In this note we investigate the considerations necessary for building a nation-wide health data infrastructure. We argue that to be able to meet the broad social objectives not only do the crucial privacy and fairness concerns have to be comprehensively addressed, but also that the theory of public good based on such an infrastructure needs to be carefully developed and the operational requirements and risks need to be clearly understood. In particular, an effective proportionality analysis by balancing the utility versus the risks becomes untenable when either the utility or the risks are inadequately or imprecisely modelled. We examine the necessary elements of a conceptual architecture required to enable such a proportionality analysis.