For executives who burned through their annual token budgets by the middle of the year: You’re not alone. Companies around the world that have embraced AI are discovering the same reality. As AI creates new opportunities for productivity and growth, it also introduces a new class of technology spending.

AI promises significant gains in productivity, innovation, and growth. But it also layers a new, highly variable cost base onto technology budgets that are already under pressure, often with limited visibility into where the money is going and what value it’s creating. In this environment, the goal isn’t necessarily to reduce technology spending. It’s to direct capital investment toward the AI deployments that will create the greatest business value. 

To meet that challenge, leading executives are extending FinOps beyond cloud cost management into a discipline for governing AI investment, linking consumption, technology costs, and business outcomes in near real time. That approach gives leaders the visibility and governance to invest with confidence and continuously review and shift capital toward the AI capabilities that create measurable advantage.

AI is both a cost driver and lever

AI presents a genuine paradox: It is simultaneously the largest emerging source of technology cost and the most powerful tool available to control those costs. Senior executives must deal with both realities at once. (See the Bain brief, “How CIOs Can Scale AI While Using It to Control Tech Costs.”) Left unchecked, rising demands for compute, data, software, security, and talent will erode margins; deployed deliberately, the same technology can simplify the estate, automate work, and reset productivity.

AI increases technology costs in multiple ways, from higher infrastructure spending to new demands on data, cybersecurity, and operating models.

• Higher infrastructure and run costs. AI workloads, particularly those powered by large language models, significantly increase spending on cloud, compute, and software licensing. Token costs are only one part of the cost equation (See Figure 1).
  
  Figure 1

  

• Greater architectural complexity. AI models and agents are being layered onto already fragmented technology environments, increasing integration effort, governance requirements, and operational dependencies and risk.
• Faster technology life cycles. AI platforms and models evolve in months rather than years, making technology investments obsolete more quickly and increasing the cost of keeping pace.
• Growing data and governance requirements. AI requires greater access to high-quality data, higher storage costs, and new guardrails for autonomous systems, all of which require investment before value is realized.
• Talent and operating model disruption. AI is compressing team sizes and reshaping roles. Organizations must absorb the costs of reskilling talent, redesigning processes and decision rights, and operating legacy and AI-enabled models in parallel during the transition.
• Higher cybersecurity investment. AI increases both the scale and speed of cyber threats, requiring organizations to strengthen security capabilities and increase ongoing cyber spending.

At the same time, many of these same AI capabilities can help organizations improve visibility, eliminate waste, and increase the productivity of technology investment.

• Better visibility on spending. AI can automatically classify invoices, map general ledger entries, and identify shadow IT, giving leaders near real-time visibility into technology spending. One global media company uncovered tens of millions of dollars in unmanaged technology spend across more than 80 general ledgers using this approach.
• Infrastructure optimization. AI-powered analytics can identify idle compute, overprovisioned storage, and inefficient cloud configurations, with documented cloud savings of 10% to 20%.
• Application rationalization. AI can scan application portfolios to flag overlapping functionality and underused software, helping organizations eliminate redundant tools and reduce software and maintenance costs by 10% to 30%.
• Faster software delivery. AI-enabled product development life cycle/software development life cycle and automated testing accelerate development cycles by 20% to 30% and reshape how global delivery capacity is deployed, reducing delivery costs while bringing new capabilities to market faster.
• More productive operations. AI embedded in customer service and IT operations automates routine work, predicts incidents, and improves issue resolution, delivering productivity gains as high as 40%.
• Lower managed service costs. Service providers are using AI to automate delivery, reuse assets, and shift toward platform-based operating models, improving productivity while reducing costs and accelerating time to value.

This tension will only become more pronounced over the next decade. Even organizations that effectively manage technology spending should expect costs to increase significantly, making disciplined investment decisions more important than ever. For example, in a typical $10 billion consumer products company, annual run spending is projected to grow from about $250 million today to $550 million by 2035 without effective cost management—and even well-managed organizations are likely to see costs reach about $450 million (see Figure 2).

Figure 2

In an environment where technology spending is structurally increasing, a FinOps approach can help leaders quantify the benefits of investments and channel spending toward the initiatives where it will deliver the greatest value to the business.

A FinOps approach to AI cost management

FinOps has provided a structure and discipline to help organizations bring greater transparency, accountability, and business alignment to cloud spending. Those same principles are even more critical in managing AI, where consumption-based pricing, rapidly evolving models, and uncertain economics make costs harder to predict and value harder to measure.

Extending FinOps to AI shifts the agenda from controlling spend to optimizing and governing investment. It provides a practical framework for understanding where money is going, measuring the value it creates, continuously improving efficiency, and embedding governance that keeps technology investments aligned with business priorities. The FinOps Foundation defines four domains for cloud cost management that focus on helping companies do just that.

