ServiceNow's $7.75B Armis deal is a lesson in what a CFO refuses to fund

ServiceNow has completed its $7.75 billion acquisition of Armis, and its president and CFO has used the deal in Fortune to lay out a test for allocating capital under AI pressure ^[1][2]. The useful part is not the purchase but the accounting behind it: the CFO describes the deal as a judgment that closing the gap between asset visibility and cyber risk mattered more right now than half a dozen other initiatives competing for the same dollars ^[3].

That is the whole argument compressed into one sentence. The hardest investment decisions are rarely between a good idea and a bad one, according to the column, but between many good ideas, each backed by smart people, credible data, and a case for urgency ^[4]. Trillions are being spent globally on AI initiatives ^[5], the list of worthy internal requests grows every quarter ^[6], and capital remains finite: money going to one area is money not going somewhere else ^[7]. If your AI request list exceeds your budget, the binding constraint is not idea generation. It is your willingness to end funding for things that would work.

The first triage filter offered is the moat: whether an investment strengthens what is hardest to copy about the business ^[8]. The reasoning is specific rather than abstract. When AI can produce functional code in minutes, a feature advantage can be matched by a competitor in weeks, which raises the bar for what deserves funding at all ^[9]. That pushes spend toward AI paired with proprietary data, hard-won expertise, and systems built over years ^[10]. The column cites JPMorgan Chase building its LLM Suite in-house and wiring it to the firm's own data and systems ^[11], and points to ServiceNow's own claim of 20-plus years and more than 100 billion customer workflows as the base it is compounding ^[12]. The candid admission is on build versus buy: even organisations with a strong build-it-ourselves culture must now be open to inorganic deals that bring in capabilities and talent faster than internal development can ^[13]. A $7.75 billion cheque from a self-described organic innovation machine is that concession priced ^[1][13].

The second filter is customer demand, with a falsifiable version attached. Customers reported fragmented AI efforts running in parallel with no central visibility or governance, and that feedback led to the AI Control Tower, a central hub for managing AI across the enterprise ^[14][15]. The stated red flag is the inverse: if you cannot trace a direct line from a customer insight to a major investment, treat it as suspect ^[16]. Applied honestly, that kills a lot of internally sponsored work. The column advertises three questions, though the text available here breaks off inside the third ^[17].

What to watch: whether ServiceNow ever names the half dozen initiatives that lost to Armis, because a triage framework is only demonstrated by the list of things it stopped. Watch too whether Armis capability shows up as differentiated pricing rather than a feature checkbox, and whether AI Control Tower customers confirm the governance gap it was built to close ^[15].