If all objects were in different positions on the same precise orbital track, they would not interact. Over time, they will drift into different orbits, because they will be perturbed at different times by lunar and solar tides. Still, while they may gain enough relative velocity to disable each other during a (rare) collision, they probably won’t produce hypervelocity shrapnel.
The problem is that all satellites aren’t in the same orbit, they have different radii and arguments of perigee and ascending nodes; an orbit is described with six parameters, which are shifted by tides and the equatorial bulge and drag. Soon enough, a constellation of satellites all in different and highly inclined orbits to provide approximately complete coverage (perhaps between 60N and 60S) will shift - without stationkeeping thrust those orbits will eventually overlap. Sooner or later, a satellite orbiting northwards will encounter another satellite orbiting southwards, making a spray of shrapnel that can damage other satellites. Look at the wikipedia page about the 2009 Iridium Cosmos collision for the most prominent example of such a collision.
No satellite should be launched if it cannot actively change its orbit, if only slightly to avoid a collision. If a satellite is due to lose that capability, it should be deorbited or recycled. After it smashes into a thousand fragments, those fragments are a thousand times as expensive to collect and deorbit. Without that, in a few centuries the exponentially increasing fragment cloud will form a Saturn-like ring, and space becomes inaccessable.
If centuries seem like a long time, imagine some careless twit doing something a millenium ago that makes your own life impossible. The tenure of earth life in the universe may span billions of years - or we may turn it into a pool of maximum entropy in a thousand. I would rather see the lazy entropy maximizers subject themselves to the treatment first.