Originally, the cationic and anionic resins were in separate tanks, and the process could not go to full completion, since some impurities were left behind from each process. The maximum resistivity that could be achieved was in the range of 2-4 megohm-cm. This was known as a "two bed" DI system, and some companies still use two bed DI modules in an attempt to convince you that they have higher capacity than mixed bed DI modules, but don't be fooled by this. While they can process more water, they can only process it to a 2-4 megohm-cm endpoint, and it still takes just as much ion exchange capacity to get it up to 18 megohm-cm! By mixing the cationic and anionic resins together in a single tank, the net result is an essentially infinite sequence of ion exchanges that do go to completion, resulting in 18 megohm-cm water.
Ion exchange resins come in a wide variety of types, and most of them are unsuitable for producing 18 megohmcm type I "ultra-pure" reagent grade water for laboratory applications. Those that do, have been specially processed and rinsed, and are usually classified as "Semiconductor Grade", since the semiconductor industry stands to lose the most if their final rinse water contains any impurities. The ion exchange or "DI" portion of the laboratory water purification system does (essentially) all of the work in removing ionic contaminants from the feed water except, of course, for the contaminants removed by the RO system, if any.