Currently this the best architecture and manufacturing technology for CPU/GPU.
The switch to FinFET was due to multiple reasons - but the major one is channel control.
Below 32/28 nm DIBL and other device parameters are such that you can no longer control your transistor. So if you wish to continue to scale down - FinFETs, Nanowires or any other form of 3D channel is a must.
The cost is another issue. FinFETs are costlier than planar devices - but it can range from a few percent up to 45%. 3D stacking is working quite well in NVM but for logic, it is much complex and has not been commercialized yet.
Bottom line - for logic (CPU/GPU/FPGA) FinFET is the only way to go below 28 nm node. In memory (flash particularly) the manufactures can stay on planar devices and make 3D stacking.
FinFET transistors - are the latest mass-produced, Si-based transistors - used in every major ULSI products.
FinFETs were invented in order to overcome various performance issues on Planar transistors. See the charts below:
FinFETs have significantly better performance - Vt, Speed, and leakage - due to a major geometrical change in the channel design. See the illustration below:
FinFETs are harder to manufacture and have more complex challenges (design, reliability, etc.) vs. the previous planar devices.
FinFETs are the current manufacturing platform for all major semiconductor companies. It is expected that future devices will be based on the FinFET platform.