Having worked at a nucleotide synthesis company, I can attest to the general issues regarding chemically synthesizing DNA / RNA, especially what was said about reagent acquisition (expensive, hard to source), handling (flammable, toxic) and disposal (again, hazardous). I can't think of anything that would compel me to try this at home.
That said, Kilobaser looks interesting. If I were thinking seriously about getting something like this, I would definitely want to ask about reagent disposal (how do you dispose of the chips and other consumables), reagent exposure (if it breaks, is there any risk to the user), and synthesis quality. They do mention the inline trityl measurements, which will give you a readout of the coupling efficiency at each cycle (and thus a decent measure of how many truncation products you'll have in the end), but doesn't tell you anything about sequence fidelity. For sensitive applications, one or two mismatched bases could be a big deal, so fidelity matters.
If you *do* end up getting one, let us know!