All cells must synthesize – a.k.a., ‘translate’ – proteins to grow and divide; faster-growing cells must thus produce proteins more rapidly. Previous work from our group showed that the mechanistic origin of this speedup Escherichia coli is a coupled change in physical crowding and molecular stoichiometry that optimizes transport and encounters between translation molecules [Maheshwari et al. (2022) mBio]. However, the initial model left a three-fold gap between modeled and in vivo rates, suggesting other mechanisms are at play. Here, we extend our computational model to incorporate attractions between translation molecules, which we show organize the cytoplasm and facilitate transport, speeding up protein synthesis even further. Crucially, we show that the interaction valency – i.e., the number of ternary complexes that can bind to each ribosome – must be limited to recover in vivo measurements and prevent arrested phase separation of the cytoplasm. Broadly, our work suggests a connection between the valency of this interaction and cell fitness, which could be tuned to either support faster growth rates in synthetic cells or slow the growth of pathogens.