FinOps Is Not Tokenomics: Observations on AI Efficiency and Impact

00:27Observations, Not Definitions

Hi folks, J.R. here from the Linux Foundation. There has been a lot of talk about what FinOps is and what the heck tokenomics is and how these things interplay, and also how the respective foundations here at the Linux Foundation work together. So I’m going to share some observations. I’m saying observations very intentionally because these are not meant to be definitions or working group outputs. Tokenomics is not yet a codified discipline, and we here at the foundation, and frankly the entire industry, are still working together to understand the full impact of AI on enterprises.

For perspective, it took about two and a half years post formation for the FinOps Foundation to publish its first framework.

01:14Twelve Weeks of Executive Conversations

And Mike Fuller and I, who co-authored the FinOps books together, each had about six years respectively of work in FinOps before we outlined the first books. So we are very early in the tokenomics journey, and I also understand that AI is accelerating the speed and the need for definitions, which is why we’re doing the work we’re doing now. So I’m going to share with you some observations from my last 12 weeks of travel. I’ve been on the road nonstop, 12 weeks, I was home on weekends thankfully: West Coast of the US, East Coast of the US, London, Europe, talking to a lot of massive organizations, Fortune 100 and FTSE 100 sized companies, about how AI economics and tokenomics are showing up.

The path of those conversations almost always follows, within a 15, 30, or 60 minute window, starting with talking about token costs and then quickly moving to the total cost of AI. Then getting into the software, the compute, the labor, and the licenses needed in an organization to manage and run that AI. Then conversations tend to go into margin management, which is how AI is impacting margins. Then they tend to get into product pricing and how AI is actually changing the way that businesses need to deliver products and services to their customers.

And then ultimately the end state in almost all of these conversations, nearly every time, is about staffing and how the workforce in the company needs to change. That last step shows up constantly when talking to executives in technology, finance, and now increasingly HR at the C level, who are really trying to figure out how AI is going to affect their labor budget and how it’s going to transform the makeup of their workforce and the skills that their workforce needs. Product pricing is a major consideration, and that is 100% true, yes, for technology companies today: how AI is changing the way they need to price and bundle and package their offerings.

But we’re starting to see product pricing and business model pricing conversations start to come out from financial services companies and regulated industries like insurance, who are starting to think about how they’re going to change the way they’re going to deliver not just customer service to their customers, but how they’re going to charge for usage of things like bank accounts in a world where people want their agents to be logging in and managing their finances.

03:36The Microeconomic Impact of AI

Now, I’ve been in FinOps for at least seven years, conservatively, and you could argue up to about 15, depending on what we want to call FinOps at this point. These conversations that I’m talking about in tokenomics and AI economics are very different than anything that I have been a part of in the last decade of FinOps work. So let’s talk about tokenomics for a minute, the practice and the discipline. When executives talk to me about tokenomics, or I ask them questions about it, they are talking about how they are using technology effectively to change how the business runs.

That means fundamental changes to organizational structures and product portfolios, because tokenomics, in my words, is really about the microeconomic impact of AI on an enterprise.

So in order to think about that microeconomic impact, you have to weigh some things. First, the total cost of capital and where they are actually getting their AI. Then, whether the AI that they want and the hardware needed are actually even going to be available. They need to look at where the AI is generated and security considerations around that, as well as data sovereignty pieces. They need to look at how their go-to-market may need to change and how they are accounting from a GAAP perspective for the AI, and looking at things like whether agentic labor can be capitalized.

Then they need to look at how work itself is getting done. Executives are describing a progression where AI is helping individuals do their jobs faster or better. But then they’re looking at building agents to take over entire functions. And then they’re looking at how managers are starting to supervise agents. And then finance teams are trying to understand: can agentic labor count as capitalized investment? How do we plan for that in our HR budgets? How do we start to look at organizational and HR structures that blend human and agentic headcount?

Because for any company out there right now, AI can be a major enabler or it can be a complete disruptor. And this is different than how cloud impacted companies over the last five to 10 years. Tokenomics covers the whole economic supply chain of AI, from producing it to the value it creates. And that’s value the way a CFO means value. When we talk about value, we are talking about revenue or cost avoidance. We’re talking about labor augmentation. We’re talking about time to market, capacity, and more. These are top-down AI economics and tokenomics conversations from CEOs, CFOs, and CIOs.

So the draft definition of tokenomics from the Tokenomics Foundation right now describes the conversion of energy and capital into AI, the efficient consumption of that intelligence, the AI itself, and the realization of measurable business value. AI economics is a synonym, in our opinion, of tokenomics. Realization in that definition is doing a lot of work as a word, realization.

It is pointing to the way that AI is changing how business value is not just being measured or reported on, but how it is impacting businesses, product structures, and teams. So the definition of tokenomics is going to keep changing. But right now, I’m stating a personal opinion: I don’t think the current definition does enough to capture how AI is affecting those business models, pricing, and staffing changes.

