Procurement's Next Crisis: Who is Paying for All This AI?

There is a massive power struggle happening in corporate boardrooms right now. Current research reveals a staggering disconnect: while 87% of companies are actively increasing their AI budgets by 2026, a mere 14% have established a clear owner for AI within the C-suite.  

This leadership vacuum has created a multi-billion-dollar tug of war over budget, risk, and strategy. In this week's Proc n Roll, hosts Conrad, Natasha, and Zachary sit down to debate the biggest question in corporate technology today: Who exactly should own AI in the enterprise?  

Here is a breakdown of the core debates, the hidden costs, and the looming risks discussed in the episode.

The Great Debate: Innovation vs. Anarchy

The battle over AI ownership often boils down to a clash between top-down governance and bottom-up innovation.  

• The Case for Decentralization: Zachary argues that AI ownership shouldn't be bottlenecked by a single executive or committee. True innovation comes directly from the employees who are doing the daily work and looking for ways to improve their own processes. Treating AI like everyday software—where the user is held accountable for the final output, just like a PowerPoint or Excel model—gives teams the freedom they need to innovate.  
• The Case for Centralization: Natasha strongly opposes the decentralized approach, warning that it will lead to corporate anarchy. Without a centralized strategy, different departments will fall completely out of sync, creating isolated "toy" agents rather than integrated business solutions. She argues that unguided AI adoption is a recipe for compliance disasters and operational chaos.  

The Looming Threat of "Shadow AI"

Because many enterprises are moving slowly to approve AI tools, employees are taking matters into their own hands.

"I honestly think that in a couple of months, a shoe will drop. One company, one employee, someone sitting somewhere in the basement... will unintentionally spill some secrets, some data." — Natasha  

Conrad highlights that this "Shadow AI" is already thriving. Employees are actively bypassing corporate restrictions by "swivel-chairing" to their personal devices or external hardware to run AI models for their daily tasks. While this bottom-up pressure forces companies to innovate, it also introduces massive, unmonitored risk to the organization.  

The "Token Tsunami" and Rising Costs

One of the most overlooked aspects of the AI boom is the hidden cost of operation.

• Tokens are the New Currency: Zachary predicts that AI token consumption—the micro-costs incurred every time an AI reads, writes, or generates content—will quickly become the single largest expense line for global companies.  
• Procurement is Asleep at the Wheel: Currently, technology teams often monitor this usage, but they do not manage the financial optimization. Zachary notes that Procurement must step up and treat AI token costs with the same rigor as Software Asset Management (SAM).  
• Questioning the Value: Natasha adds that when Procurement reviews new AI contracts, they must ask tough questions. If a company is paying for a new AI tool, leadership needs to clarify what legacy expense or system that AI is actually replacing to ensure costs don't simply pile up.  

Dividing the Responsibilities

If no single person can own all of AI, how should the enterprise divide the responsibilities? The hosts mapped out a cross-functional approach to keep companies safe and optimized:

• Security: The Chief Information Security Officer (CISO) must own the data boundaries and security policies.  
• Cost & Value: Procurement must own the budget, token consumption optimization, and vendor negotiations.  
• Overall Strategy: The Technology leadership (or Chief Transformation Officer) should own the broader strategic vision and implementation.  

However, the question of who actually writes the overarching AI Policy remains a hot potato. Ultimately, companies that foster strong cross-functional collaboration will thrive in this new era, while those operating in isolated silos will inevitably struggle.  

Transcript: Proc-N-Roll | Procurement's Next Crisis: Who is Paying for All This AI?

Conrad: Everybody, welcome back to Proc and Roll, your guide to practical procurement. Today I am with the Three Amigos again, Natasha and Zach. We got a great topic last week... We got to this big question: who owns AI in the enterprise? There's this billion-dollar tug of war happening in the boardrooms. As of 2026, the research shows that 87% of companies have increased AI budgets, but only 14% have a clear owner in the C-suite. This is creating a lot of issues, a lot of questions, a lot of uncertainty. So the question we're tackling today is, who is the owner of AI in the enterprise? Which of you want to take a crack at putting the red pill up, and then we'll take the blue pill on the other side?  

Zachary: Everyone owns AI in the enterprise. AI is only as good as the people using AI and deploying AI. I've heard of a lot of companies that are setting up these Chief of AI offices, and their jobs are to help kind of the rest of the business stand up pilots and take the best ideas and funnel them. You're not going to get AI to work that way because ultimately, if you actually want real change to happen, it needs to come from the people who are doing the job today.  

Natasha: So then, Zach, who should Chief AI report to?  

Zachary: If there was a Chief AI officer, they should report into the CEO because everyone else has got a different agenda. The CEO is the one who needs to set the AI strategy.  

Conrad: Zach, I got to ask, what does AI even mean in that context? Is it efficiency? Is it transformation?  