• Visibility: Understand usage and cost. Bain’s recent Tech Maturity Assessment survey found that only 13% of tech executives believe they have adequate transparency to make sound decisions about optimizing technology and tech spending. One area where better transparency could make a significant difference is in understanding usage and its relation to vendor pricing models. Vendors price on different models (prepaid, per-seat, capacity) and meter differently by access path, so it can be difficult to tie costs to a unit of outcome. Customers need a clear view of what they have contractually bought and where that usage occurs to connect to business outcomes and make informed decisions.
  
• Benefits: Quantify business value. For each AI use case in production, document the business metric it’s meant to improve and track it over time. The metric becomes the baseline for measuring value. Forecast and budget future spending based on key cost drivers: application programming interface (API) usage, model mix, and user adoption. Connect spending to output by calculating unit economics (divide a use case’s fully loaded cost by its outcomes). Track to see whether that cost falls as volume grows. Model selection will be critical to savings: Use the least expensive model that still clears its quality bar.
• Right-size: Optimize usage and cost. Start with the infrastructure cost optimization practices carried over from cloud FinOps. Remove idle capacity, move less-used data to cheaper storage, and lock in discounts on predictable demand. Then add practices specific to AI: Choose the most cost-effective model and inference options, reduce token consumption, cache repeated prompts to avoid paying to reprocess them, and route requests to costly models only when needed.
• Op model: Manage the FinOps practice. Treat FinOps as an ongoing capability rather than a one-off effort: Build up capabilities; train the team; and keep finance, engineering, and business stakeholders engaged with monthly meetings. Then embed the governance and AgenticOps practices that help enforce investment discipline on new AI spending, reviewing cost and value regularly.

While AI-related spend will inevitably create additional cost pressure, legacy tech cost must also still be rigorously managed.

Five reasons cost programs fail

Most companies attempt to rein in tech costs, but Bain research finds that the majority fail to meet their targets and about half miss by 50% or more. Costs that are saved typically return within a year or two.

Our work with clients identifies five structural factors that contribute to these failures:

• Poor cost visibility. Most companies lack an end-to-end view of technology spending, particularly shadow IT spending that sits outside the CIO function. As noted, only 13% of executives surveyed have enough transparency to make optimization decisions with confidence.
• Wrong trade-offs. Every cost decision involves trade-offs across customer experience, execution risk, and strategic positioning. Blanket cuts frequently damage AI, data, and product engineering—the capabilities that drive growth.
• Governance mismatched to AI speed. AI capabilities evolve in weeks; enterprise budgeting runs on annual cycles. This mismatch delays value capture and allows costs to accumulate without accountability.
• Nonlinear AI economics. Token costs are volatile, concentrated among a small number of heavy users, and resistant to efficiency gains. Falling unit prices are offset by rising usage. Cost per task often stays flat.
• No ongoing discipline. Traditional budget programs lack the continuous improvement mechanisms needed to prevent costs from creeping back. Savings erode and the cycle repeats.

Five CEO actions to make lasting change

FinOps offers a proven framework and methodology that can help manage AI and technology costs more effectively. But lasting results depend on leadership. CEOs who sustain cost discipline over time are those who treat technology spending as a strategic opportunity, not an IT initiative. Embedding accountability into the way the business allocates capital and measures performance requires action on a set of fronts.

• Treat technology costs as a capital allocation decision. Technology spending is no longer an operational line item to be managed by IT. It is one of the largest and most consequential investment decisions many companies make. Boards, CEOs, and CFOs should oversee technology costs with the same rigor applied to capital allocation, requiring clear visibility into spending, AI investments, value realization, and emerging risks.
• Establish transparency before setting targets. Most organizations lack a complete view of where technology dollars are spent and what value they create. Build a comprehensive, fact-based view of technology spending across the enterprise, including shadow IT, and connect costs to business outcomes. Without transparency, cost reduction efforts often destroy value rather than create it.
• Invest where technology creates advantage. Not every technology investment deserves protection. Where differentiation is real, protect investment. Where commodity delivery is sufficient, mandate efficiency. Clear strategic intent prevents underinvestment in growth areas and overinvestment in commodity capabilities.
• Design a self-funding AI investment engine. Treat cost optimization as a source of growth capital. Capture savings from infrastructure modernization, application rationalization, and AI-enabled productivity gains, then systematically reinvest them in the highest-value AI opportunities. Scale investments only where there is a clear value thesis, rigorous business case, and end-to-end cost transparency.
• Build a capability, not a program. Technology cost reduction programs often deliver short-term gains only to see costs rebound within a few years. Break this pattern by embedding real-time cost visibility tools, agile funding models, and ongoing accountability structures into the operating model. The goal is a permanent capability, not a one-time result.

Technology has become a defining battleground for competitive advantage, and AI accelerates both the opportunities and the costs. The companies that pull ahead won’t be those that pursue the lowest tech budgets. They’ll be the ones that invest in the capabilities that matter with the discipline to measure value, optimize spending, and continuously reallocate capital as priorities evolve. FinOps provides the operating model to make that possible. In an era when every technology dollar competes for scarce capital, success will belong to organizations that spend wisely, not just sparingly.