We propose no less, so we think, than the construction of the first genuine many-body quantum engine based on thermodynamic principles. It constitutes the vehicle of our comprehensive program of elucidating the role information plays as a notion of fuel in thermodynamics. It exhibits a complexity far transcending the few qubit regime, yet allows for flexible measurement schemes that enables a detailed study of how much fuel it takes to gain information, how that information is inevitably lost through thermalization and ultimately how information can be used as fuel. This anticipated device derives from ultra-cold atoms that in a tuneable fashion realize the full range from noninteracting to strongly correlated quantum fields. The key novel ingredient that brings this device to a new level are precisely programmable time-dependent potentials allowing to manipulate the quantum fields. In this way, operational primitives of compression, time evolution, spliing, merging, squeezing, and entangling can be implemented. antum mechanics is crucially required to capture its very functioning, and unlike machines involving a few degrees of freedom, one cannot eiciently predict all its properties. A broad range of experimental measurement techniques like single atom detection or high-order correlation measurements allow deep insights into quantum properties of the information gained. The functioning of this machine relies on thermalization dynamics and is intertwined with the long-standing puzzle of how thermalization and quantum mechanics can be precisely reconciled. This will allow us to investigate the next question: If we had more information, how could we use it to our benefit? Is information really a fuel for thermodynamic machines and how could we harness it? For a classical engine, knowing specific micro-states of some fuel particles will not allow any improvement of its eiciency – indeed, the very crucial part to this knowledge is also the ability to control systems at the level of the information one has about them. If apparent thermalization is accompanied with a loss of information, how can we, aer all, operationally exploit information? If the hypothetical Maxwell’s demon has implausibly much knowledge available leading to paradoxes, how much can practically accessible information available by an observer be teleologically exploited in thermalization control? Building upon ideas of harnessing information leads to the third core theme, asking about the thermodynamic cost of gaining knowledge. Indeed, the very fact that information can be used as fuel conjures the notion that it cannot be obtained freely. Here, we aim at unravelling the ultimate limits to obtaining knowledge, proving from first principles that as certainty of knowledge increases, the thermodynamic cost of establishing it diverges. The quantum field machines will be the first testing ground where these ideas can be flexibly tested, as the quality of information can be controlled by performing measurements of dierent levels of invasiveness by measuring only few atoms instead of the field and the obtainable information depends on the underlying emergent quantum field model, as well as the correlation functions one has experimental access to.