🎙️ The Long Table

<!-- Episode card -->
<article class="lt-card">
  <div class="lt-card__top">
  <div class="lt-card__thumb">
    <img src="https://media.doit.com/long-table/1d53f70a86f9-ep01-thumb-1080x1080-2x.webp" alt="JR Storment on The Long Table" loading="lazy">
  </div>
  <div class="lt-card__body">
    <span class="lt-card__number">Episode 01</span>
    <h3 class="lt-card__title">JR Storment on FinOps, Tokenomics, and What AI Changes Next</h3>
    <p class="lt-card__desc">JR Storment joins the inaugural episode for a conversation about the evolution of FinOps, the emergence of Tokenomics, and what AI is forcing leaders to confront about consumption, value, and accountability.</p>
    <div class="lt-card__meta">
      <span class="lt-card__date">Aug 2026</span>
      <span class="lt-card__dot" aria-hidden="true">&middot;</span>
      <span class="lt-card__duration">52 min</span>
    </div>
    <div class="lt-card__platforms">
      <a href="https://www.youtube.com/@TheLongTable_DoiT" class="lt-platform" aria-label="Watch on YouTube">YouTube</a>
      <a href="https://open.spotify.com/show/7Mt563pLWYsuYN92uFSUQ8" class="lt-platform" aria-label="Listen on Spotify">Spotify</a>
      <a href="https://podcasts.apple.com/us/podcast/the-long-table/id6807056597" class="lt-platform" aria-label="Listen on Apple Podcasts">Apple Podcasts</a>
    </div>
  </div>
</div>

  <details class="lt-card__transcript">
<summary class="lt-transcript__summary">
  <span class="lt-transcript__label">Read the full transcript</span>
  <span class="lt-transcript__chevron" aria-hidden="true">&#9660;</span>
</summary>

<div class="lt-transcript__body">
  <nav class="lt-transcript__toc" aria-label="Transcript chapters">
    <p class="lt-transcript__toclabel">Jump to a section</p>
    <ol>
      <li><a href="#cold-open">Cold open</a> <time>0:00</time></li>
      <li><a href="#welcome">Welcome to the show</a> <time>0:19</time></li>
      <li><a href="#finops-to-ai-reset">From FinOps at Citi to the AI reset</a> <time>0:49</time></li>
      <li><a href="#boards-ask-about-cost">When boards start asking about AI cost</a> <time>3:08</time></li>
      <li><a href="#token-maxing-rise-fall">The rise and fall of token maxing</a> <time>5:30</time></li>
      <li><a href="#foundation-origin">How the Tokenomics Foundation began</a> <time>8:58</time></li>
      <li><a href="#not-about-tokens">Tokenomics is not about tokens</a> <time>13:39</time></li>
      <li><a href="#gas-price-problem">The gas price problem with token costs</a> <time>18:48</time></li>
      <li><a href="#where-finops-ends">Where FinOps ends and tokenomics begins</a> <time>23:54</time></li>
      <li><a href="#model-selection">Model selection and the harness around it</a> <time>36:25</time></li>
      <li><a href="#energy-scarce-resource">Energy as the new scarce resource</a> <time>40:37</time></li>
      <li><a href="#where-it-lives">Where should tokenomics live in the org?</a> <time>43:06</time></li>
      <li><a href="#blind-men-elephant">The blind men and the elephant</a> <time>46:04</time></li>
      <li><a href="#saas-pricing-agents">SaaS pricing in the age of agents</a> <time>52:04</time></li>
      <li><a href="#shoulders-of-giants">Standing on the shoulders of giants</a> <time>59:35</time></li>
      <li><a href="#which-bucket-matters">Which bucket matters most: production, consumption, or value</a> <time>1:05:56</time></li>
      <li><a href="#shift-left-shift-up">Shift left, shift up</a> <time>1:12:52</time></li>
      <li><a href="#launch-speed">Launching amid more speed, more noise</a> <time>1:14:55</time></li>
      <li><a href="#favorite-ai-use">Favorite unconventional use of AI</a> <time>1:20:31</time></li>
    </ol>
  </nav>

  <div class="lt-transcript__content">
  <h3 class="lt-transcript__chapter" id="cold-open">Cold open <time>0:00</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">0:00</time></p>
    <p class="lt-turn__line">This is what tokenomics is right now. You ask different people who are in different parts of the order to describe it, you get wildly different answers. The thing is, they're all right, but none of them are seeing the bigger picture.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="welcome">Welcome to the show <time>0:19</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">0:19</time></p>
    <p class="lt-turn__line">Well, Jr. Thanks so much for being on the show today. I'm super excited for this conversation. I'm glad we were able to get it done between two weeks of me floating the idea, to being here in person with you. Super excited to spend the next hour and a half talking about everything tokenomics.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">0:33</time></p>
    <p class="lt-turn__line">It's really good to have you here, my home turf of Southern California, and I feel like every time I see you at a conference, and so we never actually get to sit down. And the problem with AI, I feel like, is that we all just try to make everything summaries and go fast. So I'm looking forward to digging in a little deeper level with you.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">0:48</time></p>
    <p class="lt-turn__line">Awesome, let's get into</p>
  </div>
  <h3 class="lt-transcript__chapter" id="finops-to-ai-reset">From FinOps at Citi to the AI reset <time>0:49</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">0:49</time></p>
    <p class="lt-turn__line">it. Okay, so I'd actually start with when you and I first started talking, and you were at Citi, yeah, and you were in the FinOps world, and you were doing that there, and you know, big company, lots of clouds, all those things. We watched over the years a lot of companies like Citi, you know, move through that chasm of how to do FinOps, figuring it out in the early innovators to the like early majority who had it sort of sorted, to getting really good at it, getting more mature over time. Then everything changed, right, with with AI a couple years ago, and and we kind of got reset back to the very beginning, which was how do we do all this stuff and things that you know you were doing at Citi before you you know came over to where you are now? You guys took years to figure that stuff out, right? And in the process, everybody else is also figuring that, and we're right back to essentially where FinOps was in say 2020 with all the AI stuff.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:36</time></p>
    <p class="lt-turn__line">Yeah, it's so funny. I find that even in FinOps, there was always these points where you had to reinvent the wheel, right? And so, an example is, I think 2014 VMs, S3s, buckets, and everything. The idea of ownership was really important, and we say, well, let's tag stuff so we can now actually, you know, make out some understanding out of the mess of car files that everyone has, dashboards, and then something like serverless comes along where you can tag it, and then Kubernetes comes, which is much more around multi-tenancy. And so we solved that problem. We said, well, namespaces are similar to proxies for, you know, tags. And I think with AI, though, there is this there's this big push to almost look at it to say, well, we could patch on the things that were working before. But I find that even at Citibank, when we were there, we were starting to see the seams of where the limitations were. You have a provision throughput request. Well, it sounds like a savings plan, and all of a sudden you say, "Well, you know, everyone in the company is going to be using AI, and we're going to see one line item on the bill. And you think, "Well, how do we actually think about accountability, right? And so all of a sudden we start thinking about all these different initiatives that we have in terms of should we have it like a gateway that we can track everything going in and out. Even now, I think FinOps and tokenomics is really interesting because you know initially it sounds like the same, right? And then once once I started learning a lot more about tokenomics, I realized it's it's it's almost expanding the scope, yeah, right, and expanding the scope not only from the discipline, but it's also expanding the scope from who cares about</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">3:07</time></p>
    <p class="lt-turn__line">it. Right. And that's something</p>
  </div>
  <h3 class="lt-transcript__chapter" id="boards-ask-about-cost">When boards start asking about AI cost <time>3:08</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">3:08</time></p>
    <p class="lt-turn__line">you know. This week I've been spending a lot of time with CEOs from private equity companies, and the whole topic was FinOps, or rather AI, and AI from governance and from cost perspective. And I think back like I can't remember the last time CEOs level were clamoring for more information about cost, and not just from a level of hey, you know, the cloud cost is very important to us, but more of how do we measure and how do we control and how do we not limit our companies from using AI without bankrupting the company. Yeah, right. It's a very interesting paradigm there.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">3:40</time></p>
    <p class="lt-turn__line">I would go back to just a few months ago. I started hearing from a lot of folks at that level that we're we're at a bit of an existential crisis moment in terms of where AI has landed, and I'm thinking this back to just like March or April, where you know there was still all this excitement about use all as much AI as possible all the time, the token leaderboards, all that stuff. But your point about CEOs not talking about costs and boards not thinking about costs, the AI spend ramps so dramatically that they suddenly had to look at it as an existential crisis moment around is this going to like actually clobber overall margins and and can we deliver the services that we've been delivering under the current models that we're using and actually have them make sense and so I think it did shift a lot from AI should be all about transformation and innovation, and you know how do I do the thing that we couldn't do before using AI to wait a minute, not really sure that we're actually getting those outputs we wanted, and we do need to think about cost more than value. And I think we we had a bit of a pendulum swing happen where AI was all about the promise where we swung, and I think briefly back into. I mean, right now we're living this world of budget caps and hard lines, and how much AI people can AI can use. And I was talking to a friend just this morning, you know, who's getting emails. There is a Fortune 50 company, and they're getting daily emails. Their token usage and their token costs and their total AI cost, and they're actually getting downward pressure to use less, which is like mind blowing from where we were three months ago, where we were saying use more and more and more. And so we've swung into this place of control and fear. I think about it, and I think we're going to start to swing back a little bit more toward okay, how do we use the right AI for the right things and get the most value out? But it's a weird moment in time as well. Huge investment right into the space, and also people are recognizing that use more AI is not necessarily</p>
  </div>
  <h3 class="lt-transcript__chapter" id="token-maxing-rise-fall">The rise and fall of token maxing <time>5:30</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">5:30</time></p>
    <p class="lt-turn__line">a good business model. Yeah, totally. The I think token maxing lasted about four days, where I think Meta came out and said, "Yeah, we're gonna. And you know, I get it on one level because the technology itself, once you've tried it, right on your personal ChatGPT account or anything, you see the power of it. I remember when if ChatGPT first came out, I think it was the end of Reinvent a couple of years ago, and my friends and I were using it non. We couldn't believe it. I think I had translated like a SageMaker announcement into Hindi just to see if it would work. Did it? I can't read Hindi. I text my parents. They're like, "That's good Hindi. Nice, right? So it's it's just like the art of you know when you think about public cloud, no one's really consuming it as the way we consume AI, right? And so I think there's this pressure not only to think about the use cases of how it can help companies, but really what's really interesting is that there's a pressure on the CxOs now. Of unlike public cloud, which there might be pressure to say, "Hey, move from data centers over. There's more, you know, resiliency. You can scale up and down automatically. All the good benefits of public cloud that haven't gone away. But with AI, there's this pressure of, you know, you have to have a story for the board. You have to show what are you doing with AI. And I think the age of experimentation of we're trying POCs. We're trying this and that. Very quickly, to your point, have turned into well, what's the ROI? And the ROI is not in terms of even pressure of minimizing the cost. It's more of we want to invest in the places where it's quantifiable, right? Where it's actually going to help. And if experiments are not being able to show me value, we want to know quickly when to cut them off.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">7:00</time></p>
    <p class="lt-turn__line">Yeah. Yeah.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">7:00</time></p>
    <p class="lt-turn__line">Right. And I know, like Uber and everyone that came out this year said our strategy of everyone gets to use AI, we blew through our budget already. And I always, yeah, I always find that funny. That well, how would you have known the budget in January for something technology? That's every week I go on Twitter. It's completely different,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">7:15</time></p>
    <p class="lt-turn__line">right?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">7:16</time></p>
    <p class="lt-turn__line">Right. It was OpenAI was amazing. Then Anthropic's amazing. Now open source models are amazing. Reasoning models, on and on and on. It's just it's a fire hose of information coming in for everyone.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">7:26</time></p>
