Immunity (whether acquired thorugh infection or vaccination) for most pathogens does not last “forever”, i.e. for a person’s lifespan. The common interpretation of “herd immunity” as some point at which a pathogen essentially disappears from circulation is just not correct; from an epidemiological standpoint, “herd immunity” is just the point at which enough of a population is sufficiently resistant that it is no long possible to sustain an epidemic level outbreak and the pathogen becomes endemic in the population; that is, that the effective replication number is consistantly below unity across the population.
For some pathogens this occurs naturally; seasonal Influenza A, for instance, tends to burn through an uninnoculated population until replication falls below the epidemic threshold, and then becomes endemic until it mutates or combines with another variant to develop into a new strain. Prior to comprehensive vaccination, the poliovirus responsible for poliomyelitis tended to fluctuate betwen epidemic and endemic. Varicella zoster, responsible for chickenpox, has never fallen below epidemic thresholds even though prior to a vaccine virtually every adult on the planet was exposed and had acquired immunity sufficient to prevent recurrence (in the chickenpox form, although the virus is resident in the dorsal root ganglion of sensory nerves and can reemerge to cause shingles in patients under high stress or with various autoimmune conditions).
For SARS-CoV-2, the acquired immunity through infection does not seem to provide very robust protection; at three months post-infection it appears to provide somewhere around 60% protection from reinfection, compared to between 80% t0 95% for most two shot vaccines (mRNA and adenovirus vectors), and falling dramatically thereafter. Even worse, about 1 in 5 in natural infection do not develop antibodies against either the S-protein or the nucleocapsid protein; whether that is luck of the draw or some kind of persistent genetic issue is unknown, but it appears there will be a substantial population of people who will probably not acquire natural immunity despite repeated exposures and will provide a reservoir for future infection. Vaccines appear to provide a more resilient antibody response, and it may be that with a longer sequence of vaccinations (3 or 4 shots, perhaps spread over a period of many months or even a couple of years) could provide lasting immunity, at least against the current strain of the virus.
Frankly, this pathogen is transmissible enough and capable of regular mutation that there is no reason to expect that it will ever disappear from circulation. What will likely happen is that eventually a variant will evolve that is infectious enough to outcompete other variants but only marginally pathogenic (causing only mild and transient respiratory issues, similar to other “common cold” viruses), thus preventing more pathogenic variants from taking hold. Of course, the opposite could also occur; SARS-CoV-2 could mutate to become as pathogenic as SARS-CoV(-1) or MERS-CoV, with double digit mortality rates, and potentially into a novel strain resistant to current antibodies, in which case this current pandemic would just be a light precursor to the main event of truly catastrophic proportions. This is one reason why there needs to be a system of worldwide monitoring of infection and sequencing of a statistical sampling to identify potentially threatening strains before they cannot be contained. (We also need to monitor the virus in the many animal reservoirs it is known to live in, because even though there has not yet been an instance of spillback from domestic or game animals, and the spillback incident from farmed mink was not more pathogenic than the wild-type virus, it is probably a matter of time before a more virulent variant emerges from another animal reservoir, and we can only hope that it is not as infectious as currently circulating varianrts.)
Although a lot of focus has been put onto “herd immunity” as a resolution to the pandemic, particularly in the popular press, in epidemiological and especially immunological circles more focus has gone into harm abatement, e.g. getting as many people vaccinated as possible, even with the less effective vaccines with waning immunity (which, it should be noted, are still quite effective in preventing severe disease and death) as well as developing theraputics to treat infected patients such that they do not require intensive and invasive treatments. The pandemic will be effectively over when this disease can be treated by prescribing a standard course of retroviral treatments and/or limited through comprehensive vaccination, but the expectation that this will just somehow disappear isn’t realistic, and it isn’t likely that it could be completely eradicated even by a global campaign like that which eradicated smallpox, rinderpest, or (almost) polio.
Stranger