Computing is often described as the abstract manipulation of symbols. In fact, however, every computation is a physical process that unfolds over time and is subject to corresponding limitations. This lecture argues that memory, not logic or inference, sets physical limits on computing. Using the example of quantum random access memory (qRAM), memory is understood as a thermodynamic object: a structured store of correlations that must be generated, stabilized, addressed, and renewed. From this perspective, computational performance depends less on algorithmic depth than on the ability to maintain and access memory states without thermal degradation. The familiar distinction between classical and quantum computing is thus reframed in terms of temperature, stability, and memory bandwidth. The result is a shift from algorithm-centered models to a memory-centered, physically grounded theory in which the limits of computing reflect the structure of time itself.