    <p class="lt-turn__line">And and I think we're we're going through the same thing we did with cloud, which was people were like migrate as fast as humanly possible, get it all there, and then we'll figure out the costs later. And then there was a big whiplash, which is where sort of FinOps came into being. You know, back in those days, we're in the same thing with AI, except that cloud was fundamentally a new. It was a it was a platform layer thing, right? We're moving from one type of hosting to another. Yeah, there's better services. You know, things like BigQuery and Redshift and these things. You know, are really hard to do in the data center. But it was just a different way to do this technology layer. And I think where we're hearing the interesting cases with AI. I hate to use this term because it was applied heavily in the cloud world, and I hate it then. But transformation use cases-it's-it feels like the most hand wavy term ever. Is yeah, wait a minute-we can't just say go use the AI to do what you're already doing better. I need to look at it in terms of like a 10x return instead of a 50% improvement, right? Sort of thing, which you know people are figuring it out, and and and this is where, for us, when we look at like the tokenomic space, this conversation moves suddenly from being one about let's optimize the platform layer in the FinOps to, in tokenomics land, like how does this AI spend touch every part of the business, and how do we make sure we can on one side even get enough of it, get enough of the AI on the token factory effectiveness, use Jensen's terms, to how do these scaling and very highly variable AI costs change the way that I need to repackage my entire offering to customers, and then hopefully in the middle, efficiently use it in the process.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="foundation-origin">How the Tokenomics Foundation began <time>8:58</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">8:58</time></p>
    <p class="lt-turn__line">Yeah, speaking of tokenomics, as we all know, Tokenomics Foundation launched. I think a couple of months ago now, we announced a governing board recently. I'm a member of, so hint. But I'd love to know the story of like how that came about, right? FinOps was FinOps is ubiquitous now. You know, every time I talk to someone when I first started FinOps, it was cloud financial management. It was you know optimization, on and on and on. And now we've coalesced on this word, so I'd love to hear more about what was the thinking behind starting a new foundation.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">9:25</time></p>
    <p class="lt-turn__line">I would start with the fact that I hated the name tokenomics when I heard it. I was actually really against it, and I was actually kind of late in hearing it. I think that term, now that I've done the backwards-looking research, I think started mid 2025, not applied to crypto, but applied to this area by like the semi-analysis, Dylan Patel, crew looking at token factory effectiveness, right, and how to convert NVIDIA or AMD or whatever hardware into token outputs. And then in January or so of of 2026, I think there were some others starting to put out papers about it, starting talking about this area. It was I was in actually in Manhattan where you or in New York where you live. At an event with a set of global CIOs, and this event had JPMC, and it had BNY, and it had, and these were published at the time of the event. And there was these global CIOs there having conversations about tech value, and I grabbed one of them, huge, huge company, responsible with billions and billions of budget, and pulled them aside and said, "Hey, what's your top challenge right now? And this person didn't know what I did, didn't know anything about me, and without missing a beat, said tokenomics. This is April, and my heart literally sank because I was like, "Oh, that's like an imperfect term for something that sounds a lot like what we would call FinOps for AI, and I kind of pushed it aside, and then later that day, I was sitting with a head of platforms for an asset manager in Midtown Manhattan, and they're like, "What are you doing with this tokenomics thing? And then I was at the offsite the next week for the Linux Foundation. Our CEO walks up to us, and he's in all these conversations with tech leaders all around the world, and he's like, "What are you doing with this tokenomics thing? And it was just point after point after point. And the funny thing with all this was, these are all people who they have teams multiple levels below them who do FenOps and have been doing FinOps for years. And FinOps is underpinning the efficiency of their infrastructure and and cloud and private you know cloud and and data center work now and all these different areas. But they were never talking about those areas because fundamentally, tokenomics is a discipline, and we will probably need to spend 2030, minutes here talking about what it is a discipline. It began mid-year to touch into board-level conversations. Yes, because the spending was spiking so dramatically, but also because we're in this age of scarcity. Right, these giant companies who need a lot of AI or think they need a lot of AI were getting rate limited. They were having models pulled back. They couldn't get hardware, so there was a lack of ability to get the thing that they knew they needed. And then on the flip side, they had their boards pushing them at the highest levels, and CEOs coming in. And at the same event in Manhattan in April, we had these global CIOs and CTOs saying that their their CEOs, like titans of industry, were going home on the weekends and vibe coding things and bringing it back to the global CIO and being like, why can't your teams do this? So you had CEOs saying, I need you to change everything we're doing like that, right, overnight and drive this massive transformation. And they're sitting there going, well, wait a minute, I've got the CFO next to me who's saying, "Yeah, great, do all this great stuff, but you know we have this this budget that we set last year, and we have like you know stock prices we need to think about. So you can't go spend unlimited funds on this. So they got stuck in this back to the existential crisis moment of my board says I have to transform. I'm working. My CEO is pushing me hard to deliver all this innovation. My CFO is literally sweating because this little AI cost that was this big next to the overall IT spend got to be this big, and then it got to be this big, and it was going. In some cases, we're hearing 50% month over month growth, month over month growth in AI. What do we do with that? And so we started to see some conversations happen around what was tokenomics. Now, what is tokenomics, right? And and that's the first thing we always run into with this is where does that start and stop? How does that relate to existing things like VenOps? And that's very much what we realized in April, May of this year. We needed to go figure out and define.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">13:18</time></p>
    <p class="lt-turn__line">And when when you think about tokenomics, then the name is interesting. To your point, I find that having learned about it now, initially the confusion might be that the word tokens in there, and is it is about valuation of tokens? Is it is it a discipline? To your point, is it methodology? How would you define tokenomics foundation? The purpose of tokenomics.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="not-about-tokens">Tokenomics is not about tokens <time>13:39</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">13:39</time></p>
    <p class="lt-turn__line">Tokenomics, as I'll say, maybe a discipline because everyone is figuring out what it looks like. Fundamentally, is about looking at the entire supply chain of AI. From I've created things in a AI factory at the hardware level to I've consumed these things hopefully efficiently, and I've forecasted my usage and all these things. To I've delivered value on the back end of that, so from energy to that intelligence layer outside to value. Now the value part is really tricky here because it's not about tokens, and it's not about individual services, and it's not about even task completed. It's about what are the outcomes that all this is driving. The thing I have to stress so much to people now, when when I talk to folks like you who are in this and working every day, talking to CEOs, talking to global leaders on this, tokenomics is not about tokens,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Speaker 1</strong><time class="lt-turn__ts">14:35</time></p>
    <p class="lt-turn__line">right?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">14:36</time></p>
    <p class="lt-turn__line">Weird tokenomics, not about tokens. It's right there in there, and tokenomics is certainly not about counting tokens. It is an imperfect name for a more complex concept, and I would say if I was to synonymize it, it's tokenomics is about AI economics. I didn't do maybe our community any justice with my my keynote and announce this because I focused so heavily on tokens. Because the interesting thing about tokens is they are the entire state. Of AI, you can trace them through all different levels. But as a term, tokenomics is not about token costs. It's about all the costs around AI. And to make it even more complicated, which is why we're seeing a lot of CFOs get heavily involved in this conversation. We're not just talking about the costs of the AI directly, right? We're not just talking about the costs of the hardware and the services and the hyperscaler costs and the memory. All these things we're talking about-all the people costs involved with managing the hardware and the data center, and the engineering costs to use those services, and then the labor costs on the back end. And we're seeing a lot of this cascade into conversations about how many people do I need in my organization based on the AI outputs and inputs? What is the value of an engineer? How should we compensate our engineers. What does that look like? So to draw it all back to the beginning of your question, what is tokenomics about? It's about getting the most value out of your AI, but the definition or the maybe the the boundaries of where it stops and starts are still unclear. There's a lot of definitional work to do,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">15:56</time></p>
    <p class="lt-turn__line">and in a way, I think that's it's a feature, not a bug, right? The reason we need a foundation around it, is because we need the smartest people to think about this. So somebody FinOps before, you know, everyone's trying to do their own thing. There might be a tool available. There might be you have your own spreadsheet that you're calculating stuff on. Yeah. And for me, the FinOps framework really clicked for me was the fact that it's not about doing all the things, right? It's about what does your company need now, and there's a way to get started from when you have nothing, or if you're advanced. Like I love the crawl, walk, run model around this. Right of if you really focus on optimization only, well, there's a lot of things you can do, and here's how to think about rate optimization, usage optimization, and and it's one of these things that without a sure you could probably get to it on your own, but for a lot of companies where they don't have the investment to be able to hire a FinOps person, and they don't have a person who's dedicated, or a team, or army of people for some companies that we know, where they have dedicated resources to thinking about how do we allocate costs, how do we get optimization, how do we build governance. The foundation does a lot of work in order to actually give you a lot of the homework done already by people who have solved this problem already. And I think with tokenomics, the interesting challenge is that the scope is so much bigger, right? The scope, to your point, is not about efficiency of minimizing the cost only. That's one vector that's a really important part. But I think a lot of it's also how do we know where to invest more, right? Whereas with public cloud, it'd be yes, we're going to migrate over, for example, from data centers or for a cloud native company, let's keep the costs at a level that we're happy with.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">17:24</time></p>
    <p class="lt-turn__line">Yeah.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">17:24</time></p>
    <p class="lt-turn__line">With AI, I think there's a pressure on the other side as well, and I imagine this bell curve. You know the meme where if you have an employee who's not using any AI, is on the left hand side, and you say, well, you have to be using AI, and it really connects not just engineering but marketing, sales, on and on and on because it does make you more efficient. On the other side is if you're using too much AI, the question is, well, what are you doing? You spend $10,000, and what did you get back in return? And I think in the middle is the reason for it's not a technology that you say our company doesn't need it. We'll take. We'll worry about it later, right? Where public cloud is still being built out. Let's think about it. If you have every single company from private, public, Fortune 100, Fortune 5, who has to deeply think about this because they know it can be transformational for the enterprise. But you know, I think on top of that, the idea of everyone just gets to learn AI and starts using it automatically. The reason FDEs exist everywhere is because it's not that easy, right? Integrating into your existing context of your company and making sure that I'm sure you've tried you know you try to write something with just generic Claude, it doesn't get your ideas, it doesn't understand your context, it doesn't understand your thinking, and once you start playing with it and you start building in skills files or whatever, it really does help you either brainstorm, create an output that you're okay with, and eventually get to an autonomous place where the human in the loop is not really needed.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">18:47</time></p>
