How to get budget for an AI project

Why is my approved AI project still unfunded?

Because approval and funding are separate decisions, made by different mechanisms on different calendars. The budget was locked months before your pilot existed, so there is no line with its name on it, and “great idea, let’s do it” commits nobody’s money. This is the default fate of good mid-year proposals, not a quiet rejection – though it becomes one if you leave it there. The repair is to treat the yes as the start of a second conversation: before you leave the room, ask which budget it comes out of. The sponsor’s answer – or their hesitation – tells you which of the funding paths below you are on.

Where does mid-cycle money actually come from?

Three places, in the order you should try them:

• Reallocation. Your sponsor moves money from a line that is underperforming inside their own budget. Fastest path, because it needs no one else’s consent – it just needs the sponsor to want your pilot more than whatever that line was buying.
• Underspend. Somewhere in the organization a planned initiative is running late and its money is sitting unspent. Late in the year this is the easiest money in the building: finance would rather see it produce something than watch it get clawed back. Ask your sponsor, or finance directly, what is running behind plan.
• The opex framing. A monthly subscription plus a few staff hours is operating expense; a capitalized project is governance. $500 a month and $6,000 up front can be the same money wearing different clothes, and only one of them routes through a committee. Where the accounting is honest either way, choose the framing that keeps the decision small.

How do I size an ask that clears without a committee?

Every organization has a threshold below which spend needs only a signature, and above which it needs a process. In most mid-size companies a manager or director can sign somewhere between $5,000 and $25,000 on their own authority; formal review tends to wake up above that. Find your sponsor’s number – ask directly, or look at the largest thing they approved recently without escalation – and size the pilot’s first phase under it.

The pattern is not new: one operator told us a $29-a-month tool would “not move the needle on our budget at all” and would need no business case whatsoever. Budgets have altitude bands, and spend below the band never triggers the machinery. Two cautions. Size honestly – a first phase with its own metric and exit is a real decision, while six invisible phases of the same project is salami-slicing, and finance has seen it before. And keep the ask inside one budget: a $15,000 pilot funded by one sponsor clears in a conversation; the same pilot split across three departments becomes a negotiation with three vetoes.