07:00An Imperfect Term

More broadly, the term tokenomics is imperfect. It’s not the best term out there, but it is a term that has caught on in a lot of circles and is on the lips of a lot of executives. FinOps was also an imperfect term that, in the early days, people really fought against, but it’s come to mean something in our organization. Just like we didn’t coin FinOps as a term, we just helped to define it, we didn’t coin tokenomics. It was coming up again and again at the beginning of this year, 2026, in conversations.

And folks on the data center side and the hardware side have been talking about this for 12 to 18 months, particularly through the lens of things like token outputs from energy or tokens per second from hardware. Beginning of this year, 2026, we started to see major GSIs publishing about the practice of tokenomics in the enterprise for CFOs and CIOs.

07:50Tokenomics Is Not About Tokens

So I always have to say, when we’re talking about tokenomics, every time I’m going to say it, you’re going to get tired of me saying it: tokenomics is not about tokens. It’s definitely not about counting tokens. And it’s really not even about token costs. What matters is the TCA, the total cost of AI. That includes infrastructure, it includes compute, memory, storage, data preparation, processes, security, software licenses, harness, and context around those models. Token costs, while important, are typically only 10 to 25% of the total cost of AI, the TCA.

Labor impacts costs on the way in, managing, preparing, getting ready for all of that AI, building it, but it also impacts staffing plans on the way out.

08:34From Cost per Token to CFO-Defensible Value

So value as a measurement sits at the long end of a progression that does start with cost per token, which I believe is actually not the right metric for most people to be tracking. It moves into cost per task, which is the familiar unit economics that much of FinOps has been focused on for a long time. It talks about cost per outcome, because outcomes are proof of a business state change. So things like shipping code, or code commits, or pull requests, or closed tickets may not actually deliver an outcome.

They might just be a task, because those things might not actually deliver business value. Because the value we want to talk about at the end of this chain is CFO-defensible value. I’m calling this uppercase V Value. It has impacts on revenue, cost avoidance, labor, margins, product quality, all of these areas. Most enterprises I talk to are not yet there. They want to get there, their CFOs care about it, but their data is not connecting those things together. I also want to state that we at the Tokenomics Foundation do not today believe that tokenomics is a team or an individual role in the way that we saw FinOps be that.

Tokenomics covers all the people who run the data centers, bring the energy in, with the heating, the cooling, and the code, and the hardware, and the conversions. It’s those who are managing the models and the harness, and it’s all the way up to the global CFOs and their CHRO and CIO and CTO partners. Let’s talk about FinOps for a minute.

10:06The FinOps Framework and Value

FinOps as a framework has been codified for years. The FinOps Framework itself, and I encourage you to load up the FinOps Framework while you’re listening to this on finops.org, hits three points. There are three domains that make up what FinOps is. The first domain is Understand Cost and Usage. The second domain is Optimize Cost and Usage. The third domain is Quantify Business Value.

That does not say realize, change, implement. It says quantify business value. It feeds executive decision making and input metrics around that value. FinOps has always had value as a part of its definition. The FinOps Foundation updated its mission to include all technology, but the focus on value did not change.

That value historically in FinOps has been about measuring, optimizing, and quantifying the use of resources efficiently to get the most out of them. That is a technology utilization value that feeds business value. FinOps has talked about unit economics for a long time. It talks about it through the lens of what are typically activity-based costings: cost per blank, cost per batch, cost per activity, even maybe cost per transaction. Those metrics tend to be built bottom-up rather than being board-level or CFO-level metrics. When they do reach the CFO level, they tend to be the evidence or the inputs behind those enterprise value claims that a CFO is making.

CFOs tend to talk about revenue, margins, investment, staffing, any of the things that are going to make their way into impacting an earnings call. The first book on FinOps talked about unit economics as the nirvana state. That second book, though, went back and reframed, in 2022, that unit economics is really just a means to an end. The Quantify Business Value domain already asks whether organizations and their spend are helping meet strategic goals, but most of the implementations of FinOps and that domain don’t reach the board level yet.

Where FinOps has covered data centers and labor, it is mostly focused on how those are cost and usage inputs to unit economics, which support the CFO-level value claims.

12:39AI FinOps and FinOps for AI

Let’s talk about a couple terms: AI FinOps and FinOps for AI. I believe those two are synonyms. They are synonyms for managing the efficiency of AI spend, to put it through the lens of the domains: understanding the cost and usage of AI spend, optimizing the cost and usage of AI spend, and quantifying its value. We are seeing some out there on the internets, on the socials, talk about tokenomics as a framework that does this. I got to say, I personally disagree with that framing. Some of this is coming from consultants, some of it’s coming from cloud providers, but almost all of it is through the lens of API token costs and delivery of tokens: the understanding, optimizing, and quantifying of tokens from an API.

That is AI FinOps. That is FinOps for AI, in my opinion. Optimizing AI usage is FinOps work. And a FinOps practitioner may be focused on optimizing token usage. You could call it TokenOps, I guess, if you wanted to, but that is not necessarily tokenomics as I am hearing it described in the space by executives in my observations over the last few months. FinOps is a cultural practice in many organizations that has turned into a team. And I want to say FinOps is more important than ever now.