    <p class="lt-turn__line">Yeah.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="gas-price-problem">The gas price problem with token costs <time>18:48</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">18:48</time></p>
    <p class="lt-turn__line">Let me ask you this: so around tokens and token cost and everything around there, how do you think about the unit level, right? So, you know, we talk about tokens, and I think yeah, I'm really curious to hear your take on because I have I have an analogy I will use, but</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">19:02</time></p>
    <p class="lt-turn__line">well, what's your analogy?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">19:03</time></p>
    <p class="lt-turn__line">So I you know I try to harken it back to, I think I used this with someone yesterday that looking at token costs alone is like looking at the gas price, right? You're trying to go on a road trip and you're trying to calculate how much money I need for gas, and if you look at only the gas price of $4 a gallon, it helps you on some level, but you also have to figure out what kind of car I'm driving. Do I have a race car? Do I have a hybrid? How long? How many stops we're going to have? Doesn't take into account. Well, what if we took the wrong turn and we drove 50 miles on the other way? On and on and on. And to me, looking at token costs in isolation is very similar to that, right? And it was similar to that last year, where the idea of reasoning models, agents was still taking off, but it wasn't really ubiquitous yet. And now you can have. I'm sure this has happened to you as well. You ask Claude to do something, and you walk away. You come back, and you're still thinking like, no, whoa, whoa, whoa, thinking too hard on</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">19:54</time></p>
    <p class="lt-turn__line">this, thinking too hard on this, right? And it's sort of like managing a person, right? Sometimes you'll have somebody put them off on a project. They go into rabbit hole for two weeks and come back. You're like, wait a minute! If you just checked in with me, right, with your early ideas, we could have saved a bunch of time. And you've got to, and that's this also gets into the interesting thing I think about tokenomics and just AI in general is we're now thinking about how to manage non-human workers, right? And that also touches into that far right side of the tokenomics equation, which is well, how do you think about accounting for non-human workers, and how do you can you capitalize the work an agent does, like some of the CFOs, right? And and likewise, it brings up really interesting regional conversations. The cost of engineer in San Francisco is very different than the cost of engineer in a different geo, right? And so the amount of AI you want to enable that person with is going to change, right? And so that's to your point of just doing the gas price. Sometimes you've got the you know Camry, and sometimes you've got the Ferrari. But you know the goal may be to get to one location to the next, or it might be that you're really looking to have a you know nice road trip, and all those things have to factor in.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">20:53</time></p>
    <p class="lt-turn__line">Yeah, especially I think Gartner came out last year or this year actually about the fact that if you use reasoning tokens, it might be five to 30x more than you know when it was, and I think tokens made sense from a price perspective before. Where if you have a chatbot, you know that the input tokens, the model you're using, is price to end output tokens, and now you have agents that can call and think on their own. They can do a lot of reasoning, and the output is what you really care about. But even at the output level, if a model that is cheaper per token, but it takes twice as long to get to the output because it's doing a lot of reasoning, very verbose mode, versus some model that might be way more expensive, but it can get to the outcome you wanted faster. It might be cheaper in the long run. It might be cheaper per task. It might be cheaper, whatever. So my thesis really is that the framing really has to be abstracted up a level, which was what is the point, right? What are you doing with AI? Is it to augment something? For example, if you have a call center and you want to now use AI for successful calls, right? 80% of the questions are the same thing. If you're a hotel, hey, can I check in later? Can I check out later? And if you can use AI to answer those things, well, what you really care about is not the token price of doing that. You care about what is the cost per outcome of a call that was successful. And I think without that level of information, and if you just look at a bill, it's very hard to justify. Yeah, we spent a million dollars last month on Bedrock or Vertex AI or any model. Should we invest more or not?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">22:22</time></p>
    <p class="lt-turn__line">Yeah, and I</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">22:23</time></p>
    <p class="lt-turn__line">think that conversation at the CxO level it requires this level of data clarity that is not just about the cost, but it's also the ROI side, the value side of tokenomics as well.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">22:34</time></p>
    <p class="lt-turn__line">Sometimes I feel old when I look back to the cloud days because now the cloud is like what almost 20 years old, but the the conversation that was happening in the early 2010 s around this is actually super similar here, which was before cloud, nobody could basically spin up a bunch of resources or infrastructure and cause the company a lot of cost overnight in the way that you had engineers doing that in cloud when they first got access to AWS and all these things, and that was a lot of the reaction to you know moving from this like long slow procurement cycle that existed in data centers, and you knew what you were going to spend, and then cloud suddenly blew that up, and then we got that under control because you know we can put enablement and guardrails and all these things around how engineers can spin things up, and we're we're now accelerating that journey into the AI world with now every single person in your organization can spend money, but also every single person in your organization can actually have a huge impact on what they're doing, right? And so, as we look across this journey and thinking about what I think needs to happen with tokenomics and these things, is we're getting to this place where we're rethinking the technology just as a way to improve the speed of the widget production, right? To how do we again transform? Let me spin it back to you. Where do you see the lines between tokenomics and the previous ways of doing these things, FinOps or otherwise?</p>
  </div>
  <h3 class="lt-transcript__chapter" id="where-finops-ends">Where FinOps ends and tokenomics begins <time>23:54</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">23:54</time></p>
    <p class="lt-turn__line">Yeah, interesting question. I find the there's two ways to think about it. One is that the lessons learned from FinOps can be applied here, and it's expanding. Right in FinOps, we really, to your point, didn't think about cost as anything more than a unit price for doing something. Right, serving to our customers, revenue is this, or cloud cost is blah. But there's certain things in FinOps now, even that I think because of AI, it's forcing FinOps to evolve as well, right? An example of that is this idea of ownership being the reason you charge someone back, right? Showback and chargeback is usually based on tagging. You say, well, your team spun this thing up, and and so you should own it. You should optimize it. You should govern it. You should be efficient using technology like a EKS cluster or something, and I find with AI now, you know, we already start to see the cracks in that model, right? When you think about something like like serverless or a shared Kubernetes cluster, and what happens is, you know, you get the serverless thing, your team spins it up, and another team is the one consuming it nonstop, and you get the build because. It's tagged to you. Similarly, in Kubernetes, fine. You you may be able to delegate to say this namespace or this node or whatever belongs to you. That's your cost. But there's all these auxiliary costs, right? If you have one service, for example, owned by Product TMA, and they do ton of egress costs to the internet, they're doing something there. That tag is that cost is untagged, and the best thing we probably had was well, let's divide it proportionally. Let's fudge the numbers a little bit because exactly right, and and and I think with AI especially now, that's a default and not you know something that the advanced mode of multi 10s here whatever yeah, and so we're forced to confront this infinops anyway, and so I think a lot of the stuff that'll happen on the tokenomic side of a solving ROI is another example. Where in FinOps it was unit economics, right? And look, I talking to customers. It's been 15 years or 10 years talking about unit economics. A lot of companies still aren't there yet because it's an</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">25:56</time></p>
    <p class="lt-turn__line">aspirational goal in a lot of ways,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">25:58</time></p>
    <p class="lt-turn__line">and and it's difficult because the data that sits to do unit pricing doesn't come from cloud, right? It comes from your business. And so, if, for example, the cost for something sits in Salesforce, another one lives in Excel, another lives there, trying to connect the dots together makes a lot of sense. You know, what did I get back in return? But I think with AI, especially now, it's not a nice to have anymore. Aspirational, to your point. Yeah, it's the thing that you have to solve for because otherwise, how can you invest money into this extremely expensive, extremely fast-paced environment where you know there's something that you want to do with AI in your company? Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">26:35</time></p>
    <p class="lt-turn__line">and I actually think maybe slightly controversial take given the moment in time. I actually think we're going to see more token maxing as a concept in the future because this pendulum is going to swing back. And right now, there's a there was a keep your head down if you're not using AI. Now there's a keep your head down if you're using too much. But you know we actually do want people being enabled. The other interesting thing we're starting to see around tokenomics and the sort of I'm going to get to unicos in a minute here is how this ties all the way back to capital investment in the data center and how that connects on the far right to stock prices. So last week I was reading the earnings report for SpaceX, and right tucked in there was a conversation about their revenue per watt in the data center, and they quoted a number for their revenue per watt in the previous quarter was this, and it went up. It was a huge jump. It was like 3x or something in the the current quarter, and then they were able to take that and basically say, well, if we know how many gigawatts that they're adding in capacity, we know what the revenue per watt is now. We can forward project out the AI revenue, and that actually drove was the thesis or hypothesis in this article. A lot of the stock rallies that they saw that set tokenomic supply chain in action there, which is the core unit on that side is was a was like I've got this much capacity. How much can I sell that per you know per watt in that case, and put it out. That's an easy one for an infrastructure company, right? And I think we are seeing this is probably more of a Fortune Hunter conversation instead of like, well, how can we monetize our access to GPUs and related hardware? On the everybody else side, I think the unit conversation is getting more tricky because this isn't just a like, how did I accomplish the task faster? This is how did I use the AI to drive more revenue and outputs? Or I mean, I got a call literally on my drive up here from somebody at a massive company, and they're slowly laying off a lot of staff because they're enabling more outputs. At least that's the message they're getting right from AI-enabled people, right? So, what is the right measure? And the right measure, I don't think, is a single unit metric. It never has been, but it's on a per team, on a per org, on a per product basis. How do we first get more efficiency out of those individual units? But then, how do we drive more revenue on the back end? And and we're seeing a lot of that connection, which is why we're doing this work too. Like you asked me what tokenomics is, and I wish I could give you like an easy answer, and I have an opinion on it, but I think the industry is figuring this out. What we do know, and back to your early question of you know why do we do this, is we we do know that the tokenomics term and the the idea that it encapsulates is in the zeitgeist right now at every level, and the CEOs and CFOs and CIOs and even in these earnings reports of trillion-dollar companies because it pulls together the thinking around a concept which is there's a whole supply chain around AI that starts with energy and outcome being at the end revenue and stock prices and there's a whole bunch of work in the middle we need to connect in there and right now there's a bunch of disparate disciplines, and FinOps is one of those disparate disciplines. It sits in there at that platform layer, cloud, variable technology, SaaS, software, these things, and then you've got some other data center people over here doing these things and building out capital investment, and then you've got a whole nother set of business people doing these things, and tokenomics is looking to kind of pull those together. So while it can be. Just easily described as taking energy and turning into intelligence and driving value on the back end, or you could say it's synonymous with Denops for AI. And parts of it are there's a bigger thread here that's developing, and I think we're going to see essentially a yeah new discipline that develops. And where I'm trying not to be dogmatic, and I think it's really hard as somebody who's in this space is like, what's it going to be called, and does it going to encompass the old stuff or the new stuff is a new thing, and we just don't know yet. Which is why I'm so great excited to be having these conversations like this with you to figure that out.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">30:26</time></p>
