Deciphering the nature of dark matter has great scientific importance. A leading hypothesis is that dark matter is made of Weakly Interactive Massive Particles (WIMPs), which may result from supersymmetry or additional spatial dimensions. The underground search for elastic scattering of WIMPs on suitable targets (the so-called "direct detection") is currently led by the CDMS II experiment. Its sensitivity is ten times better than any other experiment and we hope to obtain another factor ten in the coming two years. After a brief recall of our recent results (a journal club talk was recently given by Jeff Filippini), I will describe the complementarity between direct detection experiments and LHC and ILC. In order to address this rich physics we need to move rapidly towards a new stage, increasing the target mass from 5kg to 25kg. Coupled with moving to a deeper location at SNOLab this would increase the sensitivity for direct dark matter detection by another order of magnitude in the next 5 years. There are no other fully demonstrated technologies that can reach this goal within that time scale. I will finally outline the role that LBNL could play in this SuperCDMS program. While the 25kg stage requires only modest extrapolation in technology, a forceful R&D effort is necessary in the area of detectors, cold and warm electronics to extend the target mass to 150kg or even a ton. The goal would essentially cover the whole supersymmetry region of parameter space. I will compare this approach to other technologies which on the horizon.