The Hidden Costs of AI: Why Token Prices Tell Half the Story

0:00Nearly 75 % of AI leaders, according to a recent State of Tokenomics report, have little or no confidence they can demonstrate AI's value to a CFO. Now that doesn't necessarily mean their investments aren't working, but it raises a critical question. How do leaders know which investments deserve more money? As companies expand their use of AI, they're confronting costs that extend beyond the token bill, and organizational changes need it to capture that return. Today, the Tokenomics Foundation's Executive Director, J.R. Stormont, joins the show along with Jack French, who leads our applied AI team here at WWT, to examine that gap between AI spending and business value.

0:40The Tokenomics Foundation was launched earlier this year by the Linux Foundation to focus on establishing open industry standards, benchmarks, and best practices for the economics of AI infrastructure. And its early findings point to a challenge that should resonate with any business pursuing AI transformation. connecting what AI costs to what the business actually gains, and deciding who is accountable for closing that gap. This is the AI Proving Ground podcast from Worldwide Technology. Let's jump in.

1:15JR, welcome to the AI Proving Ground podcast. How are you today, sir? Doing really well. It's been a wild 12 weeks of traveling around the world talking to people about tokenomics, But this is my first day home, so glad to be here doing this with you. Well, we're grateful that you're spending some of that time with us. And Jack, bringing you here in the studio, we're honored to have you. I know you've been traveling the world as well, so thank you for spending time with us here physically. Yeah, I know this will be fun. Yeah, thanks, Brent. Okay, well, JR, I want to start with you, you know, just from what you're seeing.

1:42Because like you've mentioned, you've been traveling the world talking about tokenomics with, you know, companies of all shapes and sizes. We've certainly all seen, you know, the headlines of AI struggling to create value. Companies are struggling to figure out. So that's not to say any of this is, you know, is not intended to be hard. But, you know, why do you think right now we're all still struggling to articulate AI, ROI, you know, we're, you know, what, five years into this journey? Well, I don't know if we're five years into this journey. I feel like the moment in time we're in right now kicked off dramatically this year.

2:17I mean, yes, a lot of us are using AI in a lot of ways for a lot of things. But the acceleration we saw at the beginning of 26, you know, February, March, April with Agentech kicking in and accelerating, you know, not just token volumes, but expectations and outputs. And I'm saying outputs very specifically to your question, because I think a lot of the noise and chatter we're hearing right now from enterprises is we are getting a lot of outputs, more code, more commits, more things. but we're not quite sure how to connect those to outcomes that the business cares about. And they're definitely struggling.

2:54And we see this anecdotally and also in our survey data with connecting the outcomes to value areas that a CFO cares about. And that's really the noise and chatter I'm hearing right now is how to draw those things together from all this AI investment. Yeah, JR, I would agree with that. It used to be, obviously, token maxing isn't where you want to go. it's not how many lines of code can be shipped it's you know what are some of the other values that maybe aren't quite as in your face like you know code shipped with less bugs that reduce an out you know an outage on black friday or you know it's productivity by a development team that maybe you weren't able to truly quantify how do we start measuring that now and so yeah to your point i i would agree i think this year it's all of the tools that that engineers and and the workforce has access to has created this shift in having to change how we see value.

3:47It's no longer really it's productivity. It's okay. What does this mean to the business and how do we quantify that in a way that makes sense? So yeah, I think that's spot on. I mean, everybody is shipping more and everybody's doing more. I think it's then we're right at that point where, yeah, how do we translate that into things the business cares about? And I think this is a bit of the existential crisis that organizations and enterprises in particular are facing is, you know, they inherently know that AI can change their business. So they're getting pushed by their boards and their CEOs to innovate more.

4:21And at the same time, they have budgets that their CFOs are enforcing. And they are trying to say, you know, how do we tie this back to things that can prove that value? And value, I think you hit on some of these, right? It's not just about revenue at the end. CFOs obviously care about that, but it is about speed to market, which I think is the most obvious one we're all seeing. And it is about a lot of staffing changes and labor conversations, right? Can we, can we do more with the same people? Are we going to do more with less? If we have more people who are AI pilled or enabled, can we do 10 times as more?

