The Linux Foundation has announced the creation of the Tokenomics Foundation — a new nonprofit built to do for AI token costs what the FinOps Foundation did for cloud spend: bring discipline, transparency, and shared standards to a budget line that is becoming the largest on enterprise tech balance sheets.

The announcement lands in a week when headlines about enterprise AI costs have turned genuinely alarming. Companies like Uber blew through their entire 2026 AI coding budget before April was out. GitHub’s transition to usage-based Copilot billing on June 1 triggered immediate developer backlash after users discovered that running agentic workflows could multiply their monthly bills overnight. A separate report found enterprises are routinely running three times over their annual AI token budgets by mid-year.

The Tokenomics Foundation is being designed to address the root cause: there are no agreed-upon standards for measuring, reporting, or billing AI token consumption. Every vendor does it differently. Every CFO trying to forecast AI costs is working with incomplete, incomparable information.

What the Foundation Will Build

The Tokenomics Foundation’s work covers three areas:

Standards for token usage and billing. Right now, “tokens” mean different things across models and providers. Input tokens, output tokens, cached tokens, and tool-call tokens are counted differently depending on who you ask. The Foundation’s goal is to push for a canonical, comparable unit so that enterprises can make apples-to-apples cost comparisons across vendors.

New economic metrics. Beyond cost-per-token, the Foundation is developing metrics including “cost-per-intelligence” — what did you actually get for the spend — and “tokens-per-watt,” which connects AI economics to energy consumption. These would give procurement and engineering teams something real to benchmark.

A FinOps-style governance layer. The Foundation operates in partnership with the FinOps Foundation, extending that organization’s cloud cost discipline into AI. Companies that already have FinOps practices for AWS and Azure now need the same rigor applied to OpenAI, Anthropic, and Google AI API spending.

Supporting organizations include Accenture, Booking.com, Flexera, Google Cloud, IBM, JPMorganChase, KPMG, Microsoft, Oracle, Salesforce, SAP, and ServiceNow. The technical roadmap, initial working groups, and partnership announcements will be revealed at FinOps X in San Diego, June 8-10, 2026.

Why This Is a Bigger Deal Than It Looks

The underlying economics shifted in a way that caught most enterprises off guard. Per-token costs fell dramatically between 2023 and 2025, which led teams to assume AI costs were approaching negligible. They were not. Agentic tools, which run multiple model calls per task and chain them across workflows, drive consumption roughly 18 to 20 times higher per developer than simple chatbot interactions. Goldman Sachs research cited in the Foundation’s announcement projects that global token usage will multiply 24 times between 2026 and 2030, reaching 120 quadrillion tokens per month. The price per token is cheaper. The number of tokens being consumed is orders of magnitude higher.

Enterprises discovering mid-year that they have blown past annual AI budgets is not a technology problem. It is a governance and measurement problem — and that is exactly the gap the Tokenomics Foundation is targeting.

What This Means for Business

If you are deploying AI agents — across customer support, data analysis, internal workflows, or anywhere else — token cost management is not a future concern. It requires attention now.

A few practical steps worth taking today:

Map your current AI spend. Most businesses do not have clear visibility into which tools are calling which model APIs at what rate. Before industry standards arrive, start by documenting what you have. Which tools are running on top of foundation model APIs? At what volume?

Separate flat subscriptions from API consumption. Many AI tools charge a monthly fee but pass through API costs above a threshold. Read the fine print, or ask vendors directly how they bill for agentic workloads — multi-step tasks that chain many model calls together.

Design agents with cost awareness. Not every task needs the most capable model. Routing simpler tasks to faster, cheaper models while reserving frontier capabilities for genuinely complex reasoning can reduce token spend significantly without sacrificing output quality. This is an architectural decision, and it is worth making deliberately.

Expect pricing transparency to increase. As the Tokenomics Foundation drives toward open billing standards, AI vendor pricing structures will become more comparable. That is good for buyers long-term, but may surface cost surprises in the short term.

The businesses using AI well right now are treating it as infrastructure — with budgets, governance, and measurement practices — not as a free productivity layer added on top of existing tools. The Tokenomics Foundation is a clear signal that the rest of the industry is arriving at the same conclusion. The question is whether your business gets ahead of it or scrambles to catch up.