Skip to content
AI Tokenomics Fundamentals Certification Now Available →
Insight

Observations on Tokenomics and FinOps

J.R Storment

J.R. Storment

KEY INSIGHT: FinOps asks whether an enterprise uses AI efficiently. Tokenomics asks how AI changes the business itself: its products, pricing, margins, and workforce. The two depend on each other, but are distinct.

These are my personal observations from conversations since summer, not formal guidance from any of our Foundations.

Prefer audio? Listen on Spotify, or read the full transcript.

Over the past few months I’ve sat with dozens of executives from Fortune 100–sized companies across the United States, the UK, and Europe. The conversations have been wide-ranging, and they never stop at token costs.

The Tokenomics conversation quickly moves from token spend to the total cost of AI, then to margins, products, and pricing, and almost always toward the workforce and how AI is changing the enterprise as a whole.

I’d like to share some observations—not definitions (and, yes, the em dashes are mine, not AI)—on how Tokenomics and FinOps intersect. Tokenomics is not yet a codified discipline, and I believe it shouldn’t be rushed. FinOps took time: Mike Fuller and I had each worked in FinOps for six years before we outlined the first FinOps book in 2019.

I talk through these observations in episode 20 of The Tokenomics Brief.

Quick Take

FinOps is a bottom-up practice for managing the efficiency of technology spend, AI included. The FinOps Framework lays out three domains: Understand Cost/Usage, Optimize Cost/Usage, and Quantify Business Value. In FinOps, technology value is often expressed through simplified unit economics (e.g., cost per transaction), and this work often lives in a dedicated role and team.

Tokenomics (i.e., AI Economics) tends to be broader and top-down. It is the microeconomic impact of AI on the whole company: not whether AI spend is efficient (that’s AI FinOps), but how AI changes the product, the business model, and, yes, the staffing and workforce, with value measured in CFO terms like revenue and margin.

I don’t believe that Tokenomics is a role or a team; instead, ownership is shared by the C-suite and covers the full lifecycle of turning energy and capital into business value that a CFO can accept.

Simply put, AI FinOps asks, “Are we using AI efficiently?” Tokenomics (i.e., AI Economics) asks, “How is AI changing our business, how does it make money, and what people does it need to get the work done?”

The Executive Conversation about AI Economics

In my conversations, when Tokenomics (AI Economics) comes up, the conversation follows a consistent path:

  1. Token costs first
  2. Total cost of AI (TCA): the software, compute, labor, and licenses needed to run the AI
  3. Margin management and AI’s impact on margins
  4. Product pricing and how AI is changing products
  5. Staffing and workforce changes due to AI

That last step comes up in nearly every case at the SVP, EVP, and CxO level. Technology, finance, and HR leaders in the C-suite want to know how AI affects the labor budget and how it changes the makeup of the workforce, staffing needs, and the skills required.

Product pricing is also a major consideration. That’s true for technology companies and also for financial services firms, which have to work out how to deliver customer service and what to charge when customers use agents to access their bank accounts.

I’ve been in FinOps for seven years conservatively and fifteen liberally. These conversations are very different from anything I’ve been part of in the last decade. Let me explain.

What Is Tokenomics, Really?

The term “Tokenomics” is imperfect. We didn’t coin it. It came up again and again in executive conversations this year.

When executives talk about tokenomics (i.e., AI Economics), they mean using AI effectively to change how the business runs. That means fundamental change to how work and teams are structured across the organization, not just efficient use and attribution of the spending (where FinOps tends to focus).

Tokenomics at an enterprise is about the microeconomic impact of AI on a company. Companies have to consider:

  • Their cost of capital and where they get their AI (frontier, hybrid, on-prem, edge, etc.)
  • Whether AI and the hardware behind it are available
  • Where it is generated, data sovereignty, and security
  • How their go-to-market may need to change
  • How to account for AI, including whether agentic labor can be capitalized
  • How work gets done in a human-plus-agent world

For any given company, AI could be a major enabler or a complete disruptor, in a way cloud was not for most organizations.

Tokenomics as a macroeconomic concept covers the whole economic supply chain of AI, from producing it to consuming it to the value it creates. That’s Value (with an uppercase V) the way a CFO means it: revenue, cost avoidance, labor augmentation or replacement, capacity, time to market, and more. The conversation comes from the top down, from boards, CEOs, CFOs, CHROs, CIOs, and CTOs.

Value sits at the end of a progression:

  1. Cost per token, which is the wrong metric because token counts differ between models
  2. Cost per task, the familiar unit economics
  3. Cost per outcome, where an outcome is proof of a business state change; shipped code, pull requests, and features are outputs and may deliver no business value
  4. Value: CFO-defensible impact on revenue, cost avoidance, capacity, labor, speed to market, product quality, risk reduction, and new capabilities

(side note: my colleague Shawn Alpay has proposed a AI value data schema to codify the above, check it out)

The draft Tokenomics (i.e., AI Economics) definition describes the conversion of energy and capital into AI, the efficient consumption of that intelligence, and the realization of measurable business value. The word “realization” is doing a lot of work in the definition: it refers to how AI is changing the business beyond the value derived.