4:50Do we reskill? But then there's also, you know, capability gains and can we do new things? And so these are all the conversations we're, we're having, I think as conversations right now, but the industry and, and, and here at the foundation, we're looking to see, can we quantify that into better frameworks and then into better data schemas to really be able to prove that value confidently. Yeah, Jack, I mean, do you think that we need to settle or standardize on a kind of a new definition of what value means for the enterprise before we start moving forward? Or are we going to have to build this on parallel tracks?

5:22Interesting question. I think it's two part. One, I do think the FinOps Foundation and the Tokenomics Foundation has done a good job at trying to build a framework that we can use to have everybody looking at the same standards of, okay, here's the different costs associated with AI in environment. And let's all kind of agree in the way that that's going to flow into the organization so we can visualize it properly. But I think every organization is going to have to take their own spin on, okay, now that I can visualize and understand the spend a little bit more clear, what does that really mean for me?

5:54And how do I equate value to that? So I think it's two part. The framework has to exist that we all kind of standardize on. And I do think the The FinOps and Tokenomics Foundation is doing a great job at that. But, you know, think of worldwide technology, for example. We had a huge spike in AI costs this year. We had Tom Strunk on with you earlier kind of talking about that from our CFO's perspective. But I know for our CEO, he was okay with it because he knew that our company was getting comfortable using tools. We were all finding new ways to innovate. And it reduced kind of a fear barrier of, I don't know how to use these tools and am I comfortable talking to my own clients about this?

6:30And so that spend initially for us was big and scary, but look what it did for our organization. We're very AI forward. We've got a very comfortable workforce that's able to talk to our clients about how we're using it and innovating with it. And so, you know, how do you equate that back to cost? There's absolutely value there. Yeah. But, you know, so every organization, I think, is going to find value differently in that spend. JR, do you think that's kind of a framework or a mold that other enterprises should follow, you know, kind of understand where you want to push the boundary in terms of that token bill to see where value is being created, to see where innovation is going to be created and then react from there?

7:07Or is there a different playbook? Yeah, I mean, well, I should say with the token bill you mentioned, right, that that is an important part of the AI conversation, important part of tokenomics. But when, you know, we work with people in our community and WOT and folks, right, we're looking at the whole of AI, what we call the total cost of AI, which, you know, is the compute to support it and the storage and the networking and critically the people in labor. And that's really, I'm hitting that point again, because that's even more important in a world of open weights where companies are looking to host and train their own models, sometimes with rented hardware, sometimes with hardware in their data centers.

7:43And you all are close to the data center side as well. You know, I have noticed there's almost this, and you maybe have visibility of both sides of this based on where you sit, but there's almost this like stark line between everybody who's creating the AI, the hardware providers, the cooling providers, the energy, the data centers. they're all like let's get the tokens out and then we'll ship them out and they're going to drive revenue and but they don't really understand how the enterprises use them what use cases they go and how that drives to value and then the people receiving the tokens on the other end they don't really understand all the downstream things you know from models and vlms and quantization and and energy how that impacts the cost of the availability and all these things downstream and so there there is sort of this lack of direct tie from the creation of this ai tokens and everything else supporting it, including idle compute, which is huge now with agents all the way back to that CFO defensibility.

8:36And so, yeah, I think, you know, at the same time, I talk to enterprises sometimes and I would literally, you made a point in there, Brian, I forget the exact words, but I was on a panel a couple of days ago in Boston and this, this, this person who's a head of IT, a CTO for internal IT said, I have to remind myself that some of the smartest people in our firm are not in the IT organization. They might be outside of it in the business or other parts of it. And I want to make sure that we're enabling them with enough AI and with tokens and productivity to go innovate because we don't want it to be just us who does that, right?

