Current research in spintronics focuses on a new type of memory device where the information is stored in the resistance state of magnetic tunnel junctions. This device generates high expectations, since it combines most qualities encountered separately in existing random access memories, among which one of the key features is the non-volatility. The present writing scheme of MRAMs, based on Ampere’s law is not robust against downscaling. The solution under investigation, based on the spin-transfer torque between the free and fixed layer of the tunnel junction, solves the scalability problem but has other shortcomings. Its limited reliability and endurance derive from the fact that writing the memory element requires applying high current densities through the tunnel barrier. In this context, I propose a novel approach where, instead of using the spin-transfer between two ferromagnets, the magnetization is reversed by transferring angular momentum from the crystal lattice. During my post-doc I have discovered that Spin-Orbit effects can be exploited to create a torque on the magnetization and switch it. Since this new writing scheme uses in-plane current, parallel to the barrier, implementing this novel approach allows circumventing most of the remaining difficulties. Besides, the potential of this line of research for creating knowledge is tremendous. It is well known that the Gilbert damping is a torque caused by non-equilibrium spin-orbit interaction induced by the magnetization dynamics. Similar non-equilibrium spin-orbit interaction, this time created by the electric current, produces a new type of torque. I believe that the balance between fundamental and applied research promoted at SPINTEC offers the perfect environment for my project. While rich physical phenomena are still likely to emerge, the recently demonstrated magnetization reversal induced by the Spin-Orbit effects allows taking the first steps toward prototypes for magnetic memory applications.