Care Needed

Enterprise software is due its own “Wait, Wut?” moment as regards AI. There is going to be a Sorting Hat moment in the coming months as the dawning realization hits that Not All Enterprise Software Is Created Equal when the world gets more automated.

Broadly, we can think about enterprise software that it scales with “processes”, by which I mean both business processes - workflow, basically - and also server or other device processes like “monitor print queue”. The more processes that run at any one time within the enterprise, the greater the demand for software. In a growing economy with more and more stuff getting automated, it’s a given that unit demand for software is going to rise.

But some processes require humans and some do not. The “user login” process is driven by the number of logins per user per minute; the “monitor storage array availability” process is not at all driven by the number of users in the building or beyond.

As AI eliminates users, so too will it eliminate the number of required processes within enterprise software. It will create new ones, but since those new processes won’t be driven by users either, it is crunch time for any software company that sells predominantly on a per-head basis.

For yucks, I asked an LLM whether Dynatrace’s licensing model was human-driven. Computer said no.

Dynatrace Licensing Model

These systems monitoring companies - DataDog which just joined the S&P500, Dynatrace, Splunk and Appdynamics (the latter two both now owned by Cisco Systems) - are the types of software & services players which can win as the body count shrinks in the enterprise. Cloudflare ($NET) is another.

For me as I consider which software names I want to own and which I do not, this is a key test. Headcount-driven, bad. Complex automation driven, good.

And with that let’s look at the numbers, valuation, chart and rating.

Headlines

This is what a maturing software business looks like. Margins climbing, cash pile rising, revenue growth slowing.