Linux Foundation Analyst Summit: Stop Just Counting Tokens — TCA Is The New Economics Of AI
The management implication is that AI spending can no longer be governed as a narrow usage bill. If token charges represent only a fraction of enterprise AI cost, CIOs should treat AI as a cross-functional cost object spanning infrastructure, licenses, labor, energy, support, and embedded vendor services. That shifts oversight from optimization of unit price alone to a broader accountability model: which use cases justify their full operating footprint, and which do not.
This also changes the operating relationship with finance. Once CFOs want AI costs mapped to business units, products, and the general ledger, informal showback will not be enough. Leaders will need agreed allocation rules, consistent service definitions, and a decision-rights model for who approves AI experiments, who owns ongoing run costs, and who signs off on expected benefits. Without that discipline, enterprises risk enthusiastic AI adoption with weak benefits traceability and recurring month-end disputes over attribution.
Architecture choices are becoming economic control points, not just technical ones. In a multimodel environment, routing, vendor mix, and embedded AI services should be managed as portfolio levers balancing cost, performance, risk, and resilience. Standardization can improve control, but too much consolidation may reduce negotiating power or lock teams into suboptimal models. The practical next questions are straightforward:
- Do we have a usable definition of total cost of AI for investment decisions?
- Can our financial data model connect AI consumption to products, functions, or customer outcomes?
- Where should model-routing, vendor selection, and exception governance sit in the operating model?
The organizations that answer those questions first will be better positioned to move from AI experimentation to governed value realization.
At the October 8 analyst virtual summit — part of the Linux Foundation’s technology value umbrella — the Linux Foundation brought tokenomics, FinOps, information technology asset management (ITAM), and FinOps open cost and usage specification (FOCUS) together around a bigger question: How do enterprises connect what AI costs to the business value it creates? Seven themes stood out.
1. TCA May Become AI’s Successor To Total Cost Of Ownership
Perhaps the summit’s most useful new terminology is total cost of AI (TCA). The key insight is that tokens represent only 10–25% of total AI spend once the broader cost stack is considered. TCA expands the denominator to infrastructure, energy, capital, software and licensing, labor, and other costs required to produce and consume AI.
For decades, technology leaders have relied on total cost of ownership. By contrast, TCA could become the more relevant construct for AI because AI isn’t simply an asset to own but a dynamic production and consumption system where the economics span infrastructure, intelligence, people, and outcomes.
Moreover, the discussion is moving into the CFO’s office. The event organizers indicated growing CFO involvement in the Tokenomics Foundation’s leadership and an increased emphasis on translating technology consumption into outcomes that can be recognized financially. In addition, the Linux Foundation highlighted an upcoming December 8 CFO executive forum operating under the Chatham House Rule, creating an important venue for finance leaders to wrestle with the attribution, revenue, and value around TCA.
2. Tokenomics Is About Production, Consumption, And Value — Not Tokens Alone
The emerging definition of AI tokenomics stretches from converting energy and capital into AI capabilities, from consuming that intelligence to realizing measurable business value. Therefore, infrastructure, data centers, software, labor, and outcomes all sit inside that economic system.
3. AI Has A Value Problem More Than A Cost Problem
The Tokenomics Foundation’s State of Tokenomics report encompasses 472 respondents across 11 industries and $4.6 trillion in respondent revenue. More importantly, its bigger question is whether enterprises can connect AI consumption and investment to measurable business outcomes. That changes the conversation from “How cheaply can we run AI?” to “What business value did the investment produce?”
4. Multimodel AI Makes Economics An Architecture Decision
Enterprises no longer consume AI through one channel. In fact, the survey found that 96% use frontier models, 87% use cloud providers, and 64% use AI embedded within existing products, with half operating across at least four of six AI channels. As AI estates fragment, model choice, routing, architecture, performance, and economics increasingly become integral in making the right architecture decision that optimizes the initiative.
5. Model Routing Becomes A Value Lever
Model routing provides a mechanism for matching workloads with appropriate models rather than automatically sending every task to the most capable or expensive option. Accordingly, the summit materials report that 86% of companies are evaluating or using model routers. The emerging opportunity is dynamic optimization across cost, performance, risk, and value.
6. FinOps, ITFM, And Tokenomics Are Converging
AI crosses traditional technology management boundaries. Specifically, FinOps understands consumption and optimization; ITAM brings assets, contracts, licenses, and entitlements; information technology financial management (ITFM) connects technology economics with financial management; and tokenomics extends the lens into AI production, consumption, and value. As a result, the conversation broadens to technology value management rather than replacing one discipline with another.
7. FOCUS Is Emerging As AI’s Financial Data Layer
Cost and value management need normalized data. In that context, FOCUS is an open specification that normalizes billing datasets across AI, cloud, data centers, and vendors to reduce complexity and unify technology value. It provides an important foundation for connecting cost, usage, pricing, attribution, and financial accountability. The next challenge, however, is getting that data into the language that finance understands: business units, cost centers, products, allocation rules, the chart of accounts, and the general ledger. The briefing explicitly raised mapping total direct and indirect costs to the general ledger and chart of accounts to reduce manual month-end allocation.
The Bottom Line: The Destination Is AI Value Management
Ultimately, the summit’s message is bigger than tokenomics: Enterprises need to connect production to consumption to cost and finally to value. TCA provides a more complete denominator. Meanwhile, FOCUS helps standardize the underlying data. FinOps, ITAM, ITFM, and tokenomics contribute complementary disciplines. Finally, CFO engagement brings financial accountability.
Therefore, the next frontier is AI value management: connecting the TCA to the measurable value that AI creates. Forrester’s AI Value Matrix can help enterprises measure what matters.
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.