Abstract: The energy system of mega-cities encompasses various heterogeneous distributed energy resource units (DERs), including uncertain new distributed generation units (e.g., photovoltaic, wind turbines), flexible demand-side units (e.g., air conditioning, thermal energy storage), and energy prosumer (e.g., electric vehicles, energy storage systems). Virtual Power Plant (VPP) has emerged as an effective aggregation to reshape the energy ecosystem of mega-cities and comprehensively manage those DERs. This paper begins by addressing the construction goals and development status of VPP in mega-cities. It points out the challenges current VPP projects face in dealing with multi-agent, multi-objective, and high-uncertain scenarios. Systematic solutions are lacking, including modeling, simulation, deduction, analysis, decision-making, and validation procedures. To address this challenge, this paper combines the overall digital architecture and construction progress of urban power grids, proposing a framework for the digital twin of VPP (VPP-DT). The VPP-DT framework uses virtual spaces to simulate and analyze various developmental pathways for energy systems. Furthermore, focusing on intelligent decision-making goals, we explore its technical pathways, key technologies, mathematical tools, and more within the framework. The framework has helped the implementation of the VPP project demonstration in Lingang's new area, which provides a reference for the construction of VPP at the city level. Finally, this paper looks forward to the research direction and application prospect of digital twin VPP under megacities.