Thorium is always used in mixed cores. An accelerator-driven thorium reactor will contain uranium from breeding even if no uranium is ever loaded into it.
But anyway, 12 years is too long, even if it's feasible. Something like 4 years should be better, due to the learning curve on these. For instance, think of a reactor in the field as a ship on a long trip while technology at home continues to improve
http://en.wikipedia.org/wiki/Wait_Calculation
Edit: The lower the enrichment, and the lower the excess reactivity, the better.
And yes, we kick start using thorium with low enriched uranium or spent fuel to build up U233. Then we just keep reusing the thorium fuel over multiple refueling periods without reprocessing or separations until it is depleted.
The military is interesting because they already have small reactors in submarines and ships. And they have access to the highest grade fuel because they don't have to follow power plant regulations. So it seems like if these modular reactors would be useful to them, they could easily adapt these reactors for land use.
This startup is hard to gauge because there's so few actual details about their design. For example, how are they going to use thorium as a fuel? You can't just use thorium without a strong neutron starter source or some other fuel engineering.