This paper investigates the centralization of block production in blockchain protocols and the impact of proposer-builder separation (PBS) on decentralization. It addresses two key questions: (1) Does heterogeneity in block producers' skills and knowledge inevitably lead to centralization? (2) Does PBS reduce heterogeneity and preserve decentralization among proposers? The paper develops mathematical models to answer these questions. In a game-theoretic model with endogenous staking, heterogeneous rewards, and staking costs, it quantifies how heterogeneous rewards lead to concentration in the equilibrium staking distribution. In a stochastic model, it quantifies the rate at which stake concentrates on the most sophisticated block producers. In a model with heterogeneous proposers and specialized builders, it quantifies how PBS reduces heterogeneity in rewards across different proposers. The paper connects its analysis to contest design, Pólya urn processes, and auction theory. It shows that economic forces generally lead to an oligopolistic equilibrium rather than a "winner-take-all" scenario. It also shows that conventional intuition around PBS breaks down if the distribution of block values is sufficiently heavy-tailed. The paper also provides quantitative predictions that would be impossible without concrete mathematical models. For example, it shows that if at least 10 block producers are at least 90% as good at extracting value as the most sophisticated block producer, then at equilibrium no block producer will control more than roughly 17.5% of the stake. It also shows that even modest decreases in the performance gap between the best and second-best block producers can greatly slow down the rate of stake concentration. The paper concludes that while heterogeneity in block producers can lead to centralization, proposer-builder separation can reduce this heterogeneity and preserve decentralization among proposers, especially in competitive builder ecosystems.This paper investigates the centralization of block production in blockchain protocols and the impact of proposer-builder separation (PBS) on decentralization. It addresses two key questions: (1) Does heterogeneity in block producers' skills and knowledge inevitably lead to centralization? (2) Does PBS reduce heterogeneity and preserve decentralization among proposers? The paper develops mathematical models to answer these questions. In a game-theoretic model with endogenous staking, heterogeneous rewards, and staking costs, it quantifies how heterogeneous rewards lead to concentration in the equilibrium staking distribution. In a stochastic model, it quantifies the rate at which stake concentrates on the most sophisticated block producers. In a model with heterogeneous proposers and specialized builders, it quantifies how PBS reduces heterogeneity in rewards across different proposers. The paper connects its analysis to contest design, Pólya urn processes, and auction theory. It shows that economic forces generally lead to an oligopolistic equilibrium rather than a "winner-take-all" scenario. It also shows that conventional intuition around PBS breaks down if the distribution of block values is sufficiently heavy-tailed. The paper also provides quantitative predictions that would be impossible without concrete mathematical models. For example, it shows that if at least 10 block producers are at least 90% as good at extracting value as the most sophisticated block producer, then at equilibrium no block producer will control more than roughly 17.5% of the stake. It also shows that even modest decreases in the performance gap between the best and second-best block producers can greatly slow down the rate of stake concentration. The paper concludes that while heterogeneity in block producers can lead to centralization, proposer-builder separation can reduce this heterogeneity and preserve decentralization among proposers, especially in competitive builder ecosystems.