    <p class="lt-turn__line">Yeah, at the end of the day, it's just a name, right? I still think now FinOps is-is it financial operations? Is it financial? Oh,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">30:33</time></p>
    <p class="lt-turn__line">I hated that term. I know it's, but you know, like a terrible name, though. Also, yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">30:37</time></p>
    <p class="lt-turn__line">it's just a name, and and I think the, but I think the value of something or FinOps Foundation and now Tokenomics Foundation is that we're building out in the open in a way, right? We're not. No one solved it because, to be fair, for every different company, it's going to be a different solution. Similar to the framework was not prescriptive. It was more of well, what do you need, right? What is the thing that you're prioritizing right now? Is it being able to have clarity of data, and once you have that data, do you want to now think about how to be more efficient? And if you if you have efficiency, how do you prevent the fin oops moment? I'd love to call it. I love that one too. Yeah, and I'm sure, especially in tokenomics, like the the questions are similar, but they're very nuanced for AI specifically.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">31:18</time></p>
    <p class="lt-turn__line">Well, and there's another thing. I mean, as you're talking about the FinOps moments, people have been asking me the last few weeks that I'm saying no, no. Which is is tokenomics a rebrand of FenOps, whether that be as a at a discipline level or at a foundation level? And the answer is like absolutely not. We there was one of the moments I did when I was doing the research about whether we spend up a new foundation on this or what, just what to do with in general, I had this amazing period of like five or six days where I was able to get 20 of the Fortune 25 on the phone and folks I knew there who were in this space and people were taking calls on like a Saturday and a Sunday at seven a.m. on a Saturday talking about this stuff and one of those folks said it's so funny you're asking me this if it's the same or different because I just came out of a meeting with our, I guess it was their CIO, and they said the spinoff stuff is really important, and I need you to make sure you're you're taking care of it because it's the majority of our spend and like that's a big bulk and but I need you to convert that to be business as usual focus like that's running you've been doing this for years keep that going I need you to redirect like 80% of your smartest people to go figure out this tokenomics thing because that is scaling like mad, and our board is asking questions about it. And that kind of was a defining moment for me, where it's like you need to keep doing the FinOps work that you've gotten good at, and then you also need to go figure out this new thing. And so when I put that into the lens of what we do at the foundation was, we are keeping everything we're doing on the FenOps Foundation side. FenOps X is living; it's got a sister conference now. The training's living. The working groups are living. Everything is happening there, and our team is living there. And we're spinning up a new team to go solve this question mark of what is this and how do you do it? Because it needs a whole bunch of investment, and we can't stop doing the FenOps work because if we do that, all the costs there are going to go wild, and our forecasts are going to be off, and we're going to have issues. And so we're seeing this play out. And so over time, I think the question becomes: Is this all the same thing, or is it different as one and the other? And and that's what I think we'll figure out as we get there. Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">33:15</time></p>
    <p class="lt-turn__line">and that question was already being brought up about ITAM, right? ITFM. Are we going to have a merging messier</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">33:20</time></p>
    <p class="lt-turn__line">even now, right? Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">33:21</time></p>
    <p class="lt-turn__line">and I find the the value side of things of the ROI, for example, that also exists in in FinOps always. But it was okay to estimate because it was a controlled spend. It was a controllable thing that you could see, and you could have guardrails on everything. And really, the people who are touching cloud is engineering team, the DevOps team, the SREs, the IT team, and I think that the CXO level is like, well, I trust my CTO or CIO or IT director; they have a good handle on it. And even if the data is not perfect, even if we don't view economics, it's okay, because we know the total cloud spend is what we care about on aggregate. And at the end of the day, what we're really using cloud for is moving, doing business processes, building you know software or whatever, and with AI it's very different in that what you're doing with AI is two things. One, you might be enabling your product if you're selling it to now have your customers have a better experience, right? You might be able to say instead of learning configurations, tell me in English what you want to do, and we'll just figure it out.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">34:20</time></p>
    <p class="lt-turn__line">Yep.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">34:21</time></p>
    <p class="lt-turn__line">But then there's also this other side, which I don't think FinOps ever had to worry about, which is AI can help you with operational efficiency internally, right? You don't have marketing people spinning up EC2 instances to make their days better, but with AI, well, you do now exactly, right? And so this this always existed. Of the one wrong line of code could be a t3 micro to from a penny an hour to $99 an hour. If you made that a p5 instead on AWS, and with AI now, you know everyone gets access to a chatbot equivalent, whether it's controlled because you're limiting the scope of it, and the guardrails that exist today is just. Number, we say, hey, we'll limit you to $200 a month, and if you have a good reason, we'll expand it. And you know, if you're if you're someone using these technologies, it's not like you can just say, hey, I just have a question, figure it out in the background. You use Claude, it's it's Fable Five or this and that, and I think everyone, of course, wants the most intelligence as possible, right? They are not trained to think. Well, this task should be haiku model. It really should be want the most intelligence, of course. And I think the the experimentation phase of learning AI does require a bucket of spend to say, in order to learn AI, you can't go to a course for 40 hours and then you get it. The best way to learn is to use AI. You know, I have a lot of friends who ask me, "Hey, how do I get really, really deep skilled in quad? I'm like, "Why don't you try asking Claude? Right? Like, what is the things that I'm missing? What is the things that I can do with it? And I think this exploratory nature of AI also means it's really hard to figure out on January one when you're doing budgeting or November for the next year, what is their spend going to be? Right, Jevin's paradox. It's yes, of course, tokens will get cheaper and cheaper and cheaper. But what that means then is like, well, you know, the tokens for inference at the same level. DeepSeq just</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">36:12</time></p>
    <p class="lt-turn__line">announced increases in some of their token prices. Right, right. And so this is the highly variable nature of this. But I think there's something that you were hitting on that is always the opening to this conversation, which is we all start talking about model selection.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">36:24</time></p>
    <p class="lt-turn__line">Yeah,</p>
  </div>
  <h3 class="lt-transcript__chapter" id="model-selection">Model selection and the harness around it <time>36:25</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">36:25</time></p>
    <p class="lt-turn__line">and model selection is critical. I would say it's one of the most defining things of tokenomics as a in the optimization layer because we think about production consumption, which includes optimization, and then the value side. But one of my favorite quotes from Dario a couple months ago is that models are hitting the end of the exponential meaning. The models are not every release after release are not getting dramatically better. Like I was looking, I went from 4.8 to five. I was hoping for my life to change, and it didn't really change, right? And so, the model is an important part of this, and the model is often where we have the token conversation, and that's where a lot of the token cost conversation is. But the other part of this is the entire harness around the model, and the context layers, and the intelligence layers, and the memory, and the people, and the training of those people, and enabling the people, and so I think we are going to have in tokenomics as discipline and in the foundation a set of work that will always be about model selection, prompt select, prompt engineering, making sure you don't break your KB cache, all these things that are core. But increasingly, we're seeing people say, "Well, how do I figure out how to bring in all the context my business needs into that? Because I mean, look, I had this yesterday with somebody in my team. Great guy. I'm going to call him out, Stephen Arthur. He he's he's got a bunch of hardware at his house, and he's you know it's all protected, and he's running open weights models on it, and and it's amazing, you know, and it's like the tokens are nearly free because he spent all his money on the hardware, and I was getting some outputs from him, and I was like, this isn't right, and we realized like I have a whole Claude you know set up over here with the rest of the team that is all connected, and and his open weights model, which in some ways should be more powerful and is definitely more cost effective, was missing all this context. So I was like, say, hey man, like I need you to like move some of that stuff back over, which gets into well, this isn't about cost per token, right? Like I'm I'm I'm the budget owner in this case. I'm telling him please stop using the cheaper option because you're missing that context, and we're seeing the big companies out there. I mean, whether I mean, I'm not going to mention individual company names, but the giant technology providers who are saying, "Yeah, yeah, we can swap in whatever model you want, but we've got access to all your other data in your database, in your data store, in your email, in your calendar, in your CMDB. All this stuff is really where the value comes, and then that's also by the way we always have to go back to tokenomics is about that total like the token cost the model selection is the tip of the iceberg right and if we focus too much there we're missing the forest for the trees it's</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">38:51</time></p>
    <p class="lt-turn__line">funny you mentioned the the tokenomic side of open source models right ATT came out last week and they're earning support I think they're using up to 25% of their token spend is open source and they want to get up to like 70% or something like that. And you know, it's funny because initially, when I first saw the creation of tokens, Adams two tokens in the Tokenomics Foundation, I was pretty confused in terms of because you know my mindset was well, as a finance practitioner, I really cared about segregating the data, optimizing it, governing it. I never really cared about the energy that AWS is using to create a token, for example. But the more I learn about, the more I see there's going to be a mixture of different ways we create tokens, right? Might be cloud, obviously. It might be the you know the these service providers, but there's also these neo clouds, fireworks that can host open source models. And efficiency is not you know the same hardware running on AWS, GCP, these Neo clouds, your own you know own data center, you're gonna get very different outputs. There's a lot of efficiency gains that they're making, and how do you actually compare to say where do we invest in going forward? You have to look at the creation of tokens, right? And you have to figure out well what is the mean. That we can actually have an honest discussion about: Do we invest more? Should we go back to the data center and just buy a bunch of NVIDIA hardware, or should we continue to pay for Clod? To your point, because the ecosystem is where we want to be, and even though it's cheaper, it might require a lot more context loading or you know copying data around for us to be able to use it. Versus, it's worth the premium, premium in quotes, of using something like Cloud because it works faster. We end up using less tokens, and we are able to, you know, get to the output we really wanted without having to, you know, nine hours of going back and forth with Cloud, which is fun, but it's also exhausting sometimes.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="energy-scarce-resource">Energy as the new scarce resource <time>40:37</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">40:37</time></p>
