Imagine two coffee machines that use the same beans, the same water temperature, and the same grind size — yet one consistently brews a stronger cup. You'd never detect the difference by reading their spec sheets; you'd only notice it by drinking. This paper shows the quantum equivalent: two interferometer setups that look identical by every standard interference measure but deliver different amounts of usable work, because the order of their internal components imprints a hidden asymmetry. The committed claim: placing a quantum-controlled beam splitter before a fixed beam splitter in a Mach-Zehnder interferometer never yields less coherent ergotropy than the reverse order, even though both arrangements share the same interferometric visibility, stored energy, and incoherent ergotropy at every parameter point. The gap is governed by path predictability — a wave-particle duality quantity that standard fringe visibility does not capture. This is the first demonstration that visibility is not a complete measure of the coherence available as thermodynamic work. The architecture sits squarely in the quantum thermodynamics of small systems, specifically the ergotropy formalism introduced by Allahverdyan, Balian, and Nieuwenhuizen (2004). The paper uses a qubit prepared as a thermal quantum battery sent through two quantum-controlled interferometers — variants of the quantum switch idea where the order of operations matters. The key structural choice is decomposing total ergotropy into incoherent (population-inversion) and coherent (off-diagonal coherence) contributions, then showing these respond differently to device ordering. No exotic hardware is needed; the entire argument is analytical, backed by closed-form expressions. The integrity regime is theoretical throughout. The results are derived from exact analytical calculations on qubit systems, not simulations or experiments. The paper proves its claims via operator algebra and traces through explicit density matrices — the proofs are self-contained and checkable. There is no experimental validation, no noise model benchmarking against a physical device, and no code release. The relativistic extension (Wigner rotation correlating spin and path) adds a second scenario where decoherence from tracing out spin degrees of freedom creates coherent-ergotropy differences even where the non-relativistic versions coincide. This is elegant but remains entirely in the theoretical domain. The Landauer connection is the sharpest structural result. The coherent-ergotropy advantage of the preferred ordering comes paired with a purity deficit on the beam-splitter controller qubit, which means resetting the controller has a thermodynamic cost via Landauer's bound. The paper shows this reset cost is never larger for the advantaged ordering at the same temperature — so the ordering that extracts more work also costs no more to reset. That is a clean, non-trivial thermodynamic consistency check that strengthens the core claim. The relativistic extension is more speculative but genuinely interesting. By sending a massive particle through the interferometer at relativistic speeds, the Wigner rotation entangles spin with momentum (path), creating an effective decoherence channel when spin is traced out. This lifts a degeneracy: the two orderings now show different coherent ergotropies in regimes where their non-relativistic versions were identical. It is a proof-of-concept that relativistic effects can serve as a resource or a cost in quantum thermodynamics, though no experimental realization is proposed. The bottom line: this paper identifies a genuine blind spot in how physicists use interferometric visibility as a proxy for useful quantum coherence. The result is clean, analytically exact, and conceptually sharp. But it lives entirely in the two-qubit Hilbert space with no experimental grounding and no pathway toward a device. Its value is foundational — it changes how you think about the relationship between interference and thermodynamic work, not what you can build tomorrow.
