The volume of geostationary orbits is theoretically zero, it's a 2D circle. It's true that most applications can work within some tolerance around that mathematically perfect orbit, and it's also true that volume of all orbits at all inclinations near geostationary altitude is huge. I'll grant those caveats, but I don't think it's at all sensible to waste the best parts of those orbits, especially given the deorbit timelines. It's one thing to gamble with Kessler syndrome that will last a couple decades and cut off our generation from LEO, another if it lasts for millennia and cuts off our civilization from GEO.

The volume of geosynchronous orbits is massive, but because geostationary orbits are limited to a single plane the available volume for them is quite small.