Offerings like Uber Pool and Lyft Line are the future of the business and depend entirely on economies of scale. Putting multiple paying passengers in the car at once and reducing driver downtime completely changes the economics, but only if there is a high density of riders and drivers.
Edit:
There are two possible stories here about Uber and Lyft's losses, and we can't tell which is right from the outside.
1. Uber and Lyft are in a price war death spiral, heavily subsidizing all rides to compete. Their only hope is for all competitors to die, so they can take over the whole market and raise prices. Then vague hopes of lowering costs with self-driving tech and take the surplus as profit. The plan doesn't seem viable, since they have no moat and they can't outspend GM and other self-driving players.
2. Uber and Lyft can actually make a profit in mature markets, but choose to subsidize rides in growing markets, on the theory that growing the market size increases the long term profit opportunity. The data we need to evaluate this idea isn't public, however Uber did a "prove it" quarter in 2016 where they turned the spigots to be profitable in the US. They are now pursuing growth in the US again, expanding to a larger territory and expanding Pool and other offerings like flat-rate passes in mature cities. Under this model, Uber could at any time decide to become profitable, but the revenue growth would stop, placing a cap on the valuation. Notably, there is no "predatory pricing" here.
With internal finances, we could easily tell whether option 2 is valid. You can bet Uber is constantly doing experiments on price elasticity of demand and knows exactly where the truth is. You can bet Softbank has seen numbers we have not. If Uber wants to IPO, I would expect them to provide additional public info, possibly pivoting to profitability again in the US. But doing a pivot like that permanently reduces the size of the opportunity, allowing Lyft to capture that new market instead, so they may be reluctant to do it large scale.