I wrote a piece on stack exchange on this about what would happen if dark matter (DM) were not particle. In this I discuss some aspects of Verlinde's recent proposal for dark matter being a form of spacetime curvature. The hypothesis has a ways to go before it might be considered the primary theory. In effect the competitors have to be eliminated.
Dark matter does appear to occur outside of ordinary matter. The Einstein lensing of the Bullet cluster illustrates this. Galaxies that have collided have luminous matter that coalesced after the collision, but there the DM halos of the galaxies have continued to move apart. Luminous matter due to electromagnetic and contact forces will tend to "glue together" while the DM that interacts only by gravitation has no "stickiness" and the DM continues on its merry way. A picture below is a composite of the Hubble telescope measure of the gravity lensing on either side combined with a Chandra X-ray image of the ordinary matter of galaxies in the middle. (More below)
This data was thought to put the end to MOND, MOdified Newtonian Dynamics, which had a change in $\vec F~=~m\vec a$ for extremely small accelerations. The physics of DM and galactic halos made of DM is not tied to the ordinary matter of galaxies.
This has lead to a search for DM particles, but that search has turned up empty handed. Most theories have been for weakly interacting material particles (WIMPs), which are electrically neutral and interact by the weak interaction or weak neutral currents. The leading contender has been, or maybe better put had been, the neutralino. This is a condensate state of the supersymmetric partners of the Z weak interacting gauge boson and the Higgs particle, so called wino and higgsino. However, this has come up empty on a number of fronts. The first is the LHC is not finding any hint of low mass or TeV interaction supersymmetry. The searches for DM, which involve very sensitive detectors that measure tiny perturbations from a weak interaction, such as ZEPLIN and XENON etc have found nothing of any statistical significance. The next generation of ZEPLIN will be sensitive enough to effectively rule out WIMPs if nothing is observed. The final problem is that $\gamma$-ray observations reveal no signal from a putative DM annihilation, say if the neutralino should be a Majorana fermion that self-annihilates.
Verlinde has then jumped the gun so to speak and considered the possibility that DM is not at all due to particles, but is instead due to emergent gravity. His paper makes some interesting arguments, and I think this is something to keep at least on a back burner. This may indicate that something might radically change in our thinking about nature.
I will hint at what that might be, and this touches on some work I am doing. Let us think about the Unruh effect, and as well Hawking radiation. This physics implies that what is a vacuum in a non-accelerated frame or flat spacetime is a vacuum plus particles in an accelerated frame or curved spacetime. Hawking radiation suggests a relationship between spacetime physics and elementary particles. We also have the hypothesis that gravitation is emergent from quantum physics, such as the work of Raamsdonk. This has carried further into the ideas of Susskind and Maldacena that entanglement is equivalent to the physics of ER-bridge or nontraversable wormholes. This is called ER = EPR these days. These physical hypotheses indicate that maybe spacetime is equivalent to condensates of quantum entangled states, fields and particles. So maybe Verlinde's spacetime picture of DM is in some way dual or equivalent to a particle theory of DM.
The new physics that is emerging may mean there is a fundamentally different sort of relationship between spacetime and particles than we currently think. The QCD/gravity equivalence of Bern et al may indicate the graviton is an entangled pair of S-dual gluons that interact very weakly. Hence a colorless gluon pair has physics identical to what might be called a "strong graviton." Now enter the ER = EPR, where this entanglement may involve ER-bridges. Hence it might be half way reasonable to think of quarks, or quarks with S-dual QCD that is very weak, in a state where their QCD color lines of force are confined in an ER-bridge. This would be a relationship between gravitation and particle physics not currently explored. Of course this could be wrong, but a connection between spacetime and particles might indicate that Verlinde's idea has some particle interpretation to it.