I was on the East Coast two weeks ago at one of the largest financial institutions in the world, with dozens of FinOps people in their organization who are effectively and efficiently managing technology utilization to drive value, to understand it, to optimize it, and to quantify it. FinOps is an important set of work that impacts tokenomics. Tokenomics needs focused data. It needs allocation and attribution. It needs FinOps teams to operate, understand, and optimize costs. And FinOps practitioners are carrying an important part in this role. So I want to say very clearly that tokenomics is not meant to be an alternative to and certainly not a rebrand of FinOps.

14:34The Tokenomics Foundation’s Draft Mission

So let’s talk for a minute about the Tokenomics Foundation itself. While I’ve been talking about the definition of tokenomics, which is very broad, the Tokenomics Foundation mission and its focus areas are narrower. The governing board for the Tokenomics Foundation has not yet formally ratified a mission statement. It did meet in San Francisco a few weeks ago, in October, and spent a lot of time debating and proposing ideas. And there is a draft mission that we are circulating internally that focuses on enabling enterprises to realize measurable business value from AI investments.

So some things here: that is an enterprise focus. Enterprises mean the Fortune 500, the Global 2000, the FTSE folks. We’re not talking about the cloud hyperscalers and AI natives here, and that’s where a lot of the tokenomics production activity is happening today. Production of AI stays in the definition of tokenomics, because that’s what it is for many of those people. But much of the Tokenomics Foundation’s work is focused on how enterprises use, consume, and drive business value, realize business value, after the token is created. So it is an enterprise focus, not a people focus as the FinOps Foundation has been, which is focused on the people and teams doing FinOps.

The scope of the Tokenomics Foundation is exploring a broader range of impact of AI on enterprises. That includes how organizations need to change, from how they source their AI resources, even if they’re not the folks with the data centers and the energy, to how they are effectively consuming that, but critically, how that AI is changing their business model, staffing, and HR plans. The Tokenomics Foundation is looking at the efficient consumption of AI, and this is where there will be some overlaps with the work of FinOps.

16:17Connecting Cost Data to Value

While FinOps work is around the efficient understanding and optimization of usage, and there are capabilities like usage optimization, workload placement, et cetera, in FinOps, a lot of the work that sits in AI engineering teams that tokenomics touches is on model selection and prompt engineering and harness, and optimization of that harness and the context layers. Much of this is principal engineering and AI work that does not fall under an Understand or Optimize Cost and Usage domain of FinOps. The last thing I want to say about the work of the Tokenomics Foundation is that it is working on tying all of the cost and usage data, yes, from FOCUS and FinOps, all the way over into the telemetry data, the tasks, the outcomes, and the value, the CFO-defensible value that is needed in order for organizations to effectively change from AI rather than be disrupted by it.

A lot of that data right now is fragmented in a lot of different places. And one of the most defining sets of work that I think the Tokenomics Foundation will do in the coming year is the work we’re doing in this area, as recently outlined in a data schema of value data that we put out. So I encourage you to take a look at that right now on the Tokenomics website.

17:30The FinOps Foundation’s Mission

So, last thing I want to say, on the FinOps Foundation: the FinOps Foundation’s mission is unchanged since last year, when we added technology to cloud. But the mission is to advance the people, the people themselves that do this work. The work of advancing the people and the work of FinOps is expanding. FinOps scopes expanded and extended the reach of the types of technology that it is understanding, optimizing, and quantifying: to AI, to SaaS, to data center spend. And the FOCUS specification is becoming a cost and usage backbone.

Tokenomics is a larger discipline, not larger than FinOps, just saying it touches different areas. The FinOps Foundation supports practitioners doing FinOps. It supports practitioners doing FinOps for AI. It supports practitioners, through its sister organization, doing ITAM and SAM work. And increasingly, we’re seeing overlaps and supporting practitioners doing ITFM and TBM work as well. The FinOps Foundation and its sister organizations in the technology value umbrella, ITAM Forum, et cetera, exist to advance the people doing that work. So don’t worry if you’re doing that work. The tokenomics conversations are not a threat to you.

Tokenomics, as a discipline that touches into other areas, is giving practitioners across all these areas a path into executive conversations about bigger portions of AI value. And it is too early in my mind to say where these disciplines will land together. They may be distinct, they may converge. There is not yet an answer to this, and we are not attached to or driving toward any answer or predetermined outcome. But we are exploring all the parts, and we are discovering where all the gaps are and where the overlaps are between existing disciplines, and also recognizing that a lot of this stuff is new and has yet to be defined.

That’s the work we’re working on in the Tokenomics Foundation.

19:12Closing

So again, I want to say in closing, these are my personal observations from weeks and weeks, hundreds of conversations over the last few months. They are not formal guidance from either foundation. I encourage all of you to do what we said at the beginning of the FinOps Foundation, which is to keep breaking down the silos. Let’s reduce the silos between teams. But most importantly, let’s keep learning together and defining where all this works, because we want to make sure that you as individuals are growing in your careers, and we want to make sure that enterprises as well can realize measurable business value from their investments.