9:08So there's a tension between maintaining the cost, managing the value, and also not limiting innovation by over-capping people. Jack, the cost of AI beyond the token bill, is that something that's caught organizations off guard, do you think? I know both of you have a background in cloud and FinOps. I mean, it feels like these are lessons that are rearing their heads again and should apply. Is it fair to say it caught us off guard or is it something that we kind of saw coming? That's a good question. I think maybe caught us off guard or saw coming is difficult to answer because it's probably very subjective based on the company.

9:44But I do think that this is a significant shift in the way that we're viewing consumption or financial outputs of technology. And so, you know, very similar to cloud back in the day is all of a sudden we had this consumption-based bill that came in. It's, you know, OPEX versus CAPEX and companies are trying to figure out how to manage it. And then how now we have this hugely sporadic by the day, minute, hour, you know, new consumption type of bill that organizations are battling. So it's familiar in a sense, but I think it's, you know, exponentially greater in all of the different areas that it touches.

10:20So I don't know if anybody was quite ready for the explosion of it, but it is a familiar area, I think, for people that have been through this once before with cloud back in the day. JR, would you see it differently? Yeah, I mean, it's one of those agree and disagree. I want to say yes to all those things. Also, having gone through the cloud thing, I'm having an echo in my mind of seven or eight years ago, maybe 10, where people are saying, well, cloud's not our biggest line item, but it is the fastest growing, right? And we watched that happen and people started to care about it. And I'm hearing the same conversations today well you know for a fortune 100 or something 250 like ai is not our biggest line item but it is fastest growing but the difference now is the speed i think at which it is growing for some of these companies i sat with a huge huge firm in new york yesterday trillions upon trillions of under asked management and they were like oh yeah our ai spend is doubling and i was waiting to hear the time frame was was this person going to say every six months or a year and it's every month it's doubling every month right i was like whoa so i think what what caught people off guard to brian's question around the total cost of this stuff was as well i think a year ago everybody was really like okay it's which frontier model provider are you going to use but the fable five pullback mixed with people caring about more value all this stuff introduced a lot more focus on open weights in the conversation and suddenly hardware has become cool again everybody wants to have hardware on their desktops and they want to have hardware in their data centers, their colos.

11:47And so what caught me off guard is how much we've ended up in a massively hybrid world so quickly with AI. And that's where it's going instead of just one where, and everybody's still going to use a ton of Frontier. I'm not saying Frontier is going away, but the people looking to get their AI from everywhere now. That's such a good clue in their desktop. Oh yeah. And we spent a lot of time with enterprise organizations, Fortune 100, and a lot of them, to your point, are looking at a portfolio approach. It's not just leveraging the frontier models. It's how do I leverage cloud and colo and neo cloud?

12:16And how do I leverage open weights? And, and it's, it's not just, I'm going to go all in with this one provider. It's truly, how do I, I hedge this across multiple different providers and platforms. And it is interesting to say hardware is cool again, cause you're right. And they're looking at, okay, the build versus buy discussion is back. And what do we do? Adding to that, you can't, okay, build versus buy, you decide to buy. It's really hard sometimes to go buy what you need. And, you know, as fate would have it, I was sitting on a plane last week next to somebody who was selling a part of the data center world.

12:48I won't say which parts did I find too much, but this salesperson was saying, yeah, it's so crazy. We'll give an order to a customer. And if they wait more than 30 days to fill that, when they come, by the time they come back, the cost of this hardware we're selling is, has gone up dramatically so that we won't even sell it to that. And, and they're battling with the neoclouds and the hyperscalers, everybody else for the same hardware. So there's a real scarcity aspect there too. And at the same time, the enterprises are recognizing they want to have some things they're running in-house while they also leverage the amazing capabilities of the Frontier.

13:17That's a good point. We didn't run into that with cloud. That's a good point. That's a totally different dynamic. We did scarcity in the same way. It felt like unlimited resources. Yeah, well, JR, I mean, I know research out of the Tokenomics Foundation has found that some of the heaviest users of Frontier models are planning on shifting to more of those open weights models. And you're talking about it's definitely going to be a portfolio approach, Jack. I mean, is there a right balance that's standard across the industry or is it going to depend on each and every organization? I think it depends a lot on the workload, not even just the organization, but you think of like LLMs and SLMs and open weights and the frontier models and exactly what is the workload that an organization is leveraging AI for.