    <p class="lt-turn__line">Well, and I wasn't saying Cloud was better in that case, but in the instance of the cloud that I was using, it's got MCP connections to a bunch of other systems that we're using, and it's got a lot of shared memory, and it's got skills files, and it's got access to you know our databases, and it's all those connections that make it valuable. So we could easily swap out cloud in that case for the local model he's running if that local model had connections to all those things. But I want to go back to what you said about energy because this is a really interesting area. I think where people say, "Well, how is this fundamentally different than you know FinOps? and and we touched on this a little bit at the conference, but FinOps assumes that you can just go procure the assets that you need, and they're there, and that this infinite scale, infinite scale, they're there. You hit a button, or you write code, and it comes back. And and this also touches on ITAM, which you mentioned, right? Which is touches much more into the compliance and the security angles of it, and the procurement aspects and the sourcing of it. But when you start to think about AI services, and particularly if you're a big company at scale doing this, you've got to say, I don't know if I can get access to the hardware. I don't know even if I'm using a cloud provider. If that cloud provider will have the needed energy at the time I need to spin up a big whatever is I'm spinning up, and energy becomes. And I want to clarify as well when we're talking about energy, we're not necessarily talking about sustainability, although we might that is a subset there. But energy becomes an asset that is limited that we need to think about: Do we have enough? If we're a big provider running a lot of things, and then we also need to think about the costs from that energy, which feed into my revenue per watt in that SpaceX example or something else. And so we are seeing the first energy conversation come in. And when I talk to you know data center providers that we work with, and they're they are trying to get not just the right cost for the energy, but the right sources, right? They they need they need their diesel backup generators and the grid, and they need to buy solar and they need all the mix because they can't get enough. But all that wraps back into the idea that in AI land, there's a scarcity of resource, and some providers are slower because they don't have enough energy and enough hardware, and some of the big companies are unable to buy what they need to do that. So tokenomics is also looking at that left area of, I mean, in Jensen's words, the token factory effectiveness of how do I take energy with the right hardware and the right capital investment? You said, and I often say as well, to output tokens, but it's really to output just an AI service. Now those services can be boiled down to token in some aspect, but it's all the wrapping layers of memory and database and storage and all the things around that as well.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="where-it-lives">Where should tokenomics live in the org? <time>43:06</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">43:06</time></p>
    <p class="lt-turn__line">You bring up an interesting point around the energy side, right? Compared to FinOps, where FinOps we really were focused on cloud and leveraging it, and obviously you know we started expanding the data clouds, SaaS. I think cloud plus was a term we used before,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">43:21</time></p>
    <p class="lt-turn__line">some some did yes right we did as well for a bit yeah and</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">43:24</time></p>
    <p class="lt-turn__line">and I find that at least with AI because of these considerations you have to take it's not just saying well we're going to use open source models right it's cheaper well it's cheaper but where are you going to run it are you going to run your own hardware are you going to have the power sources to actually power this are you going to run it on a different provider so it's still essentially an infrastructure as a service, on and on and on, and I think the the question I have is in an organization in any organization that is now thinking about tokenomics, and it's not one of those things that's nice to have later, right? When I think about FinOps, oftentimes it might be like we're too small for FinOps. We don't really need a person dedicated to. We don't need tools for it. We can manage it, but tokenomics is opposite. It's like we need someone focused on this, and oftentimes it's coming from the board and the CXOs. We need to focus on tokenomics because the power of it, you know, the the ability to transform, like you said earlier, our entire business is too massive to ignore. Where do you think tokenomics as a discipline sits in any say in an enterprise? Is it in traditional with the FinOps team? Should it be a different discipline altogether, like the CAIO officers we see everywhere now? Where do you think it lands?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">44:34</time></p>
    <p class="lt-turn__line">I don't know. We're being really careful not to define unless we really see the pattern play out over and over and over again, a tokenomics practitioner in the way I think there's a FinOps one, because it touches so many different areas. If you get on the phone with somebody from a NeoCloud and ask them about tokenomics, or you get on the phone with somebody at a cloud and ask them about it, or you get on the phone with somebody in a big bank in Neo. York, who is in a finance team, you're going to get completely different answers about what it is and where it sits. And so, from where the conversations I'm having right now, we are seeing it as a term to describe a set of capabilities or discipline, you might say, but that needs to be spread across the organization, lots of different places. And so, I am hearing from a lot of our people at these big companies. They're getting the call from their essentially down from the CEO saying, "I need you to set up the tokenomics practice, and I need you to help us figure out the right personas and what we need to hire and the team. And I can't say with clarity right now exactly where that should sit or who should run it. I can see some data points. There's data points like some are setting it right next to FinOps, great, easy. But then they're usually only looking at that consumption layer in the middle. Some are setting it in the chief AI officer's team. Some, when there's a lot of hardware involved, are you know shifting it more left to those in that area. It's like that parable about the blind men and the elephant.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">46:00</time></p>
    <p class="lt-turn__line">What's the parable? I don't know. You haven't heard that. Okay, okay. I'd love to hear. Okay,</p>
  </div>
  <h3 class="lt-transcript__chapter" id="blind-men-elephant">The blind men and the elephant <time>46:04</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">46:04</time></p>
    <p class="lt-turn__line">heard this in in elementary school, but there's these six blind men who have none of them have ever seen an elephant, and they're taken to this palace to go experience an elephant for the first time. And elephants in the middle of this room, and the six blind men approach the elephant from different angles, and one of them puts his hands on the trunk and says, "Oh, an elephant! It's like a snake. And another puts his hands on the tail, and it's an elephant. It's like a rope. Another touches the leg, and they said, "Elephant! It's like a it's like a tree trunk. This is what tokenomics is right now. You ask different people who are in different parts of the order describe it, you get wildly different answers. The thing is, they're all right, but none of them are seeing the bigger picture.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">46:45</time></p>
    <p class="lt-turn__line">It's really interesting. I think it might be because FinOps is a victim of is a victim of its own success. In that, you know, when we think about FinOps when we first started the foundation, there was a lot of learnings. There's a lot of work that had to be done to get it to where it is today. And when we hear Tokenomics Foundation, I think there might be a you know understanding of like oh cool you you already solved</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">47:06</time></p>
    <p class="lt-turn__line">it right and</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">47:07</time></p>
    <p class="lt-turn__line">and people don't remember it took so much work it took so many so much brain power of people to pulling and pushing at this idea of how do you do cloud discipline to get to where we are today</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">47:18</time></p>
    <p class="lt-turn__line">absolutely and we started I mean say we like Mike Fuller, who's the co-author of the the FinOps book, and I always talk about him as the brains behind FenOps. Like we started having these conversations in 2014, right? And it wasn't until 2018 where we thought, oh, maybe there's actually a discipline here, and then we tried to go document it in 2019, and then I would say by like 2021 we felt like we had a good handle on it, and by 2223, 24, you saw mature practices, and then AI blew it all up, right, and reset. So that's where, yeah, to your point, I mean, this stuff doesn't get figured out overnight, and we're right back in that cycle with tokenomics where we're we're having these deep conversations about like, what is it, who's doing it, and and we did all this, yeah, five years ago in in FinOps, but it never had this level of focus, and it was never this-I was going to say urgent, but it was never this critical to figure out because cloud was just one piece at that point, and then people figured how to apply FinOps to other areas. But this is back to that existential crisis moment where we collectively, as industry-and I'm so glad you guys are doing this-and we've got all the other members are plugging into this. We we've got to sort this out because we don't want to see like the cloud repatriation story play out in AI, where all this investment goes in, and then people say it's not really worth it, right? I mean, that has individual consequences for companies, right? Has big consequences for the global economy as well. And figuring this out and doing this efficiency work, I sound like I'm making it a grander thing than it is, but like we really need to figure out how to harness all this investment into value and outputs, right? And to bring the people along in the process.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">48:47</time></p>
    <p class="lt-turn__line">Yeah, in in a way, I think both the foundations really what they really give you at the end of the day is the confidence to use technologies, right? The confidence to spend in the cloud without being worried about well, the bill is going up, right? Unit economics helps you figure out. Well, the unit price is actually going down. We're getting more volume. Yeah. Example I used to use is like if you're on the front of TechCrunch and you're a Series A company, and you get the alert for anomalous that says you have tripled your cloud bill in the last two days, we're all going to freak out, look at it. Someone's like, yeah, because we have tripled the traffic.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">49:15</time></p>
    <p class="lt-turn__line">It was Black Friday. We made all this money. Exactly. And that's the you know there's that whole thing about what gets measure gets improved, and I'm probably going to misquote it, but there's I think it's Goodwin's law that basically says once you put a target against a measure, the measure stops being useful because you start to. And this is actually the the challenge with unit economics as well, which is we want to get to unit economics where we're looking at business value over cost, but if we start optimizing on just cost per transaction, totally, that might be the wrong metric, right? And so this is, I think, one of the struggles with unit economics is a concept that organizations have gotten to, and I think often the unit economics conversations have ended at that platform level. Like, what is the unit economic measure for the platform, and it hasn't connected enough to the product.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">49:59</time></p>
    <p class="lt-turn__line">Yeah, and. The gamification right as soon as you have a metric for someone like Token Maxing came out because someone said you got to use more tokens, you got to learn how to use AI. Okay, well the quality of what you're doing is very different. Someone is just repeatedly asking the same questions. Someone is actually learning and applying and getting more efficient with their own the way they work to get the same output they really cared about. But if you put a measurement of saying, you know, even limitations of this group should only spend $100 a week or a month on AI. I think you kind of miss the bigger picture, which is, what does that $100 really get you? Is the point, and you want to be able to quantify and be able to measure that, because if you can't measure that, then it becomes a cost number, and that's an absolute value, right? And you know, it's funny. Like I think initially there was always maybe last year, two years ago, there was a separate budget for AI. We're gonna try a lot of stuff. You know, it's okay to experiment, POCs, whatever. And increasingly now, it's saying, yeah, either it's gonna be folded into our overall budget for cloud,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">50:56</time></p>
    <p class="lt-turn__line">or it'll be separate</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">50:58</time></p>
