- The Linux Foundation officially launched the Tokenomics Foundation on August 4, 2026, backed by roughly 30 founding members.
- Its mission is to create open, vendor-neutral standards for measuring AI costs and connecting token spend to business value.
- Initial projects include the Big-T Framework, cost-to-serve metrics, and expansion of the FOCUS billing spec with token telemetry.
As enterprise AI adoption accelerates, the gap between what companies spend on tokens and what they can actually measure in return has become a board-level concern. To address this, the Linux Foundation has taken a significant step by formally establishing the Tokenomics Foundation, a dedicated body focused on building open industry standards, benchmarks, and best practices for the economics of AI. The announcement came on August 4, 2026, with a founding roster that includes major enterprises, cloud providers, and consulting firms such as Accenture, BNY, Broadcom, Cast.ai, DoiT, Flexera, IBM, JPMorganChase, Oracle, SAP, ServiceNow, and more.
The foundation’s creation is a direct response to the rapid escalation in AI spending. Goldman Sachs projects that global token consumption will increase 24-fold by 2030, reaching 120 quadrillion tokens per month, while many organizations struggle to tie their AI budgets to tangible outcomes. The Tokenomics Foundation aims to provide a common language and measurement framework that helps both consumers and suppliers of AI understand the true total cost of ownership and the return on AI investments. With a vendor-neutral governance structure, the foundation will bring together token buyers and sellers to develop pre-competitive tools that benefit the entire industry.
Why AI Tokenomics Matters
The core challenge lies in the non-deterministic nature of token consumption. Unlike traditional cloud compute hours, tokens are not a homogeneous unit. There are input tokens, output tokens, caching tokens, and hidden reasoning tokens, each with distinct cost implications. This complexity makes it nearly impossible for CFOs to forecast AI expenses accurately. J.R. Storment, executive director of both the Tokenomics Foundation and the FinOps Foundation, pointed out that the real total cost of AI extends far beyond tokens, encompassing compute, storage, data, and human labor. The foundation’s work will therefore focus on creating a comprehensive cost model that captures the entire AI supply chain.
Several high-profile cases illustrate the urgency. Uber reportedly burned through its entire annual AI coding token budget in just a few months, while Microsoft curtailed internal use of certain coding tools due to cost overruns. These anecdotes, shared in industry media, underscore the need for better measurement and governance. The Tokenomics Foundation intends to address this by publishing reference models that break down costs in a way that CFOs can actually use, such as cost per API call rather than cost per token.
What the Foundation Will Build
The foundation’s roadmap includes several key deliverables. First, it will publish definitions for AI tokenomics, including token value and density, and establish a common vocabulary that distinguishes between input, output, reasoning, and cached tokens. Second, it is developing a full cost-of-AI reference model that places tokens in the context of nine cost layers, from compute to labor. Third, a cost-to-serve standard will express the total cost of one API call, enabling direct comparison across different model architectures and pricing structures.
Another important project is the Big-T Framework, a workload classification methodology that helps enterprises route tasks to the most cost-effective models. The framework is already in draft and available for use. Additionally, the foundation will extend the FOCUS (FinOps Open Cost and Usage Specification) to include token cost telemetry, making it easier for enterprises to normalize and compare AI costs across vendors. Education and certification programs are also on the roadmap, recognizing that AI literacy is essential for responsible scaling.
Who’s Involved and How It Works
The founding membership reflects a broad mix of AI consumers and suppliers. JPMorganChase, BNY, GoDaddy, and Lenovo represent large enterprise buyers, while infrastructure vendors like IBM, Oracle, SAP, and Broadcom bring technical expertise. Cost-optimization firms such as Cast.ai, Flexera, and DoiT add practical experience in managing cloud and AI spend. The foundation operates under the Linux Foundation’s governance model, with a Governing Board, a Technical Steering Committee, and IP-managed working groups. The board convened for the first time on July 30, and the technical committee is now being formed.
According to the foundation’s leadership, the first set of frameworks and value metric definitions will be released on a nearly monthly basis through the end of 2026. The initial in-person gathering is scheduled for Tokenomicon + FinOps X Amsterdam on September 22-23, 2026, with the flagship annual conference planned for San Diego in June 2027. This timeline reflects the urgency felt across the industry, as enterprises demand clearer ways to tie AI investments to business outcomes.
Ultimately, the Tokenomics Foundation’s success will depend on adoption by both vendors and enterprises. If the open standards gain traction, they could transform how AI is purchased and billed, much like the FOCUS spec did for cloud. With strong backing from major players and a clear roadmap, the foundation is well positioned to become the definitive reference point for AI economics. For now, organizations can start using the Big-T Framework and follow the foundation’s ongoing releases to better manage their AI spend and value.