13:59And I think that'll drive the decision, but there's concerns with lock-in or there's concerns with all these different aspects that go into it that I think it makes a ton of sense to have a portfolio approach, but it's going to continue to evolve quickly. And I think organizations that understand how to manage multiple endpoints like that will probably do the best. This episode is supported by Proofpoint. Proofpoint delivers advanced email security and compliance solutions that protect your organization from phishing, ransomware, and business email compromise using AI-powered threat detection.

14:32Defend your people and data from email-based threats with Proofpoint's comprehensive security platform. And JR, this is, I should also note, you know, this is not just a, oh, go to open waste because it's cheaper. That's, it's going to be adding to the total cost of AI. Absolutely. Can you explain a little bit about that too? Because I think that is a little bit of a misconception. Yeah, the total cost of AI or colloquially, as we're calling it, the TCA to match TCO. So in open weights, that's particularly important because I think there is a, I think those informed in the know don't say this, but there's a general sense out there in the media and other places as well that open weights models are so cheap, they're almost free.

15:12But, you know, it doesn't always factor in the hardware. It doesn't always factor in the training. If you're training them, it doesn't factor in the people of configuration and fine tuning and quantization and things you maybe need to look at. And there are times when spending those extra dollars on getting the immediate access to the frontiers latest and greatest are going to always trump the months of time or energy of getting cheaper tokens through running your own stuff. But it is a portfolio mix that is complicated right now by the massive cardinality of options. There are over three models listed on a hugging face.

15:46Now, those are all different versions of forests and flavors. But there's not only a ton of options in the frontier and open weights, but now we've seen the introduction of SLMs, right, and specialized or small language models and things like JEV, which are getting into the mix of like, well, can I use something that's really fast? And yes, air quotes, cheap by comparison, that doesn't have the trillion parameter knowledge of an LLM, but can help me with quick decisions. And I think what we're realizing is LMs can do everything, but that's sort of like using a jackhammer when all you need is a hammer, right?

16:16Sometimes you just need an SLM to do one thing really quickly. So there's a lot of work around quantifying the complexity of workloads so you can match those to the right type of model as well. Jack, any other hidden costs or hidden complexities when we start to go down that open weights route? I think JR hit a lot of them. And I love that TCA versus TCO. I've heard you say that a couple times now in different keynotes. And I got to remember that. I'm going to steal that from you. But yeah, I think adoption, you know, it's great if initially there's some savings that can be benefited from an open-weight model, but are you getting the same speed in answer?

16:53Are you getting the same creativity? Like, what was it that you were looking for for that model to provide? and making sure that going into it, knowing that you can get that result back as quick or as efficient as you expected with maybe the other models is the key to that. And then JR said it well, the operating model around that, do you have to layer in additional operational support, governance, security, are there other aspects that come in that maybe you wouldn't have by just consuming from a frontier model? Those are all the things that you'd want to think about. Yeah, yeah, yeah. If I could throw in one more on the TCA and complexity as well, so many of the conversations I'm having right now with leaders is about data and data preparation and data cleansing and data collection and data access, you know, the harness and the context layers, all of this stuff that needs to be done, particularly the bigger the enterprise, the more data, the more complex it is, and the more worries about where they put it.

17:43All that is like pre-time, pre-cost, pre-work, before things even hit the model. And, you know, it's, we're talking months-long effort for some of these organizations, right? And then they've got to decide, well, to your point of, you know, open weights, frontier, where are they going to put what? What are they comfortable putting in one place versus another? Jack, as we really start to, you know, explode with what we can accomplish through agents, and as we delegate more work to an army of agents, so to speak, how does that change the calculus around what we've been talking about so far? Yeah, I think it just continues to add to that cost layer, as well as the operational layer around it.

