Beyond finding ground states, new applications have recently emerged for Variational Quantum Algorithms (VQAs). These include dynamical simulation, metrology, solving linear systems, and principal component analysis, among others. Moreover, new tools are making VQAs more feasible, such as quantum-aware optimizers and error mitigation methods. VQAs also exhibit an intriguing resilience to certain types of noise. These signs point to VQAs soon providing quantum advantage for practically interesting problems. On the flip side, new analytical results have studied the gradient scaling for VQAs. These results paint a concerning picture and show that barren plateaus in the training landscape will emerge if care is not taken. Barren plateaus are caused by lack of structure or simply by the presence of noise. Avoiding barren plateaus has now become a crucial area of research. This talk will review both the exciting and challenging nature of VQAs for near-term quantum computing.