Yeah the title of this article is a pretty serious fail.
They aren't trapping light in the crystal, they are converting light into atomic spins and then back. They aren't stopping the light, they are converting the information to another form.
To the people asking about power storage: very very little power could be stored. Currently we have the ability to alter the spin states of nuclei. This is actually easy to do, we use the magnetic properties of electromagnetic waves (light). If we do so in an external magnetic field, we can store potential energy since we have induced "order." As the nuclei that have been pulsed by light relax in the external magnetic field, they in turn release light which reduces their "order" and induces current in a detector coil. This (very simplified) is the basis of MRI (which is really Nuclear Magnetic Resonance Imaging, but we Americans are afraid of the N word) and NMR (Nuclear Magnetic Resonance). The point being, the excitation energy is dramatically higher than the released energy. This energy gap is one of the major challenges of instrument design.
In short, atomic coherence cannot store much energy.
As a sidenote, this is also why they cannot store the information for long. The scientists have put the crystal into a higher energy state, which spontaneously relaxes through all the available means. When the crystal is transparent, it relaxes by emission of photons. However, there is not only one method for relaxation. In the article, they use the term "fizzles." Nuclear spin flipping is another mechanism of relaxation.
If anyone wants to read the original paper, it is available free here:
http://physics.aps.org/featured-article-pdf/10.1103/PhysRevLett.111.033601Edited by avesdude - 7/26/13 at 12:27am