In 1990, Moore's Law was still in full swing and every year brought drastic changes to the personal computing landscape. These days, we aren't seeing a lot of entirely new classes of applications driving local data storage needs. We've long since passed the point where text and audio data could add up to enough disk usage to worry about. Video data volume is largely stagnant, because optical discs are basically dead and local storage lost to streaming (in spite of the lower bit rates for streaming). Photos can add up to something on the order of a terabyte if you're prolific and don't use cloud storage for that.

But video games are by far the most common and easy way to burn through hundreds of GB of storage in a consumer PC. Except not every PC user is in to AAA gaming, and all of that storage capacity usage can be freed up quickly and easily when free space starts running low—again aided by cloud storage and fast internet speeds, because games can be re-downloaded with relatively little inconvenience.

Probably the biggest new class of computer applications with heavy data storage needs to crop up in recent years is generative AI. But models like Stable Diffusion are relatively small compared to mainstream consumer storage capacities, and bigger models like GPT-3 and newer are exceed local compute and memory capacity by a far wider margin than they exceed local storage capacity, so they're stuck in the cloud for now.

There are real reasons to expect consumer storage capacity needs to grow more slowly than they did in the 1990s, and there's real evidence that they have been growing more slowly in recent years. Just pointing at a trend from several decades ago without even trying to think through whether it's still relevant is not a useful observation or insight.

When I upgraded my laptop from a 1TB HDD to 2TB of flash I thought it’d be easy to fill it. I was wrong. It’s been more than two years since and my next laptop probably won’t have more than 1TB of flash.