The bulk of LEDs sold today are still fabricated using older epitaxial techniques such as LPE and VPE, but have relatively low brightness and a limited color range. The newer high brightness LEDs are fabricated from the InGaAlP and III-Nitride systems, with MOCVD being the preferred growth technique for manufacturing. While these new materials represent a significant increase in performance, they are also more expensive to grow. In this paper we consider the reasons for this, which include a less mature growth technology, lower production volumes, expensive starting materials, process efficiency, equipment throughput and cost, and safety and environmental concerns. Addressing each of these issues in turn, we examine what has already been accomplished, and what may be improved by further advances in equipment and process. A realistic COO model is of great utility in comparing product cost for different device structures, staffing schemes, reactor sizes, etc. We demonstrate that for a dedicated LED manufacturing facility, the lowest epitaxial cost is achieved by running around the clock with the highest throughput reactor that is fully utilized for the desired production level. When maintenance tasks such as cleaning and test or calibration runs are minimized, then materials costs will dominate the epi cost, which leads to the desirability of achieving both the best reproducibility and increasing the process efficiency. We show how in-situ control techniques are now capable of increasing preproducibility and thereby lowering product costs for manufacturing scale MOCVD reactors.