    <p class="lt-turn__line">for teams. Yeah, shifting gears for a second, I'd love to hear your opinion on how do you think the landscape is going to evolve for SaaS companies where traditionally it's been a seat-based model, and all of a sudden now you're saying you know with seat-based and if you're using cloud, you have unit economics, you can control it, you have margin protection. With AI, say you integrate a chatbot into your experience for your customer, I think we all remember the Chipotle thing that came out, where someone figured out that well, if I just score the chatbot for Chipotle, it's a personal GPT. It's free. Yeah, I don't have to do anything about it. And so the charge model, especially for these SaaS services that are coming out, especially if you're an existing SaaS business and you add this thing to it. There's so much variability now involved with it, right? You might have one customer who says instead of APIs, I'm going to hit your MCP server, or I'm going to ask in English every five seconds for it. Whereas before, API costs you can manage, but with you know, if you have an LLM translator in the middle, it's still going to the API to get the same data, but people are doing it a different way. It's easier for them. How do you think the industry is going to evolve when you think about SaaS pricing? I</p>
  </div>
  <h3 class="lt-transcript__chapter" id="saas-pricing-agents">SaaS pricing in the age of agents <time>52:04</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">52:04</time></p>
    <p class="lt-turn__line">saw a headline today about a company whose stock price had tanked, and and the the name they had put on them, the epitaph or whatever the term is, was oh they're just an AI wrapper, right? And and that's that's the big thing. I think the risk for a lot of the SaaS companies right now, and this gets to a more existential question for SaaS companies and software companies: is where does the value sit? Does the value sit with the underlying AI? Does it sit with the specialized knowledge you lay on top of it? Does it sit with the data you have? What does that look like? The variable consumption basis component is huge right now, and the thing that's missing in that conversation, you know, you said you can track API calls, manage those, forecast that. This is where agentic comes in. We haven't said agentic yet much, anyway. Which</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">52:50</time></p>
    <p class="lt-turn__line">is we survived an hour without seeing agent. Amazing. Which is,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">52:53</time></p>
    <p class="lt-turn__line">you know, you you ask that SaaS offerings chatbot to do a thing, and it's not just doing the thing. It's spinning up a series of agents. We're interacting with agents and going off and creating more agents and loops and retries and failures, and that can add a lot of value, and it can also just add a lot of cost on the back of that, right? And so, I think you know I'm hearing a lot of conversations I was on this morning with a giant, call them a SaaS company, you might call them a platform company, they might call themselves a cloud company, you know, and and they're looking at they use a lot of different AI models they're passing through, and they're having to reprice how they're selling to their customers and figuring out what their benefit is and what their what their margins look like on that. I hear others who are struggling with just passing through model costs directly, and you hear some that are doing it that direct, some are marking it up. Some are doing credit systems. Some are doing credits plus direct markup. It feels like we're. I mean, and to go back to the past 20 years ago, Salesforce was one of the first ones to do like a SaaS model on top of this and figure out the pricing. We're all we're we're back to ground zero with all of that right now. I mean, there's the gloves are off. There are no right and wrong ways to price at this point, but everybody has to figure out the variable component and then figure out if they are an AI company or if they're an AI wrapper company. And can you have value if you're just wrapping AI? And what does that look like? So, I think to your question ultimately of like, is seat basis going to be around? I think it will, but we're starting to see bifurcations between is a seat for a human or is a seat for an agent to use, right? I think who was it? ServiceNow made I think maybe good and bad headlines with their tollgate thing, which was if you have agents hitting it, we're going to charge you differently, and and that's actually I guess kind of makes sense because I can only type so fast, but my agents can go all night.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">54:41</time></p>
    <p class="lt-turn__line">In a way, the the idea that we had started in FinOps about SaaS class being coming in, it's speeding up FinOps adoption in a way, right? Because all of a sudden, to your point, SaaS class may not have been included in FinOps when it was seat based, because you're like, well, we're not really worrying about adding and removing seats every day, whereas. The benefit of you know FinOps efficiency is that the action you take right now, the cost is saved right now. Right, you have a Kubernetes cluster that you optimize immediately. You see the benefit. You're not waiting a year or whatever.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">55:13</time></p>
    <p class="lt-turn__line">I I already miss those days. I feel like FinOps is so clean in that way. It's so clean. Take an action, you get immediate results or pretty quickly totally traceable, and we we've lost some of that.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">55:21</time></p>
    <p class="lt-turn__line">But but now, in a way, like to your point, if they're SaaS companies or technology companies, for lack of a better word, right? SaaS may not be the right word anymore. If they're doing variable based pricing, you know, your AI cost is not just the spend that you have internally on tokens, because all these SaaS places are also being charged are start might start charging for their consumption and pass through charges to you, and so all of a sudden, when you think about, well, what's our total AI budget? How do you think about these things on the side, right? The things that are might be even superseding what you're spending internally on just pure quote unquote AI. If a company, a SaaS provider you're using, is charging you because everyone's using their shiny new chatbot to answer things, and you're getting the bill for it, right? So all of a sudden, like understanding what is the total cost of ownership, which again, so easy to say, total cost of ownership,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">56:10</time></p>
    <p class="lt-turn__line">TCO,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">56:10</time></p>
    <p class="lt-turn__line">TCO, right? And yet it was already difficult before, but I think now understanding well where does it sit in the budget, right? And I think I really do think this idea of ownership versus consumption will keep playing out again and again and again. That if a product team is using four SaaS providers, they're all you know consumption-based charges. They have AI spend. They have. They also have things that are dynamic in cloud. All of that combined together is the cost of building that product. It's not just one thing or the other and whatever. And this bifurcation of this cost sits with this team. This cost sits with that team. In a way, the you know all the conversation we start in FinOps are going to be accelerated. The conversions, for example, of ITFM, ITAM, FinOps together is because well now there's no choice.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">56:53</time></p>
    <p class="lt-turn__line">Well, and that's I was just thinking about that too because some of this TCO stuff is like classic ITFM totally roll up all the costs, make sure it's there. As a side note, one of the best descriptions of TCO I heard recently was tokens, caffeine, and outputs, which is way cooler than total cost of ownership. But but this is where I think we've all got to like lose some frigging dogma around like what the discipline is, where it sits, because you need that TCO ITFM view. You need the ITAMI license compliance stuff, procurement. You need the FinOps variable, accountable cost efficiency work, the tokenomics scale across that supply chain, and getting into models and caching and all the things that others just don't do. They've all got to come together, and it's all about technology value.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">57:39</time></p>
    <p class="lt-turn__line">Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">57:40</time></p>
    <p class="lt-turn__line">technology value is the thing, and like we even renamed our our little umbrella in the Linux Foundation to be the technology value umbrella because it has fanned out, it has tokenomics. We're doing work around ITFM now. You know, there's there's billing data and costs and usage stuff like the focus specification, which you you were on the steering committee for back in the day in your city days. But it's it's all about like we we need to get all the data together to make the decisions. And you talked about accountability. We haven't touched much on that, right? One of the most beautiful things of FenOps land, and I know you guys have some products and things in this area, was you could actually take a portion of infrastructure spend and say, "This is yours. I'm it. This is yours. And when you take an action, I can show you that that action had impact, right? This is why I'm so careful to say, hold on, tokenomics is not about token costs, right? Yes, there are token costs in there. In the same way of saying like, FinOps is not just about your instance,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Speaker 1</strong><time class="lt-turn__ts">58:32</time></p>
    <p class="lt-turn__line">right?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">58:32</time></p>
    <p class="lt-turn__line">Right, and and or if you if you're writing with your taxi driver, Uber driver, your mom, and they say, oh, you're in cloud, you you do the the storage from the iPhone, right? Your your photo. Well, no, cloud's not about storage either. It's a total cost, right? And so we've got to think about in the same way Fenops got there with covering all different things, thanks to IT FM and ITAM and all the other disciplines. And tokenomics needs to really think about that whole thing and also not lose the learnings from all these other parts of the organization that have fleshed it out. And I would say that was my biggest. When people ask, like, why did we split out tokenomics? I was like, we actually didn't split out tokenomics. That was happening in organizations. We were hearing from FinOps leaders at giant companies who were owning all the infrastructure that there was another part of the organization that was going and reinventing basic principles of accountability and consumption management in other AI areas, and not leveraging all the the knowledge that we've all done collectively over that last 20 years of this, right? And so, tokenomics needs to look at all of that. And it turns out there's some new stuff also.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="shoulders-of-giants">Standing on the shoulders of giants <time>59:35</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">59:35</time></p>
    <p class="lt-turn__line">Yeah, in a way, it's we're almost standing on the shoulder of giants. All the work we've done around ITEM, FinOps, it's not being thrown away, right? What we're really saying is that maybe it's a call to arms to say all these disparate functions that we're all individually working on improving, they kind of come together with AI. It's a unifying force to say in order to solve how do we maximize the value of AI in our company quantifiably. You need all these functions, like the Avengers of different teams of ITEM, IFI, you know, finance, procurement, all these role players that have always existed. But now there's a shared mission of okay, how do we enable AI in our company without bankrupting us and without limiting us from and falling behind?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:00:16</time></p>
    <p class="lt-turn__line">Yeah, right. It's</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:00:17</time></p>
    <p class="lt-turn__line">it's such a fine balance you have to find in the middle. Where to your point about the pendulum swinging back and forth that you know as models get cheaper, as we get out of this mindset of like oh no we're we're spending too much, I have a feeling that as we start showing the value that's being created because right now it might be unquantifiable and the fear of the unknown is like well we're spending 50% more every single month on AI and our budget from a year ago means that we already hit our limits. But if I showed you the value that you actually got out of it, all of a sudden you say, let's double down on this. Let's find places to invest more money into this versus other places to grow our company because we can see the revenue getting bigger because of it. We can see the operational efficiency because of it. And I think a lot of the the fear is not so much around you know should we use AI or not. We know we are going to use AI. The fear I think is often that we can't really tell whether we're using it properly, right? Yeah, which starts with</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:01:11</time></p>
    <p class="lt-turn__line">visibility and accountability and those things. It's part of that. You know, you did kind of lose me when you mentioned Avengers though, because I was starting to think about like which discipline would be which Avenger.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:01:20</time></p>
    <p class="lt-turn__line">Yeah, I think Hulk would definitely be. Nobody wants to be the Hulk. Yeah, exactly. But yeah, in a way, I think this conversation I remember at Finopsex for the last couple of years was like even Gartner had mentioned. Well, where's ITAM and FinOps going to be? Right. I remember. I think you talked about. They were one of the first ones that said</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:01:38</time></p>
    <p class="lt-turn__line">they would come together. Exactly. Yeah.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:01:40</time></p>