For now (and this is my personal opinion), the definition does not fully capture how AI is affecting business models, pricing, staffing, and organizational structures. The definition will keep evolving as we learn more.

How is the tokenomics term imperfect? Well, for a start, Tokenomics is not about tokens. Token counts and token costs are input metrics to a bigger conversation.

What matters is the total cost of AI (TCA): infrastructure, compute, memory, storage, data preparation, security, software licenses, and the harness and context around the models. Token costs are often only 10 to 25 percent of the total. “Labor” shows up twice: once on the way in, to prepare data and build and manage the AI, and again on the way out, in changes to staffing plans as a result of AI.

Again, I don’t believe that tokenomics is a team or a role. It’s a shared responsibility running from whoever manages data center energy to whoever manages models and the harness, all the way to the global technology CFO and the CEO approving major AI investments. Ownership is shared among the CEO, CTO or CIO, CFO, and CHRO.

What Is FinOps?

The FinOps Framework has been codified for years around three primary domains:

  1. Understand cost and usage
  2. Optimize cost and usage
  3. Quantify business value

That value has mostly been about using resources efficiently to get the most from them, a technology utilization value that feeds business value.

FinOps usually quantifies value through unit economics: cost per transaction, per job, per batch, or per any other activity. FinOps unit economics tend to be built from the bottom up rather than being board-level or CFO-level metrics. When they do reach that level, they are the evidence behind the enterprise value claims that a CFO cares about, which almost always involve revenue, margins, investments, and staffing (i.e., the things that affect earnings calls).

Our first FinOps book called unit economics “the nirvana state.” In the second edition, Mike and I reframed it as a means to an end. The Quantify Business Value FinOps domain already asks whether spend helps the organization meet its strategic objectives, but it does so by enabling executive conversations rather than taking part in them. Most implementations haven’t carried that to the board level yet. When FinOps does cover data centers and labor, it has mostly reported those costs as inputs to unit economics.

“AI FinOps” (or its synonym, “FinOps for AI”) is about managing the efficiency of AI spend: understanding it, optimizing it, and quantifying its value, as the FinOps domains describe.

Some frameworks now describe AI FinOps (or FinOps for AI) as tokenomics, but I disagree with that framing wholeheartedly. Tokenomics is not an alternative to FinOps or a rebrand of it.

Those frameworks tend to focus simply on API-based offerings like Bedrock or Foundry. “Optimizing AI Usage” is FinOps work under that domain, and a FinOps practitioner may take on token management work within their existing practice.

FinOps is a cultural practice that many organizations have also turned into a team and a role. It is more important than ever: large enterprises keep hiring FinOps practitioners to manage how efficiently they use technology.

FinOps work across its core domains is an input to the broader impact of AI economics on a company, and the dependency runs both ways: tokenomics needs FinOps Open Cost and Usage Specification (FOCUS) data, allocation, and all the capabilities that FinOps teams deliver.

The Tokenomics Foundation

While the definition of tokenomics is broad (macroeconomic), the Tokenomics Foundation’s mission is narrower (microeconomic).

The Tokenomics Governing Board has not formally ratified a mission statement. It met in San Francisco in October and spent a lot of time debating ideas. The draft mission proposed, roughly, “to enable enterprises to realize measurable business value from their AI investments.”

  • “Enterprise” focus. “Enterprise” means the Fortune 500 and Global 2000, not the neoclouds, hyperscalers, and AI natives where most tokenomics “production” activity happens today. Production stays in the definition because many enterprises are increasingly running their own AI, but much of the Foundation’s work happens after the token is created.
  • Not focused on the people themselves. We don’t believe tokenomics is, or will become, a persona or a team in most organizations the way FinOps has.
  • Focused on realized business value. It explores the broader impact of AI on enterprises. That includes how organizations need to change, from how they source AI resources to how AI affects their business models, staffing, and HR.

The FinOps Foundation

The FinOps Foundation’s mission is still to “advance the people who manage the value of technology,” and that work is expanding. FinOps Scopes extend the practice to AI, SaaS, and data center spend, and FOCUS is becoming the cost and usage backbone that tokenomics relies on. The Foundation supports FinOps practitioners and people doing FinOps for AI, and it increasingly supports people doing IT asset management (ITAM), software asset management (SAM), IT financial management (ITFM), and Technology Business Management (TBM) as well.

Tokenomics doesn’t threaten FinOps; it gives practitioners a path into bigger executive discussions about AI value. It is too early to say where these disciplines land. The Foundations may stay distinct or converge, and we are not attached to any one answer.

But we are exploring all the parts and finding the gaps and overlaps between these existing disciplines and the new ones.

Let’s keep learning together while we all figure this out. Progress, not perfection.