Sustainability in social robotics is often addressed at the level of individual artifacts, design choices, or efficiency metrics. However, social robots are embedded over time in socio- technical systems composed of users, infrastructures, routines, and shared practices, within which their (re-)use, maintenance, and disposal are collectively shaped. Under these conditions, sustainability can be adequately explored not only at the level of robotic artifacts, but also at the level of the socio-technical systems in which robots are embedded. Starting from this premise, the paper investigates technological discontinuation in social robotics as a critical site for analyzing sustainability. It explores the conditions under which post-discontinuation configurations can generate system-level regulation leading to sustainability-relevant outcomes after support withdrawal. The analysis builds on prior work on AIBO, where post-discontinuation sustainability was interpreted through a self-organizational framework describing system-level regulation involving users, repair practices, and material infrastructures. While this case showed the heuristic power of a systemic approach, it left unanswered whether self-organization can function as a general descriptive framework for system sustainability in social robotics. To address this issue, the paper subjects the self-organizational framework to a stress test through a controlled comparison of the post-discontinuation cases of AIBO and Moxie, characterized by sharply different organizational dependencies. The comparison suggests that self-organization operates not as a general framework, but as a diagnostic analytical device, identifying the organizational conditions under which system-level regulation can stabilize, or remain inhibited. On this basis, the paper examines when post-discontinuation sustainability can, and cannot, emerge in hybrid human–robot ecologies. [This manuscript has been submitted (04.02.2026) and is currently under peer review.]