18:20I mean, this is essentially introducing all sorts of new headcount into an organization that you've got to manage. They need an identity. They need access and security. They need governance. There has to be an operating model around them. And that's all, for the most part, going to be human-driven initially. How are you setting this up? How are you creating kind of the environment for them to operate in? And how do you quantify that value, essentially? You're taking somebody and training a whole new workforce, essentially. So there's that upfront soft cost of onboarding all of these new agentic employees, but then the ongoing cost of ensuring that they're performing efficiently, they're, you know, making sure that the tasks are being accomplished in the way that you expected.

19:01And then even going back and, and ensuring that the output of those agents conforms to your company policy. I mean, who's responsible if a customer's like success agent yells at one of your clients, you know, and how do you, how do you manage that? And so that there still has to be kind of human oversight to all of that and ensuring that you're looking at that full scope for cost is important. And JR, you brought such a good point up before, the data behind it. I mean, all these agents need data and context. And so who is creating that environment to make sure that they can perform effectively?

19:35That's all costs that you've got to think of, not just the token bill of what they're consuming. Yeah. JR, I mean, build on that. Where should ownership, where should governance lie when there's so many different stakeholders at the table? Yeah, that's a really tricky question, I think, to high five Jack on his comments about the labor and changes. Those ownership questions and the labor conversations about where it sits in ownership is impacting nearly every part of the business, right? You've got parts of the business that are potentially being disrupted by AI, but then what's different now as compared to cloud is that cloud had this decentralized ability of engineers to spend company money, right?

20:19And before that, in a pure data center world, somebody had to procure something and order it and put it in a rack and all that stuff. Nobody was sneaking into your data center to install and turn on extra servers overnight in the way that cloud engineers do that. But with AI, you've got everybody in the organization consuming AI, but also spending money, company money. And so the ownership layer necessarily has to be very broad and attach a lot of areas. we are seeing in our survey data and then also anecdotally, the initial sort of governance, if you will, of this stuff is living within a technology org, CTO, CIO.

20:57When we talk about the tokenomics work, which we don't define as a person or individual team, it's all the economics of AI across the company, but they're very closely partnered with, and we see these people on our governing board for the Tokenomics Foundation, tech CFOs specifically. So there's this aspect of, you know, those tech CFOs are responsible for ensuring that all this AI usage and investment fits in their margins and budgets and hits their pricing models in the right ways. But the engineering orgs, AI enablement orgs are putting in a lot of governance and centralized accountability support methods.

21:31But the final answer of ownership, I think I'm hearing from, and I just did an interview with the CEO and CTO of the London Stock Exchange Group that I published a couple of days ago. He hit the nail on the head, which is ultimately this needs to end up and the P &L owners, right? So Central Group will provide tooling, visibility and stuff, but the business owners need to make their own decisions about what they're going to use and align against their plans. So truly shared accountability, but sitting right now between tech CFOs and the CIOCTO org. Jay, I know you speak with a lot of different organizations.

21:58That last point I think is spot on is the desire is to get that ownership of outcome out into the P &L or to the business owners or product owners. We're seeing it's not quite there yet. Are you seeing organizations be successful with that yet? What is kind of the turnover for you right now? Still, that's the dream that we're going for, or are you seeing anyone accomplish? It's a tough question. I mean, one of the things I like about the Athletics Foundation is, you know, we don't build services or products, and we don't ask our member vendors to bring in case studies. We want to hear the real-world learnings.

22:32And so, all that being said, you know, people are saying it's being passed out to the P &L. I don't know if that's aspirational or if it's actually happening, you know, but it is where it's going. I think just even the last week of meeting with a lot of FSIs across the spectrum and on the East Coast, you know, there's some that are, you know, more in the hedge fund realm. We're printing money with AI. I think, well, I just go and use it and we're going to keep track, but there's no spends and caps. And then there's some of the really big complex organizations. And I think, you know, it was widely reported a couple weeks ago that, you know, JP Morgan had set some spend caps, right?

