Knowledge infrastructures are not merely technical supports; they are sociotechnical configurations that shape how knowledge is produced, circulated, preserved and legitimised. Mounier and Dumas Primbault argue that digitalisation has made visible what often remained in the background: networks, platforms, archives, standards, protocols, institutions and human practices that enable contemporary research. Their central claim is that such infrastructures should not be understood as neutral tools, since they embed values, hierarchies and forms of power. Open access, for instance, depends on repositories, identifiers, indexing services and funding models that determine who can publish, consult or reuse knowledge. The case of open science reveals this tension clearly: although it promises to democratise knowledge, it also requires sustainable governance systems capable of resisting commercial capture, technical exclusion and dependence on private platforms. In this sense, infrastructure becomes a space of negotiation among researchers, institutions, funders, technicians and diverse publics. Its apparent stability conceals permanent fragility, because it requires maintenance, legitimacy, resources and continuous adaptation. Therefore, governing knowledge infrastructures involves far more than administering technology: it means deciding which forms of knowledge count, who participates in their circulation, and under what conditions the epistemic common good can be sustained.