I think you're probably right. And in order to bootstrap the necessary volume (and starve their competitors), they were willing to actually pay more than market

I think you're probably right. And in order to bootstrap the necessary volume (and starve their competitors), they were willing to actually pay more than market (similar to Uber's investor-subsidized rides in their early years). So the problem isn't so much that they lost money (which they probably expected for the first phase of the program), but that they lost more money than expected (and could tolerate). The additional cost and carrying time associated with labor and supply shortages in the construction industry may have contributed in that case, though I agree with the article that the adversarial market aspect was probably the biggest factor: in a competitive market, particularly bad offers were particularly likely to be accepted.