    <p class="lt-turn__line">And they laid out like three different potential things that could happen. ITEM absorbs FinOps, FinOps absorbs whatever, or they merge or whatever. And to me, it's like this is a point where it won't merge because they're different disciplines. We realize that it's all the same discipline.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:01:55</time></p>
    <p class="lt-turn__line">It and it's it's we got to stop the same discipline too. And this this was one of the things. So we had this executive forum that we started as the Finance Foundation, and then you know we brought in ITAM and pre tokenomics, and and they were all about talking about how all these different things fit together, and they had all these every every time we do an executive forum, there'd be like 10 topics, and then around like March of this year, they're like, yeah, we don't care about anything, anything except the scaling AI costs, and and as we started to say, well, is that you know ITFM or there's the TBM flavors or is it ITAM or SAAM or ITSM or FinOps or pick your acronym? These leaders who are getting the push from the board were like, we don't we don't we don't care like like don't ask us to have any more conversations about where this sits. We we just want to like show the value from the things, right? And and amazingly, if you think about some of these conversations we're having at that point, if I go ask a a global CIO, hey, who does that FinOps for you? They were like, okay, there's this team, and they sit in the platform engineering team or their you know infrastructure team. I ask that same. This is an actual example. I can't say the company, but you know, they gave us the FinOps person who does tokenomics. Like, well, I got to introduce you to the the CFO team. I got to introduce you to the chief AI officer team. I got to introduce you over here. There was a much broader reach, right? And and I think in that aspect of what we heard in the executive forums, we're seeing a real integration of technology into the business now. Like every every company is now a technology company. Banks are technology companies, and AI is making that so. So we can't think about any of these things in individual disciplines anymore.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:03:28</time></p>
    <p class="lt-turn__line">Yeah, I think AI kind of brings them together in that I need this level of data and accountability from use, say, in FinOps, right? Because often I find that what what gets ignored about FinOps and AI is that if you have an agent, right, and we think about efficiency in FinOps, say like a database or VMs or Kubernetes, is that we want to make sure we're only consuming what we need, and utilization is a good metric for it. And so imagine a world where your Kubernetes cluster now scales up 1000 times in a month. The team that owns it will say, "Well, yeah, we're efficient though. It's 80% utilization. We're scaling appropriately, and all of a sudden you realize, "Okay, well, the cost went up, but what do we get back in return? It might be some rogue agent that someone created says, "Hey, I need to scale, or I need to call this thing 1000 times a day. Right? This individual team is not getting that bill. Yeah, the one that is actually creating the cause of it. The ownership lies with this platform team, and they say, "Well, we're scaling appropriately. What do you want us to do? We can't shut down someone else's server, right? And the team you go to that owns the server says, "Well, we're not the one consuming this. We're providing a service in our company. And so often I find that even just looking at AI spend oftentimes is an educational lesson, and well, what AI spend is is a culmination of these four different things.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:04:46</time></p>
    <p class="lt-turn__line">So true, so true, and and I think if you look try to token maxing,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:04:49</time></p>
    <p class="lt-turn__line">yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:04:50</time></p>
    <p class="lt-turn__line">that was. I mean, token maxing is a great example of if you just set a target on a single metric, you're going to get the wrong output. You know, if you set a a target on, you know, using. A hammer, you're going to have a lot of smashed windows in your office, but you're going to crush that hammer metric, right? And and that's the same thing with Kubernetes utilization metrics. Is like great, it's highly utilized, but it's being utilized for things that don't actually add value. And this is where we've got to break out of disciplines and silos around technology and connect it back to the business.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:05:16</time></p>
    <p class="lt-turn__line">Out of assume that you know between the tokenomics there, so there's a creation of tokens, there's a consumption value. Which one do you think will have the biggest impact? Say in the next year, I'm part of the governing board, and there's a lot of emphasis on where do we work. What is the priority to work on? Is there one that stands out to you the most in terms of we got to get this thing right as soon as possible, and these things are really important. But if we solve this one first, out of the three disciplines-not disciplines, three</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:05:46</time></p>
    <p class="lt-turn__line">buckets, buckets,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:05:47</time></p>
    <p class="lt-turn__line">buckets of tokenomics- which one stands out to you the most, if any?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:05:51</time></p>
    <p class="lt-turn__line">I'll admit, as somebody on the governing board, I'm going to ask you, what do you think we should tackle first?</p>
  </div>
  <h3 class="lt-transcript__chapter" id="which-bucket-matters">Which bucket matters most: production, consumption, or value <time>1:05:56</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:05:56</time></p>
    <p class="lt-turn__line">Yeah, good question. I think value is where it ends, right? Because if you don't understand consumption, if you don't understand creation, it's really hard to get to the value side. But the the biggest ROI, no pun intended, I know it's probably part of the value conversation, is the thing of like the so what, answering the we did all we did the spend, and is it good and bad? You need the value side to say, well, what are we getting back in return, right? Are we more productive? Are we growing revenue because of it? Are we able to have operational efficiency? And I think again, the nice to have in unit economics and FinOps. I've talked to a lot of customers who are like, yeah, that's a 2030 thing, dude. Right now, I just want to optimize. Right now, I just want you know showback. I just want teams to feel accountable. I think with AI, it's like yeah, we can show you all the data, but we need to answer the question at the C level and the board level. The so what should we do more of this, less of this? Should we double down on our AI spend because even though it's increasing, it should be increasing more? Because look at what we're getting back in return. Right. So for me, it's value, and I think that's the hardest one, given that it's so qualitative sometimes, right? If I use a model A versus model B, and my marketing copy is way better with model B, but it's 10 times more expensive, it's a lot more homework for companies to figure out how do I connect the dots to this, right? Say, for example, marketing. I spent $1,000 on Claude to do one pagers for a conference. Not that I would spend 1000, maybe whatever. If anyone's listening, hypothetically, hypothetically, right? But to tie back to value, it's not going to be the number. It's not just the hours I saved by doing using Claude, for example. Maybe the metric is MQLs. Maybe it's how many people now, because of the copy that salient that connects, might be more personalized to every single email that I send out. You know, getting those things is not easy, right? It's a lot of on the value side. I think it's tying so many disparate events that happen to get to the actual thing. That is where I think a lot of the value, no pun intended, again is really going to come from. From tokenomics of what is a framework? If you're a SaaS company and you're sell, you know, you're serving up customers, what does value mean for you? Versus imagine a discipline that's a bank, right? You're doing a lot of a you know analysis for your grid workloads or HPC workloads. Maybe it's per transaction, but if you're retail, right, and you're doing customer support, call centers are now being more efficient, so I think value will not just be one thing. Where we thought about unit economics of well, find the divisor, right? Million dollars divided by n is your unit economics. Here, it's going to be a lot more of well, what are the different pieces to get to the end goal to define value? And there might be, yeah, step A is just getting the token price per unit level, or rather the outcome, and then it might be well that's connected to our saved productivity value. What's the revenue value? What is the x? It might be multiple things that we define as value as well.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:08:52</time></p>
    <p class="lt-turn__line">Value is absolutely, I think, where we need to focus. I I would counter a little bit that we've kind of got to do all of it because all the buckets are deeply connected, and this goes back to the numerator, denominator, whatever set up in unit economics. Is people underestimate how hard it is to get that total cost on the thing, whether it be AI spend or cloud spend or anything else. And so, there's work to be done there, right? About connecting, tracing through some use of technology to connect all the costs associated with it, and then having the conversation internally about with finance teams of like, are we rolling in the actual people costs associated with that? And I think we're also seeing on that production side the conversion of energy and capital into services. That there's a whole bunch of work there that historically was just done by a couple big hyperscaler companies that we didn't have to think about in say FinOps land. That now we're having to think more about. You mentioned open weights models earlier. We could spend a whole bunch of time on that. I think that's an important part of the mix. You know, Linux Foundation was on that. Big initiative a couple weeks ago that was about you know ensuring that open weights is part of the conversation, and that's also just one of the arrows in the quiver, right? Like a big company is going to have some weights and some frontier models and some just good old fashioned you know machine learning stuff on the back end, and so I think we've got to look at how to be more efficient in that. We've got to look at the consumption allocation optimization accountability layer. The value side is probably the most green field, I think, in a lot of ways. And we're looking to see, like, are are there consistent things we can do there on the value side, consumption to value, because some companies have similar measures. But there's also going to be industry specific stuff underneath that we have essentially a fourth working group, a fourth bucket, if you will, production consumption value underneath, which is the definitional terminology, basic core frameworks, personas, who's doing it work that needs to be developed, and with all of this to tie it back to the fourth part of our little group of the tech value umbrella at the Linux Foundation is there. There's the focus specification, which is actually the only piece that I would say is now standardized because there's all the hyperscalers use it, and you know a bunch of data generators. But what what do we need to put into the focus spec to carry through some of these AI concepts? Yes, token spend concepts, yes, LLM concepts, but also there's a real challenge right now where where people are procuring the same piece of AI. In this case, we could say a token or something else, but from multiple different places. So I might get you know if I'm using I might get my clod from four different sources right now. Might be directly, it might be resold, it might be credit card enterprise, it might be you know used under the covers by some other service that is reselling that to me, and so this is why it all fits together with the tokenomic supply chain, if you will. Is that if anything goes awry at the data center level, this doesn't just cascade into one company's pricing; it cascades into every company's pricing because it's also interconnected, right? And so, I think we talked about the blind men and the elephant, and them all having a different view of it. The other elephant metaphor here is how do you eat an elephant? You have one bite at a time, right? So we're spending up different working groups to each work on each of those. But but at the core of all this is education, right? First and foremost, let's get aligned so that when the board member goes to the CEO, and the CEO goes to CFO, and the CFO goes to everybody else, and says, "What do we do about this tokenomics thing? We all need to make sure we're talking about the same elephant, right? And we're not talking about one component of it, and then figure out the responsibility, right? Who who needs to do which parts of it? Because yeah, I mean, token costs matter, but that's a tiny slice of it. Model selection matters, but that's a tiny slice, and then all the other work that all these other parts of the organization are doing matter. The other thing we haven't talked about is another term. I always wait to see how long it is for somebody says that to shift left.</p>
  </div>
  <h3 class="lt-transcript__chapter" id="shift-left-shift-up">Shift left, shift up <time>1:12:52</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:12:52</time></p>
    <p class="lt-turn__line">Ah, yes, my favorite friend, shift left. Yes.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:12:55</time></p>
    <p class="lt-turn__line">So, what's your view on on shift left and tokenomics and AI and how that's changing things?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:13:01</time></p>
