As artificial intelligence technologies become ever more integrated into business operations, a new challenge has emerged: how to effectively price and manage the costs associated with AI tokenomics. Companies leveraging AI services, particularly those built on large language models (LLMs) like ChatGPT and Anthropic’s Claude, are facing unprecedented uncertainty in budgeting and billing. This uncertainty arises from the unpredictable nature of token consumption—the fundamental units of computation that power these AI systems—and the lack of established payment models that can keep pace with the rapid evolution of AI capabilities.
The Complexity Behind AI Token Pricing
At the heart of the problem lies the token itself. When a user submits a prompt to an AI model, it is converted into tokens—mathematical units that the AI processes to generate a response, which itself is returned as tokens. These tokens are the currency of AI computation, and their consumption directly translates into costs for users and service providers alike.
However, unlike traditional software licensing or cloud computing services where costs can be reasonably forecasted, token usage is inherently unpredictable. A slight change in how a prompt is phrased can lead to large variations in token consumption. Furthermore, different AI models respond differently to the same input, and agentic AI systems—where multiple AI agents collaborate to complete tasks—multiply the complexity and token demand.
This variability makes it difficult for companies to lock in prices or predict expenses over standard contract periods such as 12 or 24 months. As Simon Gooch from Saviynt notes, committing to fixed cost models is impractical when the underlying economics of token usage are still in flux.
Exploding Token Consumption and Its Impact on Budgets
Despite the cost per token falling sharply due to technological advances and competition among AI providers, the volume of tokens consumed is skyrocketing. Goldman Sachs projects a 24-fold increase in external token consumption by 2030, reaching an astonishing 120 quadrillion tokens per month. This surge is driven by businesses increasingly adopting AI agents for automation and decision-making.
Yet, many organizations remain in the dark about their actual token usage until they receive unexpectedly high bills. Even tech giants like Microsoft have had to curb internal use of third-party AI coding tools due to budget overruns. Uber reportedly exhausted its annual token budget for AI coding within just a few months, highlighting how token costs can spiral out of control without careful oversight.
Strategies Businesses Are Adopting to Manage AI Costs
In response to these challenges, companies are experimenting with various approaches to control spending and pricing. Smaller firms sometimes use personal flat-fee accounts to avoid unpredictable charges, but this practice is unlikely to be sustainable as major AI vendors tighten their policies to protect profit margins.
Experts suggest that businesses must become more strategic in how they use AI. This includes selecting the most cost-effective AI models for specific tasks and crafting more precise prompts to minimize unnecessary token consumption. Rob Steele, CFO at UK accounting software firm iplicit, likens this to giving clear instructions for a shopping list—vague or broad prompts lead to wasted resources and higher costs.
Moreover, when AI is embedded in products serving thousands of users, token costs can balloon rapidly. Companies must factor in not only development and deployment tokens but also tokens used for testing, security checks, and implementing safeguards. The ease of scaling AI agents with a simple click contrasts with the slower, more deliberate process of expanding human teams, making token cost management even more critical.
The Uncertain Future of AI Payment Models
Despite these efforts, there is no consensus yet on how to best price AI services. Options under consideration include raising overall prices, charging based on outcomes, or offering bundled pricing for specific AI-driven incidents. However, these models risk being disrupted by frequent changes in pricing strategies from AI model providers themselves, adding another layer of unpredictability for customers trying to budget for AI expenses.
Bill Peterson of Sumo Logic highlights the ongoing internal debates within companies about how to charge for agentic AI services. The dynamic nature of AI pricing, with costs changing every few months, clashes with traditional budgeting methods that favor stability and predictability.
Balancing Innovation with Financial Control
As AI continues its rapid advance, businesses face the dual challenge of harnessing its transformative potential while keeping an eye on cost control. The non-deterministic nature of AI output means that higher token consumption can yield better results, but companies must weigh these benefits against the risk of runaway expenses.
Ultimately, developing robust, transparent, and flexible payment models will be essential to sustain AI adoption across industries. Until then, organizations must navigate a landscape marked by uncertain token economics, evolving pricing strategies, and the pressing need to translate AI’s promise into measurable business value without breaking the bank.
Recommended reading
For more context, see related Peack News coverage and explainers linked below.