September is associated with football, foliage, flannel and, for some, the Financial Plan.
As we put pen to paper (or agents to harnesses), there’s a few core elements that have always driven the P&L outlook for the following year: rep productivity and new product releases driving new sales, expansion and contraction against the install base, employee roster changes, and discretionary spend.
For each one I can name the driver and the owner and tell you how far the number moves when the assumption changes.
AI doesn’t typically have that simple correlation of input to output. It also doesn’t live in a singular place on the P&L. In fact, AI has the potential to impact every line in the P&L: AI cost to serve influencing pricing structure (Revenue), delivering AI-native products & features to customers (COGS), engineering development and capitalizing new product creation via AI (R&D Opex), AI-driven prospecting/customer support/workflow evolution (Sales, G&A Opex).
So, the fastest growing, furthest-reaching input has the largest variability of outcomes. A bit unsettling.
More importantly, the Financial Plan is a proxy for what the company is focusing on and the associated expectations of achieving it. I’d assume your company is focusing on AI in some significant way. So, the stakes are not just the next 15 months’ financial outlook, but the probability of your company succeeding in a new era.
The AI adoption curve
Since early 2026 we have run as an AI-first company across the business, not just engineering: sales, customer success, marketing, my finance team. That shift is why the line grew into one worth explaining. And until recently, all I had to explain it with was what we had spent.
Ask me six months ago what our AI spend would be in Q3 and my answer was the run rate plus a cushion. I could not tell you which teams were driving the growth, so I couldn’t say whether it would hold with confidence. When it came in over plan, I couldn’t explain the variance with any depth, which is the uncomfortable part of this job. Numbers move, but I have to explain why.
That gap changed how decisions got made. Because I was the only one who could see the total, every AI question routed to me: a department head asking for more budget, a manager wondering whether a pilot should keep running. I was arbitrating calls about work I wasn’t close to, armed with a total and nothing underneath it.
What my team did once they could see it
Emily Allen, our Director of Finance and Accounting, went into CloudZero to find out why our AI costs had roughly doubled in the P&L. She traced most of the increase to heavy use of a premium model and that to one account development rep building a go-to-market play with it. Moving part of that usage to a cheaper model was worth $13,000 to $30,000 a month and she was able to do the analysis in 10 minutes.
The savings aren’t the only point. Emily brought me a finding and a recommendation, and I approved it in a quick conversation. That is how every other cost category in my business already works and it is the first time AI has worked that way.
She has since built her own view in CloudZero of AI spend against revenue, gross margin, and each department’s salary base. Last quarter she capitalized the AI spend that went into new feature development, with CloudZero as the support. You cannot capitalize what you cannot attribute, and a provider invoice does not know which feature it paid for.
Customer success spends far more of its salary base on AI than sales. A year ago that would have been my problem to adjudicate from a P&L. Instead Matt Katz, who runs customer success, brought me the case: his spend tripled from March to July, and he tied it to renewal rate, platform engagement, and the number of customers his team reaches. He put the return at four to five times what he spent. I did not have to take it on faith and I did not have to build it for him.
Confidence in deploying the next dollar
I can see which departments are growing their AI use and how fast. The driver is cost per unit of work. Each team knows its own: what one customer review costs Matt’s team to produce, what one close costs in sales, and how many of those each team plans to do in 2027. I can see what the money bought, not just that it was spent. Spend shows up as the work it paid for, account research, renewal planning, code review, so I know whether a team is building toward something or a job is still running after the project ended. There is an owner on every dollar, so the number gets built with that person instead of imposed on them.
That is enough to model the line, and enough to answer when someone asks what happens if it doubles.
Budget owners can advocate for their own numbers
This is the part I would underline for anyone heading into a planning cycle. Every budget owner here sees their own AI numbers on the same basis I see mine. They are not hearing their own consumption from me for the first time in a quarterly review, and they do not need me to translate it.
So a department head asking for more AI budget for 2027 brings me a case I can evaluate in the same terms we use for everything else. Some I will fund, some I will not. Either way I am making a decision rather than picking a side.
What this changes for my 2027 plan
We released new AI Signals capabilities today, and Scott Castle, our chief product officer, has written about what they are and why we built them. The new Overview dashboard in AI Signals gives finance and other leaders a single picture of AI cost and usage across every team, model and provider in the company.
The cost of AI is now something I can plan and something I can hand to the people accountable for it. I can measure AI against revenue, margin, and payroll because Emily built that view.
Where I would start
If you are building a 2027 plan around an AI line you cannot break down, you are in good company. We surveyed 260 senior finance leaders this year, and companies were twice as likely to overspend their AI budget as to hit it.
Before you commit that number, find out where you stand. Visit our website to request a demo, read more, or see a short video highlighting how AI Signals works. It takes a few minutes, and it will tell you whether your next planning cycle starts from evidence or a cushion.
