What you describe is the standard paradigm in cosmological physics, so it has been studied extensively. The basic consequence of dark matter not having significant nongravitational interactions is that it has no way to shed its orbital energy. Dark matter particles that cluster due to gravity cannot coalesce into disks or compact objects, because they retain the orbital velocities that they acquired upon first infall into a system. Thus, dark matter remains in the form of hot, diffuse "halos".
By the way, this remains true even in places where ordinary matter is also present, since the dark and ordinary matter do not interact with each other nongravitationally. The formation of disks and compact objects in the ordinary matter does not qualitatively alter the distribution of dark matter.
Also, dark matter is not unique in this behavior. Due to their tiny cross sections, stars also effectively only interact gravitationally. So the structures of dark matter halos are similar to those of globular star clusters and gas-poor elliptical galaxies, only differing due to their different initial conditions and the possibility of gravitational few-body interactions in star clusters.
More detail on the dark matter distribution
Since two-body interactions are negligible, dark matter systems do not thermalize, so their structures are tightly connected to the initial conditions. For example, material accreted earlier settles into lower orbits while material accreted later settles into higher orbits, which implies that the radial distribution of material of a halo is tightly connected to its assembly history. Let $M_\mathrm{acc}(\rho)$ be the mass accreted by the halo when the density of the universe was larger than $\rho$, keeping in mind that the density of the universe drops over time. Also, let $M_\mathrm{halo}(\rho)$ be the halo mass inside radii for which the average enclosed density exceeds $\rho$. Ludlow et al (2013) noted that
$$M_\mathrm{halo}(800\rho)\simeq M_\mathrm{acc}(\rho),$$
that is, material accreted when the universe was denser than $\rho$ settles into regions denser than $\sim 800\rho$.
Despite this, the structures of dark matter halos tend to be pretty consistent.
- The mass distribution is pretty close to spherical.
- At each position, the distribution of particle velocities is pretty close to isotropic, although this becomes less true toward the edge of the system (where radial or tangential velocities might be favored, depending on the halo's current accretion rate).
- As a function of radius, the mass distribution is reasonably well approximated by the NFW density profile. The density of a halo diverges toward the center of the system, although it's expected to converge to a finite value at radii smaller than what is resolved in simulations. It decreases at larger radii.
- The gravitationally orbiting system ends at roughly the virial radius, where the enclosed density is $\sim 200$ times the average density of the universe. Outside this radius lie still accreting particles that have not completed full orbits yet.