By CJ Combs, Sr. Sales Director for Data & AI at Columbus
The companies still talking about AI in 2027 are the ones still piloting it. The ones that actually deployed it are the ones who stopped saying the word because it disappeared into the function it serves.
As AI moves out of experimental pilots and into core business operations, it’s time to treat it like any other tool in the business: something with an owner, a cost and a measurable job to do.
Let’s unpack what that means for your budget.
The hype slowly settles
AI hype has been driven largely by the fear of missing out (FOMO). Leaders see competitors announcing AI initiatives and feel pressure to announce their own, whether or not it’s attached to a clear business problem.
Much of that FOMO is fed by the market itself: the IPO pipeline, data center build-outs and the marketing behind both. I expect the hype to fade once those IPOs close and the data center boom levels off. Until then, there’s the hype, and there’s the reality, and they won’t line up for a while.
In the meantime, leaders can’t wait for the market to calm down on its own. They need a discipline that cuts through the noise today.
AI becomes an operating expense
Most organizations split spending into two broad buckets. Capital expenses are big, one-time investments, like building a data center. Operating expenses are the ongoing costs of running the business, like salaries, software subscriptions and electricity.
For the last few years, most AI spending has been an operating expense on paper but managed like a special project: a standalone “AI budget” controlled centrally by IT or an innovation team. That structure was the right fit for the experimental phase, but once AI is doing real work every day, it gets much harder to justify.
In 2027, I predict AI will stop being a central budget line that gets charged back to the departments it serves. It will become a real operational tool, and each department will own the cost directly. When the accounts payable team uses an agent to process invoices, that department will pay for it. When customer support uses an AI agent to resolve tickets, they own that cost.
This is the crux of AI operationalization: the process of moving AI from a demo environment into the daily workflows that run a business. And once AI is running daily, the costs will look very different. You’ll see agents running around the clock, workloads spread across multiple models and token consumption (the usage-based charges AI providers bill for every chunk of text a model reads and writes). Those three things will turn AI into a CFO conversation.
The other wrinkle is that for most business tasks, leading AI models have converged enough in capability that picking one is the easy part. Swapping one for another is largely a configuration change. The real money goes to integration, workflow redesign, governance and monitoring, all of which continue after launch. AI is an operating expense with a long tail.
Value receipts end pilot purgatory
Many teams are stuck in pilot purgatory: pilots that never reach production and never officially shut down, so they keep consuming budget. McKinsey’s 2026 State of AI survey found that nearly nine in ten organizations use AI regularly in at least one business function, yet only 37% attribute any impact on earnings to it, essentially unchanged from the year before. Usage is close to universal; returns haven’t moved.
This can be remedied by defining target outcomes and budgeting for them. Every project gets a defined value target at the start. Here are two tools that help:
1) Cost-per-outcome numbers. Before approving anything, require a clear figure for what each unit of work costs: each invoice processed, support ticket resolved or contract reviewed. That figure becomes the foundation for AI budget optimization, because you can compare it directly with what the same work currently costs.
2) Value receipts. Think of this as a running record of what the AI system did, how much time it would have taken a human to do, and what that time is worth. A value receipt makes ROI visible and captures value that shows up early, like faster cycle times, lighter workloads for staff and other capabilities the business didn’t previously have. Headcount is the slowest and most political number to move. A receipt shows progress long before it does.
Every AI agent and governance program gets an owner
The basis of AI cost accountability is a named owner for every agent and a business line that pays for it. That owner also funds the data work up front, since an agent is only as good as the data it can reach. Governance needs the same treatment in 2027: a named owner and its own budget line.
Many agents run on the permissions of the person who built them, including access the agent never needed. Because their actions are logged under that person’s name, there’s no clean audit trail of what the agent decided. Business users are also connecting their own agents to company systems through integrations nobody approved, because nothing required approval. It’s like letting someone through the front door without a background check. Whoever connected the agent owns the risk, and every document it can reach becomes part of the attack surface.
One of our clients in Jamaica addressed this by pairing a governance program with an AI lab where employees could experiment and bring ideas. A central team reviewed every proposal and worked out how to run it safely, and business teams had a say before any project moved into production. Central oversight, departmental ownership: they had both in place before agent sprawl became a problem.
Stop saying “AI”
Enterprise AI adoption is entering its grown-up era. To prepare for 2027:
- Retire the standalone AI budget line and move AI costs into the budgets of the departments it serves.
- Require a cost-per-outcome number and a defined value target before approving any project.
- Track value receipts so ROI is visible month by month.
- Assign a named owner to every agent in production, and give governance its own budget.
- Audit the permissions your agents have inherited and what each one can reach.
When all of that is in place, people stop calling it “AI.” The finance team just calls it their close process, and it runs faster.
##
ABOUT THE AUTHOR
As Sr. Sales Director for Data & AI at Columbus US, Christopher (CJ) Combs helps organizations lead with AI and data to solve real-world business challenges. With over 25 years of experience in AI, ML, and automation, CJ is known for bridging strategy and technology to deliver clear, measurable outcomes. He partners with enterprises to design and guide AI initiatives that cut waste, accelerate progress, and integrate seamlessly with existing systems while maintaining security and compliance. A trusted advisor, CJ brings deep expertise in large language models, copilots, and custom AI solutions, empowering businesses to stay competitive and future-ready in the digital era.
