The technology deposits DNA fragments, many copies of a single fragment, in each of millions of wells on a chip. A, G, C, or T are flooded over the chip sequentially, and if one is incorporated by the polymerase, the reaction, not the chip, emits a proton and the sensor beneath each well picks up the change in voltage. The cycle is repeated hundreds of times, until the end of the fragment is reached or the quality of the reads falls off. Lets say you get a few million fragments of a few hundred basepairs, you get close to 1B bp per chip.
One of Jonathan Rothberg's insights was that the large, well funded core labs with the largest most expensive instruments, could not give the individual investigator the turnaround time she needed. So, he sought to deliver the solution in a low cost instrument that allows the investigator to decouple herself from the core labs. I've heard him say that Ion has the largest installed base of any sequencer, and I've heard from many investigators that what appeals to them is this very independence.