Zachary: It's not efficiency, right? I mean, it can be used for efficiency, but it can be used for all kinds of things, it can be used for creativity. That's precisely the point, is everyone needs to use it. So can it really be owned and the use case decided by and governed by a Chief AI office? No... I say every budget holder should have a budget for AI.  

Natasha: I'm falling away from Zach yet again. Because I think that if it's completely decentralized, then how are you going to drive compliance? How can you make sure that your guys in the distribution center somewhere in the field or in the plant will not create their own purchase orders? If someone is falling massively behind, then they will be out of sync with the entire enterprise. So I'm not a big fan of decentralization. I am a fan of centralized strategy and autonomous implementation, but under some set of rules and guidance. In the absence of a Chief AI Officer, if the company has a Chief Transformation Officer then it should fall under their responsibility, or Chief Strategy Officer. If there is no chief strategy and no chief transformation, then I would say it absolutely has to go under CPO.  

Zachary: The best people who are using AI the best that I know are honestly not using it on their corporate machines... outside of work, they're subscribing to all these tools. They're testing the boundaries. If companies don't give their employees the frontier models, let them get tools that are really working in their industry as quickly as possible into the business... the risk is some people get ahead of others in the business, and you don't want people falling behind.  

Conrad: I see the same thing... you go back to most procurement organizations and ask them how they're using AI, they'll talk about generative AI, maybe copilot... but at an enterprise level they're opening the doors to these things more cautiously. Because it's not like there's no risk here. It's so easy for a hallucination to find its way into a document, into a contract, into a relationship. I'm not surprised and I'm not sure that I have a problem with the enterprise moving slower than the garage.  

Zachary: Look, don't strip away the accountability. Let's talk about people using AI in their day-to-day role... you can stay accountable for your outputs. You use Microsoft Word, Microsoft PowerPoint, Microsoft Excel today. If you're building a model in Excel, let's say, and you present it to the CFO, the expectation is that you own that model. Keep that same kind of accountability, keep that same kind of expectation, but let people do what they must to do the best job they can.  

Natasha: I think developing building agents without strategy makes no sense. I see companies building agents, investing a lot of money and a lot of time... and it's like two agents dropped in the middle of the field, and they do something that has no connection to the entire operation. It's cool, it's sexy, it's slick, it's fun. It's like a toy, but it's a toy. It does not provide a holistic solution. I am still of a belief that strategy should come first.  

Zachary: Maybe I think I'm converging on two things here. So one is you using AI in your day-to-day... you set up 12 prompts that really help you do your job better... I think that layer is pretty safe, and that layer is pretty easy to keep someone accountable for. But the thing that Natasha is talking about is where you have agents executing on a platform or a process, in which case I would totally agree. If you have a P2P platform, you can't have one manager create an agent and another manager create an agent that does something similar, but they're not exactly the same. That would be a disaster. You definitely want some more governance, you need rules.  

Conrad: As I've been playing with these agents, it's easy to burn through... 50 to $100 a day in AI consumption. Who's going to manage the budget and allocate the budget? Finance is going to want some control over all of this new spend that's coming in.  

Zachary: The single biggest expense line in every single company right now around the world is token consumption and AI token cost. As usual, procurement feels like it's asleep at the wheel. The people who are monitoring consumption and usage are in tech because they own the tool. But procurement should be all over the costs. It's sort of parallel to software asset management in a lot of companies.  

Natasha: Because a good procurement person, good CPO, when they see some AI contracts, they will ask questions. What are you planning to retire? If you're bringing this additional expense, what are you giving up? What is it replacing?  

Zachary: Here's where I'm at so far: Security policy for AI, CISO. Cost, Procurement. Strategy, Tech. If CISO says you're not ever allowed to put this type of data in an AI tool, then functional leadership should just live with that.  

Natasha: Who should own the AI policy? I honestly think that like in a couple of months, a shoe will drop. One company, one employee, someone who's sitting somewhere in the basement and just creating something on their own will absolutely unintentionally spill some secrets, some data. It's already happening in small things, but we just didn't have a big shoe drop with a big ripple effect. I think it's coming.  

Conrad: It is complicated because the kind of things that Zach is doing in his garage are learning and setting a standard. Without that bottoms up pressure, I think it's going to be a really slow process to get the level of innovation into AI in the enterprise that's truly possible.  

Conrad: All right, everybody, this is awesome. We'll be continued. We want to bring in a couple of experts in this space and get their perspective about how it sits. Good to catch up. Have an amazing rest of your day, Zach. Looking forward to seeing you in London here in a couple of weeks. And yeah everybody thanks for joining us. Don't forget to like, comment and share. Love you all.  

Natasha: Thank you guys, be well, bye.

This transcript has been edited for clarity while maintaining all substantive content