Short answer: when the captured knowledge in question, creates ongoing costs which gradually impact local, state and national budgets over time. Some of the most serious budget issues occur, when knowledge capture protects skills as utilized in basic forms of time based product. The most obvious example is physician compensation, as the highest income in the U.S. This can also be thought of as an instance where the private sector actually encouraged government "crowding out", in order to preserve limits in skills use. In other words, the private sector is not ready to assume a broader healthcare marketplace on monetary terms.
However, it is possible for final product which is independent of time value, to negatively impact budget revenue as well. In particular, costs associated with pharmaceuticals are a stark exception to normally positive tradable goods production - given the fact much of their organizational capacity isn't radically different from other tradable sectors. In the U.S., pharmaceutical companies were able to free ride on the special privileges granted to healthcare associations in the twentieth century. As Washington provided assistance for costs which were becoming prohibitive, the result was higher costs - alongside a tangled web of insurance subsidies and employer obligations which in turn made it more difficult to hire employees.
Knowledge capture in healthcare, has become a formidable general equilibrium problem. This form of wealth capture represents a continuous drain, from fiscal policy which could have been more closely aligned with normal government obligations and infrastructure needs. Healthcare's hidden subsidies matter. They have reduced options which otherwise would have been available to Washington for overcoming economic stagnation, and made it unnecessarily difficult for government to play a role in overcoming the limits of the zero bound.
Government healthcare obligations also contribute to local fiscal problems - especially those embedded in the health benefits of public employee pensions. Too many communities - in order to fulfill already outstanding pension commitments - end up reducing both public services and levels of public sector employment. Still these are temporary measures, before the thicket of subsidies becomes unsustainable. Should asymmetric compensation remain the only option for healthcare services production, it's not difficult to imagine societies replacing ongoing time based physician product with robots, in order to (ultimately) address budgetary concerns.
Clearly, such an approach would also be problematic at an existential level. Already, patients have less time with physicians and other providers than they would prefer. Only consider how often pills are substituted for one on one time with healthcare providers and time based care. Possibly the biggest difference between alternative medical approaches and those of today's traditional care, is the degree to which alternative care is more time intensive.
Yet it would take a marketplace for time value, before time intensive service provision could once again be considered institutionally viable. Only a marketplace for time value would provide greater knowledge diffusion for innovation, as well. How so? Quite unlike tradable sector product, innovation in healthcare thus far tends to create more costs, for time based services which are asymmetrically compensated.
Even though it is widely assumed that physician's work is "safe" from robots, this may be true so long as governments can balance healthcare obligations with other (equally pressing) budgetary demands. Of late, policy makers hardly dare speak of looming government cutbacks for healthcare - even though nothing has occurred yet which would preserve the healthcare marketplace for the long term. As a result, the same knowledge capture and protection which makes these vital skills so monetarily lucrative, also limits the ability of healthcare as the profession it now represents, to remain a viable part of progress and prosperity.