RESEARCH REPORT
The CIO’s guide to AI tokenomics
How to see, control and account for AI token spend at scale
10-MINUTE READ
September 10, 2026
In brief
- Four in five dollars of AI token spend lack a quantified link to business outcomes
- 80% of executives say AI creates value yet less than 20% of token spend is linked to outcomes
- CIOs face a double bind: they can’t predict consumption or show what it creates. They have a token bill no one can forecast and value no one can prove
- The CIO’s new mandate: Turn token consumption from an unpredictable expense into an investment that CEOs can quantify and boards can evaluate
AI practitioners have spent the past couple of years building the investment argument. Now they have to answer for the economics: token costs are arriving as a new and material line item with a growth rate that is outrunning the ability to justify it. Tokenomics is a boardroom concern today, as the costs of every prompt, inference and agentic interaction jump faster than a company’s ability to prove the value they create.
We surveyed 750 senior global executives across 17 countries and interviewed 15 technology and finance leaders at Fortune 500 companies. Less than one dollar in five of enterprise token spend can be traced to a quantified financial outcome—revenue influenced, cost avoided, or productivity converted into a figure finance can act on.
The concern stems from two gaps: companies cannot reliably predict token volume and spend quarter to quarter, and they cannot explain the return once the money is spent. This creates a double bind. When costs are unpredictable, finance hesitates to approve larger AI budgets. When value is invisible, boards press CEOs to explain, they turn to CFOs for control, and CFOs look to CIOs to make AI consumption predictable, economically disciplined and demonstrably valuable.
Four out of five executives say AI is delivering measurable outcomes, but fewer than one in five can connect value to token spend. Most struggle to explain such a spend in financial terms that a CFO can act on, the CEO can quantify and a board can evaluate.
What is tokenomics?
The Tokenomics Foundation defines AI tokenomics as the discipline of converting energy and capital into AI capabilities and efficiently consuming that intelligence across the organization, to realize measurable business value. Tokens are the units of data that AI models process. Every prompt, response, retrieval step and agent interaction consumes them. This report and the disciplines it outlines focus on token cost and value.
The economic question is not … how many [tokens] were consumed, but what did that token actually do?
The economic question is not … how many [tokens] were consumed, but what did that token actually do?
Field CTO for AI, Cybersecurity and Data, Global Technology Infrastructure Company
78%
The expected growth in token consumption over the next 24 months
1 in 3
organizations exhaust their token budgets before year-end
Just 35%
of companies can calculate the cost per business outcome even for their largest AI use case
The trajectory is the problem
The numbers make the double bind hard to ignore. Enterprises in our survey spent roughly $2.5 billion on AI tokens last year. Token spend now ranks third among major AI cost drivers, behind infrastructure and software development and maintenance.
The trajectory is the real concern: companies expect token consumption volume to grow 78% over the next 24 months. Even after accounting for a predicted 19% price decline, executives expect that, over the same period, aggregate token spend will approach $3.6 billion, with no optimization applied. Behind that aggregate, the range is vast: a global technology company spending north of $20 million annually, a European insurer with an AI inference budget of $150-200 million, a government contractor with a global AI spend ceiling of $250 million.
Today, token spend may be flying below the board’s radar. But the trajectory of that cost is concerning. CIOs and their finance colleagues may find that operating expense becomes a material financial risk surprisingly quickly.
In our interviews, a technology leader at a major US financial institution describes a 5-10x increase in token spend in six months. A large European bank moved from near-zero to €5 million in annual token spend in 12 months, with a single workload burning €200,000 a week.
It is a real problem … growing by a minute, by a day … keeping track of [token spend] is definitely a challenge, especially if it's distributed across multiple organizations, multiple teams.
It is a real problem … growing by a minute, by a day … keeping track of [token spend] is definitely a challenge, especially if it's distributed across multiple organizations, multiple teams.
Chief Technology Officer, Global Financial Data and Media Company
Falling AI prices will not ride to the rescue. Nine in 10 executives say cheaper tokens will lead them to expand use cases rather than cut spend. Asked what they would do if prices fell a further 25%, only one in 20 say they would bank the savings.
