There is a shift in where the money comes from that most go-to-market motions have not caught up to. For fifteen years, the ceiling on what any software company could sell was some version of the software budget. That is no longer true for AI. As one closely-read analyst put it this month, the addressable ceiling for AI is not the software budget anymore, it is the headcount and consulting budgets, because AI is being bought to do work that people and outside firms used to do. Gartner's numbers point the same way: total IT spending is growing about 14% this year and the software line was just revised down, while AI spending is running up nearly 50%. The money is not appearing in the place software money used to live.
It is appearing in the operating budgets of the functions whose work is being automated. That single fact rearranges who a partner should be selling to. If the budget for an AI agent comes out of the support organization's cost line or the finance team's headcount plan, then the person who owns that budget, and owns a number it has to move, is the buyer. Not the central IT function that owns the platform. The last article was about how to price the outcome. This one is about who actually buys it, and the answer is increasingly the function, not the firm.
The Budget Moved, and the Buyer Moved With It
The old enterprise-software motion sold a horizontal platform to a central technology buyer, who then rolled it out to the business. That made sense when the thing being sold was infrastructure that every function would eventually use, funded from a shared IT budget. AI bought to do a specific job does not fit that shape. Its budget sits with the function feeling the specific pain, and its success is measured by that function's own metric, not by a platform-adoption dashboard.
Selling the horizontal platform into central IT now lands you in the slowest and most crowded buying center in the company. It is where every vendor is pitching, where the decision is about architecture rather than a result, and where the outcome is a proof-of-concept that stalls. Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027, and the ones most likely to die are exactly these: technically interesting, owned by no one who feels the cost, attached to no number anyone is accountable for. A project a function leader is running to hit their own target does not get cancelled the same way, because someone's result depends on it.
Sell the Function's Number, Not the Firm's Platform
Function-led selling means choosing a function and selling the workflow consequence for that function, priced against the metric its leader already owns. Not "an AI platform for the enterprise," but "we will take your support-resolution time down and stand behind it," sold to the person whose review depends on support-resolution time. The offer is scoped to one function's outcome, funded from that function's budget, and measured by that function's number.
This is the horizontal cut across the market. A function like customer support, finance operations, revenue operations, or IT service management exists in almost every company regardless of industry, which means a play built for one instance of it travels to the next customer with little rework. That repeatability is the whole economic advantage of going function-first. It is a different axis from concentrating by industry, which is its own decision and the subject of the final article. The point here is narrower: whatever industry a customer is in, the door you knock on is a function, and the key is that function's number.
Land in One Function, Then Take the Neighbors
The motion is to land narrow and expand along the seams (the land and expand approach reworked for AI, where agentic workforces that deliver on an identified metric are added in place of seats or modules). Win one function where the outcome is measurable and the budget-holder feels the pain daily, prove the result, and then move to the adjacent functions that share the same data and process. Support borders billing, which borders collections, which borders finance. A partner that has proven itself in one of these has a warm path to the next, because the reference is not a slide, it is a number the neighboring function's leader can already see.
The strongest recent examples of AI landing inside a company follow exactly this shape. The most talked-about internal deployments this year did not begin as a company-wide platform decision; they started in one function, threaded AI through the tools that function already used, and became the connective tissue between neighboring teams one at a time. The lesson for a partner is that you do not need to win the enterprise to get in. You need to win a function, and let the result open the next door.
Why This Reaches the SMB-to-Corporate Market
Function-led selling is not only viable in the SMB-to-Corporate market, it is the natural motion there. In a small business up through roughly $1 billion in revenue, the head of a function is often the budget owner and the decision-maker at the same time, without the layers of central procurement and architecture review that slow an enterprise deal. The person who feels the pain can say yes. And because central IT in these companies is thin and stretched, waiting for an IT-led platform strategy is waiting for something that may never come, while the function leader has a number to hit this quarter and a budget to spend against it.
That also protects the partner. A deal anchored to a function's owned metric, funded from that function's budget, is not a discretionary experiment that dies in the next cost review. It is tied to a result someone is measured on, which is the same durability the whole series has pointed at from different angles. The buyer who owns the number is the buyer who keeps paying.
The Common Thread
Every article in this series has come back to the same place: value follows accountability for a result. Go-to-market is where that becomes a question of who to call. The budget for AI has moved out of the shared software line and into the operating budget of the function whose work is changing, and the buyer moved with it. A partner still pitching a horizontal platform to central IT is selling to the office that owns the least urgency and the slowest yes. A partner selling a specific outcome to the function that owns the number is selling to the person who has both the budget and the reason to spend it. Find the function that feels the pain. Sell the result it is measured on. Everything else is knocking on the wrong door.
Part of a series
This post is one part of the Insights series, our post-by-post working through of the thesis the 2026 Channel Forecast sets out in full.
Sources
- The argument that AI's addressable ceiling has moved from software budgets to headcount and consulting budgets: Jamin Ball, Clouded Judgement (Altimeter), August 2026.
- IT and AI spending growth, including software-line deceleration against ~47% AI-spend growth, and the forecast that more than 40% of agentic AI projects will be cancelled by end-2027: Gartner, 2026.
- Illustrations of AI landing function-by-function and becoming connective tissue across neighboring teams rather than a single platform rollout: company go-to-market accounts and general trade coverage, treated as directional.
- The migration of agent monetization away from the seat toward consumption and credits, which reinforces that spend is tracked to a function's usage rather than a firmwide license: company earnings coverage, treated as directional.
