At the threshold of 5G, much is promised based on the predicate of hyper-accurate location-aware communications. Although the unique spectral real estate associated with 5G implies the possibility of such a persistent localization scheme, we submit that precision range based localization is not a generational inevitability but rather achievable only under the correct architectural philosophy. To show this, we consider a fundamental component that limits or enables any positioning system: the underlying infrastructure geometry. The significance of this research follows from several significant results. First, cell ultra densification, although the primary catalyst of Long Term Evolution (LTE)/LTE-Advanced capacity improvement and a major disruptive technology in 5G networks, will not alone provide for more accurate user positioning. To show this, we derive a closed-form solution to the Cramér-Rao lower bound specific to 5G networks given that infrastructure is distributed via the Poisson point process (PPP). This result is used to justify a positioning services architecture that fundamentally embraces a variable number of supporting access points. In fact, we show that without this architecture, a foundational decline in positioning performance may inadvertently be realized. Finally, by numerically comparing our results with other point processes common in cellular network modeling, we present justification for the PPP as an appropriate model for analyzing positioning performance in ultra dense 5G infrastructures.