The global energy transition necessitates sophisticated planning tools to navigate the complex interplay between policymakers and market agents. Bilevel optimization provides a natural framework for modeling this hierarchical decision-making process, where a central authority (leader) sets policies, such as carbon taxes or renewable energy subsidies, and energy producers (followers) optimize their investments and operations in response. However, the resulting large-scale mixed-integer bilevel problems are notoriously difficult to solve with monolithic approaches. This paper proposes an exact decomposition method, based on the principles of Benders decomposition, specifically tailored for strategic energy transition planning. The method iteratively refines a master problem, representing the leader's decisions, by generating optimality and feasibility cuts from a subproblem, which models the follower's market response. This decomposition approach breaks the problem into smaller, more tractable components, enabling the solution of realistic, large-scale instances that are intractable for conventional solvers. We formulate a generic bilevel model for long-term energy system expansion and present the detailed algorithmic framework of the proposed decomposition. The methodology is validated through a comprehensive case study of a national energy system, analyzing optimal transition pathways under different policy scenarios. The results demonstrate the computational efficiency and scalability of the proposed method, highlighting its capability to provide crucial insights for policymakers by quantifying the impact of strategic policies on the evolution of the energy mix.