23:08That's more the norm we're seeing is spend caps happening with this stuff as a way of enforcing accountability. They're saying it, everybody's saying it should be in the P &L owner, but a lot of the approvals for going over spend caps in these large organizations is still centralized. sometimes with a tech CFO, sometimes with AI enablement team, but they are, you know, going in and using these spin caps to have conversations about if what the AI is being used for is valuable. And sometimes it's a vibe check, like, yeah, that seems good. And other times they're asking for business cases, but it's, it's again, that healthy tension of we don't want to stifle innovation, but we also need to make sure you're not using an LLM for something that you could just be using, you know, spreadsheet for.

23:49Right. Is it Jack, based on what JR just said, I mean, to distinguish between legitimate innovative use versus just, I guess, on the other end of that spectrum, token maxing and not necessarily driving value. Is it just being scrutinous about that and asking questions? Or is there a good way that we've stumbled upon to understand how can we separate value AI versus just wasting tokens? Sure. And even to build on what JR just mentioned, am I using AI for something I could use a spreadsheet for? I think there's several scenarios that as you really start digging into it, well, this could have been just standard automation that we would normally build into a company that we would have done years ago.

24:28But you don't get to that point, to your point of the question, without asking questions and digging in to figure out exactly what organizations are leveraging AI for. There are new tools coming out weekly, it feels like, that give better visibility into that. Daily is probably right. The frontier labs are creating better enterprise kind of governance and policies and visibility into what's happening with those tools. So I think we'll see that continue to evolve where the CTO of an organization or a business leader will have better visibility into how their team is using AI. And then you can make better informed decisions on, hey, this is great.

25:03It's innovative. Thank you for thinking of a way to improve on this, but we should shift this out of this tool into something else, maybe that we build natively in the organization or this is just a spreadsheet. We just need to build some automation into it for you or something like that. So, yeah, I mean, even beyond that, JR, I mean, Jack's talking about, you know, the model providers are coming out with more and more data, but they're still offering different views, different sets. What do you think about from a standardization standpoint? I mean, is that an urgent call right now to get to some point of standardization?

25:34Or is that something that we have to just work on collectively and make a best effort? yeah that that was one of the things i was most happy to see in our state or tokenomics data that we launched at that amsterdam event that jack mentioned was that there was a real ask in the open-ended questions where we asked what do you need from the model providers there were a lot of folks saying we need standardization of reporting we need something like we have the focus spec that we've been doing for the last few years with the foundation we need something like that in that area and and that area is really just about standardizing the consistent reporting of cost and usage.

26:07And you're seeing all the frontier model providers now have, and this is fairly new in some number of months, that you can now report per user or per different categories of things. And we're watching them go through the same journey that we saw Amazon and Google and Microsoft do in the 20 teens with more granular billing data. But it's still all specific to them. And very few big enterprises are using just one model provider. And so there's a real need to normalize the reporting against them from the cost and usage. But then the next step and where we're seeing a A lot of the conversations our community go just in the last week is, OK, well, how do we connect that cost and usage data to the open telemetry data that tells us more about what it's doing?

26:44And then how do we build consistent data schemes for quantifying outcomes? We talked about the beginning of like, what is how do you codify an outcome into some code like structure that you can connect back to that? And then how do you then tie that to value and value could be revenue or could be the things we talked about labor, but it could even just be an executive's OKRs. You know, how do we say that we got to these, this outcome that is part of our annual plan faster as a result of the AI? So I think that's where we need to head toward it. And the base of all that, your question is standardizing the cost and usage and telemetry data so that we can look to say apples to apples, what does it look like?

27:22And say, well, truly, if we are trying open weights, that we are looking at, you know, apples to apples numbers, not apples to peacocks or whatever. apples to peacocks i like that that would definitely be an awkward one well i know we're coming short on time so we will wrap up here fairly shortly we're getting close towards the the end of the year which always kind of lends itself to getting out the crystal ball for what you see happening in 2027 i won't necessarily ask that because 2026 was certainly an up and down year it was an explosion of tokens it was pulling it back it was talking about caps i mean a lot of the stuff that we're talking about here what over the next you know two to three months jack do you think leaders need to focus on to set themselves up for success, knowing that the future is fairly undefined and, you know, at any point things can kind of change on us.

