify continued spending. Only 7.5% of enterprises have baked FinOps into their AI initiatives from the start, according to IDC. This gap leaves many companies without the tools to calculate unit economics—the cost per transaction or per output generated by an AI system.

For example, a bank processing home loans could establish a baseline cost of $8 per loan for handling 1,000 loans monthly. After implementing an AI-driven automation system, the bank might see loan processing time decrease and volume increase. The ideal outcome would be a lower unit cost, say $7.20 per loan, representing a 10% reduction. This kind of granular measurement requires robust cost allocation frameworks. The Technology Business Management (TBM) model offers a structured approach, combining traditional IT financial management (ITFM) with FinOps to provide a single pane of glass for chargebacks across SaaS, on-premises, and AI services.

AI-Driven Solutions for Cost Management Ironies

Ironically, managing AI costs often requires more AI. Pinter predicts that AI-driven anomaly detection will become essential to prevent bill shocks from misconfigured cloud instances or unexpected usage spikes. Natural language chatbots could replace traditional business intelligence dashboards, allowing executives to ask questions in plain English and receive instant insights about spending patterns. These tools can automatically flag unusual token consumption, GPU idle time, or inefficiencies in model deployment.

However, technology alone cannot overcome the primary barrier to effective FinOps adoption: organizational resistance. Whether in mature cloud markets like Australia or tech hubs like Taiwan and Singapore, the biggest challenge is cultural change. Executives may not fully support cost management initiatives, and engineers may view FinOps as a bureaucratic hurdle rather than an enabler. Getting everyone on board—from the C-suite to developers—requires clear communication, training, and incentives that align cost optimization with business outcomes.

The Expanding Scope of FinOps Teams

As AI reshapes the cost landscape, FinOps teams are expanding their reach beyond public cloud. They now engage with platform engineering and enterprise architecture groups to build pricing calculators and offer pre-deployment guidance. This proactive approach helps organizations avoid costly mistakes before workloads go live. Additionally, FinOps professionals are working with ESG teams to track and reduce the carbon footprint of AI operations, linking cost savings directly to sustainability goals.

The 2026 State of FinOps report also highlights that 60% of enterprises have formal FinOps teams dedicated to cloud cost optimization. These teams are increasingly focused on AI-related activities, including monitoring GPU utilization, managing token budgets, and calculating unit economics for machine learning pipelines. As AI becomes embedded in every aspect of business, the role of FinOps will continue to evolve, requiring new skills and closer collaboration with data scientists, AI engineers, and sustainability officers.

Source: ComputerWeekly.com News