Since the 1965 power blackout in North America, power companies have made huge efforts to prevent and mitigate cascading power outages in power grids but catastrophic blackouts continued to happen in many countries of the world. At present, ever-growing penetration of intermittent renewable resources in an interconnected power grid will further increase the complexity of the grid, change its dynamic characteristics, and bring more uncertainties and challenges to daily grid operations. It is important for the operator sitting in the grid control room to be aware of the accurate real-time margin to instability for the current grid condition and any foreseen credible disturbance. However, a real-world power grid is an extremely high-dimensional nonlinear network system, so fast determination of the accurate stability boundary of the grid in its state space is fundamentally difficult. For instance, a US power company typically builds the mathematical grid model having 50,000+ nodes and lines modeled by nonlinear algebraic equations and 5,000+ generators and other dynamic devices modeled by high order nonlinear differential equations. Solution of the resultant whole set of nonlinear differential-algebraic equations (DAEs) can be time-consuming for any given single disturbance. One of today’s best industry practices in online power system stability assessment simulates predefined 1000-3000 contingencies every 5-15 minutes based on the system state estimated from a real-time state estimator but still have a big gap to truly real-time simulation. In fact, the iterative mechanism of traditional numerical methods for solving nonlinear DAEs is a key barrier to faster stability assessment. Moreover, in the next 5-10 years, a 30%-50% high-penetration of renewables and other intermittent energy resources is foreseen in many power grids, which may further stress transmission networks and bring much more complexities, uncertainties and instabilities to real-time grid operations. A power grid will experience more frequent disturbances and more diversified operating conditions than ever and may have to be modeled by stochastic DAEs to extremely increase the difficulty of real-time solution. The power industry and the research community have been looking forward to new technology enabling “faster-than-real-time” stability assessment and closed-loop control against blackouts. The presenter will share his visions in this field and introduce potential approaches to faster-than-real-time stability assessment and control: 1) wide area measurement system (WAMS) based grid stability assessment; 2) non-iterative semi-analytical solutions of power grid DAE models; 3) application of high-performance supercomputers together with techniques on spatial and temporal partitions of the power grid DAE model for faster computation.