Starlink is just under 600km in order to target a specific voluntary threshold of 25 year decay. Most of their finished birds are going to make around 5 years without propulsion or impacts.

Three things about that -

Profitability generates competition, which may or may not respect precedent. Right now Starlink is only really worried about collisions with other parts of Starlink. We cannot afford a Starlink-inspired future to happen at 600km.

Debris generating events can spin out an object with much higher cross sectional mass than an intact satellite. Think of it in terms of what we use drag for on Earth, like a kid building kites. A heavy metal bolt works worse as a kite than a long thin panel.

These calendar decay timelines are blind to density. If there are a billion satellites with a natural lifespan of 5 years flying at 551km then they are going to go into an exponential cascade in a matter of weeks. If you plan to launch very large constellations, you need very fast decay timelines to keep that safe. It is much safer at very low altitude. There is decay 'room' for >10x as many satellites at 300km as at 400km, and >10x as many at 400km as at 500km, and >10x as many at 500km as 600km.

...

It would also be nice to set aside something for manned spaceflight. Unlike with a satellite collision, if a pressure vessel gets penetrated everybody dies and nobody wants to go back. The ISS and Tianhe are going to have to deal with debris risk slowly raining down from a collision at 971km.