Gartner says AI agents put $234 billion in software spending at risk. Read the fine print before you cancel anything.

On July 1, Gartner said $234 billion in enterprise application software spending is at risk from agentic AI by 2030, roughly a fifth of what companies pay for SaaS. The reposts stopped reading there. The report did not.

If I only handed you that number, I would be selling you something. So here is the case against acting on it, made as strongly as I can put it.

First, Gartner itself refuses to call this an apocalypse. Its own framing is that SaaS "will not be destroyed; it will emerge in a different form." A change of shape, not a funeral. Second, and more damning for the hype, Gartner separately predicts that more than 40 percent of agentic AI projects will be canceled by the end of 2027, on rising cost, unclear business value, and weak controls. Third, by Gartner's own read, most products with the word agent on them are "agent washing," ordinary software relabeled, with only a small fraction doing anything a real agent does. Put those together and the honest summary is this. Agents will reshape software spending eventually, most agent projects attempted today will fail, and nobody should rip out working systems this quarter to chase it.

I agree with all three points. So why do I still think the $234 billion number should matter to you.

Because the number is not really about agents. It is about a bill you already pay and have never examined.

Strip the AI framing out completely and the underlying fact survives. You pay per seat for software that does a fixed set of things. Some of it earns the price. A lot of it does not, and most companies have never sat down and separated the two. You are paying for seats nobody logs into, a tier you turned on for one feature two years ago, and two tools that overlap because different people bought them in different budgets. That waste existed before agents, and it will still be there if 40 percent of agent projects die exactly on Gartner's schedule.

This is the bet behind MojoSoftware, and I want to state it carefully, because the skeptics above are right that hype kills projects. Our line is: why rent when you can buy. For a narrow, well-understood problem, a custom tool you own outright can replace a stack of subscriptions that grows every time you hire. You own the code. You own the data. The bill does not climb with headcount.

The catch, and the skeptics earned this, is that custom is often the wrong answer. Rebuilding a mature product that already fits you is a waste of money. Betting your operation on an immature agent because a report said $234 billion is the exact mistake Gartner is warning against. The skill is knowing which of your subscriptions are worth replacing, which are fine to keep, and which problems are not ready for automation at all. That judgment does not require believing any particular AI forecast. It is just arithmetic and honesty about your own operation.

Gartner handed the software industry a frightening number and, in the same breath, three reasons not to panic. Both halves are true at once. The move is not to cancel your stack, and it is not to ignore the shift. It is to find out, concretely, which of your monthly bills is buying you something real and which is just habit.

If you want a straight read on which subscriptions are worth replacing and which are not, that is a good first project. Send us your stack and we will tell you what we would actually do.