Telecommunications providers have spent decades managing finite resources such as spectrum, bandwidth, compute, storage and network capacity. Today, AI tokens have emerged as a new strategic resource requiring the same level of executive oversight and discipline.
As telecommunications providers scale AI across customer service, network operations, cybersecurity, fraud management and other business functions, token consumption is becoming a material driver of cost, performance, governance and business value. According to industry reporting, T-Mobile has surpassed a benchmark of processing more than 1 trillion AI tokens annually as it integrates AI more deeply into its operations, illustrating how quickly token consumption can reach telecommunications scale. Vodafone, meanwhile, reports that its AI-powered assistants now handle approximately 60 million customer conversations each month across Europe.
For CFOs and CIOs, the challenge is how to govern, monitor and optimize AI investments so that every token consumed delivers measurable business outcomes while supporting operational resilience and long-term growth.
Beyond managing costs, leaders must also prepare for a new generation of AI performance metrics, including time to first token (TTFT), token throughput and energy consumption per inference. Building these capabilities depends on more than technology investments. It requires AI observability, clear ownership, effective controls and the governance frameworks needed to manage AI at scale.
How CFOs can manage AI as a consumption-based utility
Telco CFOs are familiar with usage-based economic models. Whether they are managing cloud consumption, network transport costs or customer usage plans, success depends on understanding what is driving demand, controlling costs and ensuring that spending produces measurable business value.
AI introduces a similar challenge but at a much larger and less predictable scale.
Every customer inquiry handled by a virtual assistant, every network anomaly analyzed by an AI model, every cybersecurity workflow executed by an AI agent and every employee interaction with an AI copilot consumes tokens. Individually, the cost of a single interaction may appear insignificant. Across a large telecommunications enterprise processing millions of customer conversations, network events, security alerts and operational decisions each day, token consumption can quickly become material and increasingly difficult to forecast.
The challenge is compounded by the fact that AI costs behave differently from traditional software licensing. Instead of predictable annual licensing fees, organizations pay based on consumption. A successful AI initiative that delivers strong business outcomes can also generate significant increases in consumption as adoption scales across the enterprise.
Effective AI governance starts with asking the right questions. For telecommunications leaders, those questions must go beyond how many tokens are being used and focus on what business outcomes those tokens are enabling, where consumption is accelerating and whether the organization has the controls needed to scale AI responsibly. Here are some key questions leaders should be asking about AI consumption:
- Do we know which AI use cases consume the most tokens? Identify the applications, models and business functions driving the highest consumption and costs.
- Can we measure the cost per customer interaction? Assess whether AI-enabled customer service is delivering improved efficiency, experience and outcomes relative to its cost.
- Can we quantify the cost per network incident resolved? Determine whether AI investments are accelerating detection, diagnosis and remediation activities while lowering operational costs.
- Which AI initiatives are producing measurable business value? Track outcomes such as revenue growth, productivity gains, customer retention, reduced operating expenses and EBITDA improvement.
- Where are we generating unnecessary token consumption? Look for excessive prompt sizes, overuse of premium models, inefficient workflows, duplicate AI requests and uncontrolled agent activity.
- Can we forecast future token demand and costs? Understand how adoption, model changes and new AI initiatives may affect future spending.
- Do we have the controls needed to govern AI at scale? Ensure that policies, monitoring, approval processes, spending thresholds and risk controls are in place.
- Can we connect AI consumption directly to business outcomes? Move beyond tracking token usage to measure how AI contributes to customer experience, operational resilience and financial performance.
Industry analysts increasingly describe tokenomics as the next evolution of FinOps.
Organizations must manage token economics with the same discipline applied to cloud spending, establishing visibility into who is using AI, which models are being used, what those interactions cost and whether they are delivering business value. Without that visibility, leaders risk treating AI spending as a growing operating expense rather than a strategic investment.
For telecommunications providers, the goal should not be to minimize token consumption, but to optimize it. For example, a customer service assistant that consumes more tokens may still deliver substantial value if it reduces call-handling times, improves customer satisfaction and increases first-contact resolution rates. Similarly, an AI-enabled network operations platform may justify higher consumption if it improves outage detection, accelerates incident response or reduces operational costs.
For CFOs, the priority is to move beyond expense tracking and build an operating model that links AI consumption to value, performance and risk. Working with CIOs, they can apply financial operations discipline to gain usage visibility, assign spending accountability and establish a scalable cost model for AI and cloud consumption.
The CIO perspective: Governing AI at carrier scale
At carrier scale, even small inefficiencies can multiply quickly. Without proper oversight, organizations may face rising token consumption, inconsistent model usage, security risks, compliance challenges and limited visibility into how AI is being used across the enterprise. This is why CIOs must establish a governance framework that provides visibility, accountability and control across the AI ecosystem.
The framework should incorporate these five key priorities:
Establish an enterprise AI inventory. This inventory is crucial to help control shadow AI or unauthorized use of AI tools and provide visibility into AI assets, including:
- AI applications and copilots
- Autonomous AI agents
- Third-party AI services
- Business-owned AI solutions.
Implement AI token governance policies. This would provide clear guardrails around:
- Approved AI models
- Usage thresholds and budgets
- Monitoring and reporting requirements
- Escalation and exception procedures.
CIOs should think of tokens the same way they’ve managed bandwidth: measurable, governed and aligned to business priorities.
Optimize model selection. Not every use case requires the most advanced or expensive model. For example, customer-call summarization, knowledge search, documentation and routine reporting can often be handled by smaller, lower-cost models without sacrificing quality.
Secure AI agents as enterprise identities. AI agents need the same governance as human users. Governing these agents should include:
- Identity and access management (IAM) integration
- Least-privilege access controls
- Audit logging and monitoring
- Lifecycle management and human oversight.
Build AI observability. Leaders should be able to answer:
- Which applications consume the most tokens?
- Which business units drive AI spending?
- Are AI systems delivering expected outcomes?
- Where are risks, inefficiencies or compliance concerns emerging?
The CFO-CIO partnership: AI governance as a shared leadership responsibility
Leading telco providers understand that AI token management is both a finance and a technology challenge. They are building joint operating models that balance innovation, cost discipline, operational performance and risk management.
This includes establishing cross-functional AI governance boards that regularly review consumption trends, business outcomes, risk exposure, model performance and optimization opportunities.
AI governance is most effective when CFOs and CIOs share responsibility for cost, value, performance and risk. While CFOs focus on value realization and financial accountability, CIOs provide the governance, technology oversight and operational controls needed to scale AI responsibly. Together, they can build token governance that will enable their organizations to scale AI confidently, control cost, improve performance and convert AI experimentation into measurable enterprise value.
