The rapid ascent of Artificial Intelligence within enterprises has ushered in an era of unprecedented technological advancement, but it has also created a significant and often opaque financial challenge. As AI adoption accelerates, the costs associated with its consumption are escalating, outpacing the development of robust accountability mechanisms. In response to this growing concern, the Linux Foundation has announced its intention to launch the Tokenomics Foundation, a new initiative dedicated to establishing open standards, benchmarks, and best practices for managing the economics of AI token consumption. This announcement marks a pivotal moment in the ongoing effort to bring financial clarity and control to the burgeoning AI economy.
The formal launch of the Tokenomics Foundation is slated for FinOps X in San Diego later in June. At this event, leaders of the new foundation are expected to unveil further details regarding its technical roadmap and the formation of associated working groups. The initiative has already garnered significant initial support from a diverse array of industry leaders, including technology giants like Google, Microsoft, and IBM, as well as financial institutions such as JPMorgan Chase, and major consulting and software firms like KPMG, Oracle, and Salesforce. This broad coalition underscores the widespread recognition of the need for standardized approaches to AI cost management.
A New Unit of Enterprise Spend: Understanding AI Tokens
For those less familiar with the intricacies of AI, a "token" represents a fundamental unit of text processed by an AI model. These tokens are not merely abstract data points; they are the very currency of the AI economy, forming the bedrock of how models "think," how data centers meter usage, and ultimately, how enterprises incur costs and derive value. Every interaction with an AI, from generating text to analyzing complex datasets, is quantified and billed in terms of tokens.
"Tokens don’t behave like any cost category finance teams have dealt with before—even cloud, which took years to tame, had more predictable usage patterns," highlighted an industry expert. This statement captures the essence of the challenge. Unlike traditional cloud computing costs, which, while complex, eventually yielded to established FinOps practices and predictable scaling, token-based AI costs present a fundamentally different financial paradigm. The elasticity and often unpredictable nature of AI workloads mean that expenditures can surge rapidly, catching finance departments off guard.
The urgency of this issue was recently underscored by fintech giant Ramp, which in April announced its capability to pull token-level data directly from AI providers. This move was designed to offer finance teams much-needed visibility into the generation and allocation of AI costs. The move came in response to a mounting point of friction: AI costs, unlike many traditional software contracts, are directly tied to consumption. This consumption-driven model can lead to swift and substantial cost escalations, a stark contrast to the more predictable billing structures of legacy systems.
Ramp’s internal data has painted a concerning picture, revealing a thirteen-fold increase in average monthly token spend since January 2025. For businesses with particularly heavy AI usage, costs have seen jumps of 50% or more within a single quarter. These figures are not isolated incidents. Data published by Goldman Sachs in May corroborates this trend, projecting a staggering 24-fold increase in global token usage between 2026 and 2030, with monthly consumption anticipated to reach 120 quadrillion tokens.
This projected trajectory is already reshaping the AI market’s pricing and sales strategies. GitHub’s recent decision to transition away from its flat-rate subscription model for Copilot in favor of token-based billing serves as a clear indicator that previous economic models are becoming untenable. As AI-powered coding sessions grew longer and more resource-intensive, GitHub absorbed significant inference costs, a model that proved unsustainable. The community’s reaction to this change was swift and largely negative, with many Copilot subscribers perceiving it as a "bait-and-switch." Reports emerged of projected monthly bills skyrocketing tenfold overnight, illustrating the direct financial impact of this shift on end-users.
It is precisely this kind of anxiety and financial uncertainty that the Tokenomics Foundation aims to address. By establishing open standards, the foundation seeks to introduce order and transparency into a cost structure that is currently opaque for both buyers and sellers of AI services. J.R. Storment, executive director of the FinOps Foundation, articulated the core problem as fragmentation. "Each hyperscaler and each model provider and each hardware provider will have their own approach, their own data, their own value metrics," Storment stated in an interview with The New Stack. "We aim to align consistent models between them as we’ve done previously."