How executives will respond to falling token prices
A bill you can’t forecast with value you can’t prove
Enterprises can’t see token consumption at the workload level. Spend that isn’t tied to a workload can’t be predicted, and value that isn’t tied to a workload can’t be proven. Essentially, companies have an AI bill that nobody can forecast, and value that nobody can prove. This gap comes from six blind spots that are quietly breaking AI budgets.
Unknowable task cost
Tokens have no intuitive equivalent: cost emerges only after tasks run.
Frontier by default
People choose the most capable model even when lower-cost models are sufficient.
Agentic multiplier
Chained calls, expanding context and repeated processing multiply consumption.
Outdated budgets
Plans predate mainstream agentic tools and usage growth.
Fragmented pricing
Multiple providers, models and pricing structures prevent a consolidated view.
Accountability gaps
Showback dominates; limited chargeback weakens cost ownership.
What this asks of the CIO
The strategic way forward is to fix the root cause with workload-level visibility, ownership and financial accountability. And five disciplines separate the companies pulling ahead. Visibility comes first, because the other four depend on it. And there is one lever that outperforms all the others – yet less than 10% of companies are using it.
Make every workload observable before it scales
Deploy an AI gateway or observability layer that captures, for every AI interaction: the workload name, the owning team, the model used, the token cost and the output. Without this, everything else is advisory.
DATA
53%
of token usage can be tracked to a particular user or team.
Make teams pay for what they use
Bill consuming teams for their token costs. Start with engineering and the top three or four consuming functions.
Showback makes consumption visible but attaches no financial consequence. It shifts awareness, and changes behavior only when tied to other incentives. Chargeback bills the consuming team for what it uses. It is the single strongest predictor of cost optimization and value attribution in this research yet a fraction of companies use it today. Its effectiveness depends on the surrounding governance capabilities being in place.
The mechanism is direct. When a team pays for its own consumption, managers ask whether the spend is justified, and engineers start selecting models with cost in mind.
DATA
Just 7%
of companies use a chargeback model for tracking AI costs.
Send each task to the model it actually needs
Identify the approximately 10% of workloads that genuinely need frontier capability (with complex reasoning, multi-source synthesis and high-stakes generation), and route everything else to capable mid-tier or open-weight models. Enforce this routing at the gateway, not as guidance left to each developer’s judgment.
DATA
90%
The share of more than 9,000 tasks that our research found do not require frontier-model capability – a capable mid-tier model is sufficient.
Demand a value case before any workload reaches production
Define three things before you commit capital to any new AI workload: what the process costs today, in money or time; what specific financial outcome AI will improve; and how that outcome will be measured in dollars. Then hold the gate.
This does not need to be complex. But it needs to exist before the spend, not while trying to justify it afterward.
What counts as value varies by function: the deflection rate in customer operations, cycle time in software development, hours per transaction in back-office processing. The definition changes. The requirement does not.
DATA
<20%
of token spend can be attributed to a quantified financial outcome.
Build AI skills across the organization
Most routing mistakes happen because people lack clear guidance, not good judgment. Developers and employees choose the most capable model because they can’t see the cost difference – and often don’t have a simple rule for when a less expensive model is enough.
Make cost visibility available to those who want it. Set the default so that the right choice is also the easy choice, and reserve human judgment for the cases where it truly matters.
DATA
54%
of AI requests are misrouted and sent to a more capable and more expensive model than the task actually needs.
Discipline is the differentiator
AI leadership comes down to control and management discipline. The companies that win manage each token before it becomes a cost to justify, then connect it to a value case they can explain.
None of this is the CIO's job alone. It needs the CFO to make measurable value a condition of AI investment, the CHRO to build cost awareness into workforce expectations, and business leaders to give every workload a named owner from day one.
The questions are already traveling: from board to CEO, CEO to CFO, CFO to CIO. The CIOs who build workload visibility, financial accountability and routing discipline now will have an answer and the confidence to scale AI on evidence rather than hope.
About this research
This research surveyed 750 senior executives from enterprises with annual revenues exceeding $1 billion across 17 countries, coupled with 15 in-depth interviews conducted with Fortune 500 technology and finance leaders between June and July 2026.
Ajoy Menon
Digital Core Reinvention Partner Lead
Lan Guan
Chief AI and Data Officer; AI and Data Reinvention Engine Lead
James Burrows
Technology Strategy & Transformation Lead
Surya Mukherjee
Principal Director – Accenture Research
