JPMorgan and Accenture Back a New Group to Standardize AI Token Costs

The Linux Foundation launched the Tokenomics Foundation in June 2026 to standardize AI token cost measurement, backed by JPMorgan Chase, Accenture, IBM, Microsoft, Oracle, Google Cloud, Salesforce, SAP, ServiceNow, Booking.com, KPMG, and Flexera. Accenture's internal AI platform processes 8.7 trillion tokens weekly, costing about $51.6 million at frontier API pricing, and Fortune reported that average enterprise AI budgets grew from $1.2 million in 2024 to $7 million in 2026, with only 23% of C-suite leaders seeing widespread value. The group will expand the FOCUS billing spec to cover token-based spending, aiming to help CFOs manage rising AI costs as Goldman Sachs projects token usage will multiply 24-fold by 2030.

The Linux Foundation is building a "FinOps for AI," backed by JPMorgan Chase, Accenture, Microsoft and Oracle, because most enterprises still can't tell you what a chatbot answer or an AI agent's afternoon of work actually costs. The Linux Foundation announced in June 2026 that it's launching the Tokenomics Foundation, a new open-standards body meant to measure and manage how companies spend on AI tokens. JPMorgan Chase, Accenture, IBM, Microsoft, Oracle, Google Cloud, Salesforce, SAP, ServiceNow, Booking.com, KPMG and Flexera have all signed on as initial backers, according to the Linux Foundation's own announcement. That's a serious backer list. The group's first job: expand FOCUS, the open billing specification the FinOps Foundation built for cloud costs, so it can describe token-based spending too. Here's the problem it's trying to solve. Accenture's own internal AI platform processes roughly 8.7 trillion tokens a week, an equivalent workload that would cost about $51.6 million weekly at frontier API pricing, according to Accenture. Its Chief AI and Data Officer, Lan Guan, has argued that companies are underestimating what scaling AI actually costs them. And Accenture is one of the more sophisticated operators here. Most CFOs don't have that kind of internal telemetry at all. Fortune reported on July 29 that CFOs are hitting a "cost wall" on AI spending, with many executives unable to trace where their token bills are even coming from. The average enterprise AI budget grew from about $1.2 million a year in 2024 to roughly $7 million in 2026, Fortune reported. That's not a rounding error. Some Fortune 500 firms are now seeing monthly inference bills in the tens of millions of dollars. Yet only 23% of C-suite leaders report widespread, sustained business value from AI across their organizations. You're paying more, for something you can't yet prove is working. The FinOps playbook, again Anyone who lived through the 2010s cloud migration has seen this movie before. Companies moved to AWS, Azure and Google Cloud expecting savings. It rarely worked out that way. They got blindsided by egress fees, idle instances, surprise line items and bills nobody on the finance team could parse. The FinOps Foundation formed in 2019 to fix that: a shared vocabulary and a shared spec, FOCUS, for what a cloud invoice actually means. The Tokenomics Foundation is built on that same playbook, and it's leaning on the same institution to do it. The difference is speed. Cloud spending grew steadily for years before anyone built the standard. Token consumption is compounding almost overnight. Goldman Sachs projects that global token usage will multiply 24-fold between 2026 and 2030, reaching roughly 120 quadrillion tokens a month, with the AI inference market expanding from about $106 billion in 2025 to $255 billion by 2030. That's not a gradual cost creep. That's a bill that will be unrecognizable in four years if nobody agrees on how to measure it now. What changed the math for enterprises isn't just per-token pricing coming down. It's that AI shifted from single queries to multi-step workflows, where one user action can trigger dozens of model calls behind the scenes. The price per token fell. Cheaper tokens multiplied by exploding call volume still add up to a bigger bill. That's the trap a lot of CFOs walked into this year, expecting falling API prices to translate into falling spend, and getting the opposite. Who controls the meter There's a power question buried in all this, too. Whoever controls the metric controls the upper hand at the negotiating table. If enterprises can finally benchmark token efficiency across the field, from OpenAI and Anthropic to Google and the open-weight models nipping at their heels, that's a real check on frontier labs' pricing power, the same way FOCUS let enterprises play cloud vendors off against each other instead of trusting whatever number showed up on the invoice. Smaller AI labs and infrastructure vendors that can prove genuine cost efficiency, rather than just quoting a lower headline price per million tokens, stand to gain the most once buyers have a standard way to check the math. None of that exists yet. The Tokenomics Foundation is still an intent to launch, with a Governing Board and Technical Committee to be staffed and specifications still to be written. Whether JPMorgan and Accenture actually converge on numbers everyone trusts, or just produce another dashboard nobody reads, is the real story to watch from here. Also read: White House Flip-Flops on AI Regulation as California Law Takes Effect https://startupfortune.com/white-house-flip-flops-on-ai-regulation-as-california-law-takes-effect/ • Michael Burry's Palantir Bet Is Paying Off but His Nvidia Short Isn't https://startupfortune.com/michael-burrys-palantir-bet-is-paying-off-but-his-nvidia-short-isnt/ • Wall Street Banks Are Being Asked to Bankroll Trillions in AI Data Centers https://startupfortune.com/wall-street-banks-are-being-asked-to-bankroll-trillions-in-ai-data-centers/