In my opinion, it is similar to Parkinson's Law [1] about work filling into the time given to it, but replace work with bureaucracy filling the aggregate enterprise revenue on offer. Atlassian's work, one might argue, is "done" but has such cashflow from business customers that they can continue to spend above and beyond what is needed to maintain what has been built to service their customer base. They could be 37signals/Basecamp, but they are enabled beyond that (from the business customer cashflows mentioned), and so these actions occur until an innovator comes along to replace them (and potentially, the cycle repeats due to enterprise sales cycle durations, inertia, etc).

You see this with all manner of large enterprise in my experience, where what they continue to do is "good enough" to allow for these inefficiencies and actions because they are, on some spectrum, "money printers" due to their moat and inertia. Creative destruction is not a forgone conclusion, nor fast. Is the incumbent exploring the problem spaces adjacent to their core business(es) to increase their TAM to increase shareholder value? Are they innovating? Or are they just churning and burning up revenue on meaningless work?

All of these companies doing layoffs to invest in AI is not about AI specifically, it is about reaching for profits and yield in a challenging business landscape and macro post zero interest rate policy ("ZIRP") imho. They are desperate for productivity growth, whether that is doing more with less people, AI, offshoring, whatever because money now has a cost.

[1] https://en.wikipedia.org/wiki/Parkinson%27s_law