28:09Yeah, I would go even to what JR just touched on. Do you have an ability to properly visualize and attribute all of the different areas of AI that you're exploring in your organization right now or that are deployed? I think especially as you bring on open weights or you bring on new endpoints of AI, having a framework that you can plug it into to make sure that you understand, okay, how am I going to test this against my current metrics? How am I going to, you know, if this is my value in the organization, how will this new system or new tool that inevitably is going to come out for people to test?

28:40How am I going to measure it against what I have today? And you can't do that unless you have some sort of system in place today for visualization, normalization, and a framework set up. And it may look slightly different for each organization, But I think that's first. And it goes back to the early days of FinOps. Like, let's make sure we all understand how to visualize data so then we know how to act on it. I will absolutely not take a stab at what the AI market's going to look like in six months because it's changing too quickly. But I can tell you that making sure you have a good framework for visualization of what your spend is, how you attribute value to that, and what your real goals are for AI in your organization, I think that's standard.

29:20And JR, I know that you're as plugged in as anybody on these conversations as it relates to tokenomics. So I am going to ask you a little bit more poignantly, like, what do you expect to unfold here? But, you know, build on what Jack just said about how we can set ourselves up for success and give our listeners a little bit of an understanding of where you think the conversation is going to go as we move into 2027. Well, I'm glad you didn't ask me a question. I got asked earlier in the week by some folks not near AI, which was they asked me, where do you think it's been five years? I don't know.

29:51impossible you know the the foundations that jack laid out are right and and some of this is obviously following on the principles that we work through as a community at finops is you've got to get visibility first and that's what we're seeing with spin caps in these things and then you need to move to toward attribution as distinct from allocation attribution is who actually did the thing, not just who are we charging for the thing. But then the connection back to the business, I think, is going to be unique per organization. But I'm encouraging folks to, you know, build bridges in the org. If you've got a FinOps team, obviously work with them for the things they're good at, you know, looking at spend and visualizing and forecasting and, you know, anomaly detection of those things.

30:32But understand that a lot of the AI conversation has shifted to different parts of work and that it is really a top-down concern that is coming down right from boards and whatnot. So if you're on the ground floor of it, you want to be building up that visibility attribution, but you also need to be aligned to what the top is doing. And I guess just for individuals, I think it's so important right now to recognize that although AI may not be coming for your job, somebody else who's better at AI than you are is coming for your job. And that if you're not running ahead to keep up, you're falling behind.

31:07And all of these things we're doing, all these industries and disciplines is going to be impacted in some way. I think there's great opportunity, right, for us all to do this. But for our work is really about, okay, how do we get the right smart people in the room together to have the conversation so we can document the frameworks, figure out the specifications to help all the enterprises get better at confidently proving measurable business value from their AI investments. And so we're going to do that in the ways we do. You're going to do it in the services you provide to your customers. Everybody's got a role play in it but yeah i'd invite people to you know get involved with the journey and you know don't don't let the air train pass you by as it is moving exceptionally fast no exciting time fast moving time um and excellent insights from from the both of you in terms of you know what we can do on an individual a team all the way up to an organizational uh view jack like i said thank you so much for joining me here uh in studio and jr thank you so much for joining virtually certainly looking forward to more of the research and insights that i can see coming out of the Tokenomics Foundation.

32:04So continue the great work there. Appreciate the both of you joining. All right. Thanks to JR and Jack for joining us on the show today. It's clear that understanding the AI bill is only the beginning. The harder work is connecting that spending to better business performance and giving someone the authority to make the changes that requires. This episode of the AI Proving Ground podcast was co-produced by Nas Baker and Kara Koon. Our audio video engineer is John Knobloch. My name is Brian Felt. Thanks for listening and we'll see you next time. Thank you.