A Different Operational Muscle for a New Era of Spend
The Tokenomics Foundation is set to operate in close collaboration with the FinOps Foundation, a long-standing nonprofit under the Linux Foundation umbrella that has been instrumental in developing a shared discipline around cloud cost management since 2020. The ambition is to replicate the success of FinOps in the cloud realm by applying similar rigor and standardization to AI token spend.
However, the analogy between cloud cost management and AI token economics, while useful, has its limitations. Token economics introduces layers of complexity that cloud computing never presented. Factors such as the differential pricing of input and output tokens, the distinct billing for cached tokens, and the wide variation in pricing structures across providers make direct vendor comparisons exceptionally challenging.
Nishant Gupta, Chief Availability Officer at Salesforce, emphasized this point in a press release accompanying the announcement. Gupta argued that token economics represents a categorically more difficult problem than cloud cost management. He asserted that solving it will necessitate collective industry experimentation and knowledge pooling, rather than allowing individual companies to independently reinvent solutions. "Token economics is fundamentally more abstract and more opaque than anything we’ve managed at this scale before," Gupta stated. "It requires a different operational muscle than the one the industry built for cloud, and that muscle should evolve through broad experimentation across the industry, with the best ideas and practices contributed back so we can collectively establish durable standards around it."
Foundation in Action: Building the Frameworks
While many of the operational specifics are still under development, with further details anticipated at FinOps X on June 8, the foundational structure of the Tokenomics Foundation is taking shape. A technical committee will be responsible for developing common specifications and benchmarks for measuring and reporting token costs. This will include extending FOCUS, an open billing format already in use across cloud providers, to encompass AI token spending. A governing board will be established to set the strategic direction of the foundation and allocate resources.
The list of founding supporters reflects the diverse ecosystem of AI stakeholders. Provisional commitments have been made by major players such as Accenture, Booking.com, Flexera, Google Cloud, IBM, JPMorgan Chase, KPMG, Microsoft, Oracle, Salesforce, SAP, and ServiceNow. The extent of financial support from entities like Google and Microsoft is still under evaluation, according to a spokesperson for the foundation, indicating that the funding models are being finalized.
Who’s Absent and Why It Matters
A notable omission from the initial list of backers are the leading frontier model providers, including Anthropic and OpenAI, whose pricing structures are at the heart of the current cost crisis. This absence is significant, as enterprise budgets are already feeling the strain of these advanced model costs. A recent analysis of the AI token pricing crisis highlighted this issue, citing an instance where Uber’s CTO revealed that the company had exhausted its entire 2026 AI budget within just four months, largely due to the surging adoption of Claude Code across its engineering teams.
The Linux Foundation’s initiative seeks to alleviate this pressure by providing organizations with a consistent, vendor-neutral method for measuring AI expenditures, comparing costs across different providers, and making informed decisions about AI deployment strategies. The lack of such standardization has left many enterprises navigating a complex and costly landscape with little guidance.
The Challenge of Wildly Varying Token Pricing
The core challenge in establishing universal standards lies in the extreme variability of token pricing. Input tokens, output tokens, cached tokens, and different multiplier and structural approaches across various models and vendors create a complex web of costs. The question of how to build a common standard across such diverse offerings, and crucially, how to gain buy-in from frontier model providers who are not currently at the table, remains a significant hurdle.
J.R. Storment draws a parallel to the development of cloud billing standards. He argues that the hyperscalers themselves did not initially participate in the creation of FOCUS, yet they all adopted it once their customers demanded it. "We did this already with cloud—we put out consistent frameworks and specs for cloud billing data and now every single hyperscaler supports the standards," Storment explained. "The clouds didn’t start in the room on day one, but based on their customers being there, they all joined. We expect the same pattern here."
The significance of the Tokenomics Foundation extends beyond the technical standards themselves. The presence of major AI spenders at the table presents a unique opportunity to establish shared frameworks before the market becomes entrenched in unilaterally imposed vendor practices. "The large token consumers coming together to agree on the best approaches to maximize their consumption of the token-based services will be the fastest win in terms of frameworks, metrics, and guidance for efficient use to drive value and business outcomes," Storment concluded. This collective approach holds the promise of fostering a more predictable, manageable, and ultimately more valuable AI ecosystem for all participants.
