and unprecedented speed of hardware/software iteration that comes as a benefit of vertical integration. It will ramp much faster than expected, in order to capitalize on their first mover advantage. GM agreed with our suggestion that large scale commercial operations would start relatively soon (we expect this in 2020). GM's operations will first target the most lucrative markets (densely populated cities). which are also arguably the most challenging for AVs. And this will be done in a highly disruptive manner (sub $1 per mile). We believe that this will make GM's service overwhelmingly attractive vs. incumbents such as Uber and Lyft ($1.53+ per mile). But the real goal is to make it more compelling than private vehicle ownership in dense urban centers (Uber, Lyft, etc account for just 0.1% of miles driven; in NYC private vehicle ownership costs >$3 per mite; in SF, Philadelphia, Washington, Boston avg. costs are >$1 per mile). A few details gleaned from our discussions led us to conclude that GM is most likely planning to go to market through their own Transportation as a Service platform, competing with Lyft and Uber (GM will be highly disruptive to those players; GM does not intend to sell the technology to unaffiliated third parties; and GM appears to have thought through many details of how the TaaS service will operate). We also believe that the faster ramp could evolve into a competitive moat, as the company's Al develops more rapidly (i.e. Their product will be better/more capable than most others), and their network establishes natural monopolies in key cities. The economics of their plan appear even more attractive than we had modeled. Useful life/avg miles per vehicle is a key cost driver (depreciation accounts for $0.24 of the $0.53 cost per mile assumed in our model). And GM believes that this cost is likely to be much more favorable than we've assumed, since their platform will be EV based, they will be engineered for long life cycle