AI Deserves Its Own Budget

I’ll say the quiet part first: no health-system CEO is looking to spend more on IT. It already anchors a significant share of the expense line, and most of the conversations that reach that desk are about how to spend less, not more. Margins are thin. Staffing is expensive. I know this from the inside — I’ve been the CIO trying to keep all the green lights blinking.

From that seat you also see the whole board. There is always a need to invest in new revenue streams and new programs, and always pressure to cut expense so the existing ones live long enough to reach the maturity everyone — including finance — hoped for, the point where they finally start throwing off a return. Across an entire health system, on any given day, everything sits at some pivot point. Collectively, those pivot points are the thin margin: a lot of hope that things will turn, and real uncertainty about which will turn when.

The IT budget is the wrong home for AI

IT budgets exist to keep the lights on — to run what is already in flight and aligned to how the core business functions. That is what they are for, and that is what they should stay for.

AI is not that. It is a cultural and behavioral pivot: new tools that have to be used in a new way, changing how information is accessed and distributed, and changing what roles do in a way we haven’t seen before. Fold that into keep-the-lights-on and it gets managed like maintenance — and starved like maintenance.

It merits its own budget for a second reason: it carries its own return case, tied to transforming the workforce, examining processes, and changing the way roles function. That can read as scary or as innovative. The difference is whether everyone is on deck with an open mind and a lot of new information that isn’t at anyone’s fingertips yet.

Outcomes, not impact

Wherever the dollars sit, be careful with them, and decide up front how you’ll measure what they do. I don’t even want the word impact. I want outcomes — the same discipline we apply to clinical results, applied to technology results.

And that discipline is not a pilot. A pilot is permission to stay vague. I mean an investment in one area, with specificity and full resources, to achieve a defined result. While that result is landing, the next outcome zone is already sequenced — so the work grows in a way that manages risk and keeps the returns coming in a line, not in a lottery.

Months, not years

Healthcare IT has always measured return over years. That habit comes from the EMR era, and it was honest then — digitizing a health system is a years-long project. But we’re already there. Everything is digitized. AI sits on top of that finished work, which is exactly why it can move at a speed EMRs never could. The returns shouldn’t be measured in years anymore. They should land in months.

That’s the world I lived in, and it’s the same discipline I bring now: the research to understand what’s at stake inside the four walls, the research to know what’s moving in the technology market, and the judgment to align the two — carefully, a step at a time, with the outcome named before the dollar leaves.