Brief IA
›

The Tokenomics Foundation Aims to Link AI Costs and Outcomes

Regulation & Ethics·Tom Levy·

The Tokenomics Foundation Aims to Link AI Costs and Outcomes

The Tokenomics Foundation Aims to Link AI Costs and Outcomes
Key Takeaways
1The Tokenomics Foundation was launched in August 2026 as a project of the Linux Foundation.
2Big-T describes the growth of token consumption alongside the number of users.
3Work on billing mainly involves an extension of the FOCUS format.
Why it matters : the Foundation aims to link AI spending to business outcomes, beyond just the price of tokens.

Launched in August 2026 as a project of the Linux Foundation, the Tokenomics Foundation aims to bridge the gap between artificial intelligence spending and business outcomes. It proposes common methods to examine usage and costs, without marketing its own measurement tools.

A request can mask multiple calls

A question directed at an application can trigger five to ten model calls, without the user seeing this background activity. The Big-T notation aims to make this multiplication identifiable by teams before receiving the bill. It describes how token consumption evolves with the number of users, based on the Big-O principle, which characterizes the growth of code execution time with data volume.

The ranking starts with T(1): a cached response is returned without invoking a model. Next is T(n), with one call per request, followed by T(n·k), when each request triggers multiple calls. At the T(n·k·a) level, agents trigger others, multiplying the calls. Finally, T(∞) denotes a loop that does not stop without external intervention.

The Foundation is working on common standards

The actual measurement falls to market tools, particularly AI gateways and observability platforms. The Foundation is working to create a common vocabulary for them. It collaborates with the FinOps Foundation, which specializes in cloud spending, and brings together banks, retailers, as well as cloud and model providers.

Its working groups disseminate open resources, independent of any vendor: definitions, methods, references, training, and certifications. The definitions particularly focus on inputs, outputs, reasoning, and cached tokens. For cost data, the work mainly involves extending FOCUS, the FinOps billing format, to standardize how AI providers report usage.

La lettre du soirGratuit · 19 h

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

The Foundation aims to transition experimentation spending to responsibly managed operations. Its goal focuses on the profitability of AI, not just on reducing the bill.

Seven layers detail cost items

Tokenomics applied to AI covers the economics of its execution and its relationship to business outcomes. It differs from the use of the same term in cryptocurrencies, where it refers to the supply and incentives of a currency. Companies pay for tokens, API calls, and GPUs; both requests and responses are broken down into tokens, each associated with a cost. For English, the rough order of magnitude is about 4 characters or 0.75 words per token.

The scope defined by the Foundation ranges from energy to commercial value, including hardware. Its 7-layer framework maps the factors that make up the cost of a response. Layer 6 examines demand: who is requesting the application, for what purpose, and with how much context. Layer 5 concerns access to models, including the gateway, routing, budgets, and safeguards.

Layer 4 focuses on the chosen model and its sizing relative to the task. Layer 3 examines inference, that is, the software serving the model, including caching and agent loops. Layer 2 looks at the portion of hardware actually mobilized for useful work.

Layer 1 covers the GPUs and data centers that convert energy into tokens. Layer 0 deals with the price and supply of electricity as well as cooling.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.