The second law of thermodynamics governs the "arrow of time" problem: why broken glass never recovered again, why heat is flowing always from hot to cold, the friction converts the ordered energy into heat and never back on, etc...: Another consequence of the second law is the efficiency bound on heat engines -- the celebrated efficiency value of the Carnot cycle that appeared in 1824 and states that not all heat can be converted to work. This is an instance of the power-efficiency dilemma for heat engines: if their efficiency (usefully delivered energy over the input energy) is large, then the power (useful energy over the cycle time) is low and vice versa. This dilemma restricts the efficiency of practical engines to (at best) some 40 % of the maximal efficiency, i.e. 60 % of the available energy is wasted. Is the dilemma a temporary complication or a fundamental property of nature? This question was addressed in a recent paper that appeared in the Physical Review Letters [*]. The authors show that the dilemma is fundamental for natural heat engines, the ones that do not have purposefully designed interactions with thermal baths. The authors performed a fully quantum mechanical analysis of a n-level heat engine reciprocating between two thermal baths. The structure of the optimal engine is such that the engine's relaxation (a necessary step of the cycle) is equivalent to searching for a marked item in an unstructured database of n items. The latter problem has high complexity so that the time necessary to solve it grows with n nullifying the power whenever the Carnot efficiency is reached. As is proven in the paper, it is not possible to circumvent totally this fundamental hurdle. And yet one can spot an analogy with protein folding processes and show how to use that evolution-optimized mechanisms to construct efficiency-power compromised but still efficient heat engines. In the mesoscopic scale (n ~ 100) the structure of the machine and its relaxation pathway is similar to a protein folding process, which, as the mere existence of life implies, progresses in finite time. Inspired by this analogy, the authors propose a reasonably optimal mesoscopic heat engine which makes, e.g., 1 second long cycles while being highly efficient (e.g. 90% of the Carnot value) and giving out almost the half of the maximal work. [*] A.E. Allahverdyan, K.V. Hovhannisyan, A.V. Melkikh, S.G. Gevorkian, Powerful Carnot cycle and attainability of the maximal efficiency, Phys. Rev. Lett. 111, 050601 (2013).