    <p class="lt-turn__line">You know, I think what's really happening is shifting left. I think we're really what we've done is what we talked about last year, which is we're shifting up, right? And it was a big problem that we had. Where, you know, as a finance practitioner, you may go to the room and you start talking about savings plans and efficiency ratings, and the Cxo is like, tell me that this complete different thing, right? Tell me about the thing that I care about, which is well, I have to forecast for the next three years. Where's the investment going to be? I think with AI, it started at the up level, right? There's no need to shift up because it started up here, right? The CXOs are the one putting pressure to say, I need this level of granularity, I need this valuation, I need this X level of data from all the spend we're having, and it's not us trying to, as a finance practitioner, trying to convince a CXO to care about as bottoms up. It was bottoms up exactly. Dynamics</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:13:50</time></p>
    <p class="lt-turn__line">has been top down,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:13:51</time></p>
    <p class="lt-turn__line">right? So you know what I think with shift left, it's the promise of shift left to me meant that the right data to the right person at the right time, so you can actually make a better decision, and that means something different for every. It wasn't just you know shifting left of engineers get the data so they can make better decisions. It was also procurement. They could figure out which SKUs have they have to discount. It could be product teams to get the data of, yeah, in a non-prod environment, this new feature we're launching is too expensive, right? So shifting left was always giving you access to that data, this rich amount of data that we get every single hour from the cloud service providers and what have you, and gleaning out of it the information that's relevant to the right person, and you know we got there a little bit, but to me with AI, it's it's not only shift left, it's also shift up at the same time, and it's the speed of when we have to enact these things isn't going to be like a five-year horizon, it has to be done yesterday, right?</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:14:44</time></p>
    <p class="lt-turn__line">Yep.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:14:44</time></p>
    <p class="lt-turn__line">So a question for you is really compared to when you started the FinOps Foundation, how's it been different this time around with the Tokenomics Foundation?</p>
  </div>
  <h3 class="lt-transcript__chapter" id="launch-speed">Launching amid more speed, more noise <time>1:14:55</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:14:55</time></p>
    <p class="lt-turn__line">Yeah, I mean that's sort of a cliche answer, but this the speed has been really.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:15:00</time></p>
    <p class="lt-turn__line">Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:15:02</time></p>
    <p class="lt-turn__line">crazy. You know, look. Last week we we did the official launch, and when we launched the sorry the FinOps Foundation, like people who are in this space took note when we launched Tokenomics Foundation, and we were in some of these articles, not others. But in the in the week that we launched, the BBC, the New York Times, the Wall Street Journal, Fortune. Every everybody covered this stuff. I mean, I got more outreach about the New York Times article than you know anything. And people who I hadn't heard from in 15 years reached out and were like, "Oh, tokenomics thing. This is really cool that you guys are doing this. FinOps never hit that level because it was about the platform layer. It was about the technology, and and so is every other discipline in this area. So the the speed and interest has been really cool to to see and watch, right? That happen, but but it also introduces the challenge of just how do you how do you move fast enough as a foundation who's around consensus based open source development to have an impact, right? And I think these are the to bring in a third elephant metaphor. That's the elephant in the room: is like how do how do we how do we how do we do standards and spec work when the ground is shifting so suddenly? And so my my view on that is there's there's three parts to the approach, right? There's the we've got to get out in front of just getting information out, like what we're doing right now. Information radiation conversations about this. You know, you said, "Well, why don't you guys have a weekly newsletter? You know, earlier I was like, "Yeah, we need to do that. We put out that info, and everybody's getting that because there's model changes, there's AI pricing changes, there's copper pricing changes, and power pricing changes that affect all this all the time. So that's the first thing. The second thing is we're going to do medium-term working groups that put out fundamental best practices. I'm not saying specs because specs entry is the whole thing we get to in a minute, but how to think about this, and that's likely going to take the form of yes, the boring white papers that nobody actually reads because everybody summarizes them now, but it's going to take the form of education. It's going to take the form of explainers, but it's also going to take the form of skills files and you know agentic enablers and things that we can drop a bit of knowledge into that you can access with your cloud and ask the questions. The longer term is the specification standards work, and this gets into the focus spec. It also gets into look. Everything's moving super fast, but you know, somebody asked me when I leaned into this, are you sure tokens are even going to be around in a couple years? And it's like, yeah, I know everything's moving quickly, but there's some core fundamental concepts that we all need to align on, right? And I'll repeat for the fifth time, tokenomics is not about token costs, but there are some underlying primitives like tokens that exist there that you're using to decorate with a lot of other pieces of spend to get to total cost of whatever it is. There's long-term work we need to do. The focus spec is in its 1.5 edition. We're looking at 1.6 right now. How do we normalize the language around all of this, and how do we get everybody putting out their billing costs, uses data in a way that you can get to that unit economic conversation? But longer term, I think we're all sitting, and I don't tokenomics foundation is going to change the world, but I but I would hope to make the conversations easier for the CFO team, the CTO team, the CIO team, the procurement team, the product team, the AI team, and the organization, so they can at least use the same language, and at least when they look and approach this elephant, that they can have the context right to do that. But it's tricky. I mean, to how it's different, it's it's moving faster. There's more interest, which is great. It's exciting, but it also means we have less time to to figure it out.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:18:39</time></p>
    <p class="lt-turn__line">I think the Tokenomics Foundation mascot has to be an elephant. Now we've we figured it out. You know, every year Finov sacks Tokenomicon will have an elephant ride. But we announced</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:18:49</time></p>
    <p class="lt-turn__line">Toki, who's the Tokenomic cat, right? The Tokenomic cat. So now we I don't know if Tokenomics elephant is.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:18:55</time></p>
    <p class="lt-turn__line">You know, it's a to your point though. I I think tokenomics itself, it's not about one small thing, right? It's not about the tokens. It's not about X. It really is how do you get value from AI in a way you can justify</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:19:09</time></p>
    <p class="lt-turn__line">tokenomics is how do you get value from AI.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:19:11</time></p>
    <p class="lt-turn__line">Yeah,</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:19:12</time></p>
    <p class="lt-turn__line">and that's different, I think, than FinOps for AI, which is how do I efficiently tune, optimize, forecast, allocate that service, right? I want to efficiently use my clod. I think that's one of the lines, and this is a blurry area. Whereas the tokenomics aspect is touching much more into the business case. It's also going. It is a bit T-shaped, right? It's going further out into the product conversations and the margins and the earnings. It's going left into the data center and the capital efficiency and and investments in energy, but it's also going deeper into like we have a lot of principal engineers involved in the Tokenomics Foundation now. Ben Ops is sad in engineering a lot of ways, but it often didn't get down to the layers of BLLM and PyTorch and KB Cash and these like deeper quantization. This is also an area that these folks and their work they're doing at those levels, which a lot of that has been in the Linux Foundation more broadly for a while, is now intersecting with a CFO conversation in a way it never has before.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:20:12</time></p>
    <p class="lt-turn__line">Yeah, CFOs start talking about distillation and quantization, and you're like, whoa, right? Very different than you know anything we ever saw in FinOps. There, I think we're almost at time. So to wrap it up, I love asking this question to everyone: What is your favorite use of AI that isn't a standard use case that you love?</p>
  </div>
  <h3 class="lt-transcript__chapter" id="favorite-ai-use">Favorite unconventional use of AI <time>1:20:31</time></h3>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:20:31</time></p>
    <p class="lt-turn__line">I'm really bad about following up on things. I have a very ADHD brain, which which is great for connecting disparate concepts, but it's it's not great for following through things down to a deeper level. So, I mean, maybe this is a standard use case, but I'm I'm a big brain dumper into it, right? And give me the things I need to follow and chase. Like, if there's one thing that's I think had the biggest impact for me, it's it's ridiculous, but it's like just like the calendar management aspects of how do I take these six things I need to chase down, align it against what I'm doing, and then make sure I get the right outputs at the right time. I don't know what's one for you. I feel like this is what I hear everybody else is doing too, but maybe I haven't come up with that unique one yet.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:21:15</time></p>
    <p class="lt-turn__line">No, it's it's similar, but I live in New York, and I find that every time I type something to AI. I get in editing mode. I want to change stuff around. I want to give. So what I found is, if I go for a walk, I'll go voice mode, and I'll just say I'm going to brain dump an idea. It's unstructured.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:21:31</time></p>
    <p class="lt-turn__line">Similar. It's</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:21:31</time></p>
    <p class="lt-turn__line">word vomit. Exactly. And at the end of it, what I find is that you know, if I explain to a friend or a colleague about some idea, one, they don't have the patience to listen to me for an hour, rambling about some concept, but I think talking through stuff also gives me a lot of clarity about these salient concepts that, at the end, they just click. But it took a lot of work to get there. And what I find is like, if I can just talk for an hour at something and say, "Your job is to listen. Don't give me 1800 tokens of output of like that's a great idea, whatever, and then start challenging my assumptions, right? What did I miss? What am I? What's the elephant in the room, right? That I have haven't looked at or touched upon, and oftentimes I think the clarity of thinking that I'm able to get by doing this is incredible, right? So a lot of times when I, you know, I'm doing anything around marketing or product or even even this conversation, I'm like, what should we talk about? Yesterday I was in California, just I walk around for like half an hour, brain dumped my ideas, and I find this this ability to actually listen, connect the dots for me, or point out things that I you know everyone has biases, but be able to actually do that without having to put pen to paper and think about I have to write an essay that's going to be a you know skills file or context file. No, I just I just talk. The gift of gab, you know. So I find that to be so useful for me to actually structure my thoughts in a much more efficient way than having to you know write paragraphs and paragraphs and delete and edit and everything. Yeah.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:22:54</time></p>
    <p class="lt-turn__line">Yeah. So now that everything's going to be watermarked, you're going to have to be more more careful about that. Exactly.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:22:58</time></p>
    <p class="lt-turn__line">Yeah. Well, Jr. Thanks so much for spending the time today. Really happy we could make this work.</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:23:02</time></p>
    <p class="lt-turn__line">Really fun. Thanks. I think I</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>Amit Kinha</strong><span class="lt-turn__role">Field CTO, DoiT</span><time class="lt-turn__ts">1:23:03</time></p>
    <p class="lt-turn__line">reached out to you about two weeks ago and said, "Hey, I'll be in California. He said, "I will be there. I'll drive</p>
  </div>
  <div class="lt-turn">
    <p class="lt-turn__speaker"><strong>JR Storment</strong><span class="lt-turn__role">exec. dir, Tokenomics Foundation</span><time class="lt-turn__ts">1:23:07</time></p>
    <p class="lt-turn__line">up to LA, and we'll make it happen. Awesome. Thanks for having me, man. Great conversation.</p>
  </div>
  </div>
</div>