More often than you might think: low latency on interactions, loads, responses, etc is a high priority for e.g. social media sites and online marketplaces, where fast, smooth interaction measurably increases user retention and spending ([link](https://www.youtube.com/watch?v=x2EOOJg8FkA)).

Also the case for desktop applications: lower latency on input directly improves user experience and satisfaction. Some applications directly market based on this: half of the Zed editor's marketing is 'coding at 4k 144FPS'.

Though you might not immediately think of it as 'software' gamedev is one of the most performance-focused regions of the industry, especially in graphics, especially especially for studios building or customizing their engines. Reducing memory consumption lets you comfortably increase scope, making the engine more efficient increases your frame count. A bad framerate on console/mid-range PCs alone can be the touch of death for a project.

In cases where the software isn't end-user facing, performance is still often a concern even when there isn't a strong real-time constraint, not just because of temporal latency, but because slow software means more time spent running the software, which means paying for more servers or a higher cloud bill. A badly optimized core cloud pipeline can devour an org's budget, where a well-optimized one can run for peanuts.