It still seems likely that they were offering more than they were charging. If the cost to convert is $200,000 and the cost to operate is $200,000 a year - but

It still seems likely that they were offering more than they were charging. If the cost to convert is $200,000 and the cost to operate is $200,000 a year - but they can only realistically bring in $200,000 - then no one is going to buy that model. It worked when investors didn't care about losses because the losses were hidden as technology investments that they thought would get smaller with scale.