Remember that they're collecting field test data from all of those 2nd-gen autopilot cars, over the air. They are the only ones who can do that. They have a forward-looking very underutilized compute platform in the car, and they use it to test all of the processing flows they're working on. Field testing of sensor data processing is a big cost for everyone but Tesla. Nobody other than them can do self-driving research on a fleet of tens of thousands of cars, in all sorts of real-world conditions. Market analysts with no grounding in tech seem not to pay much attention to that. Tesla flleet's telemetry is worth way more than $1B/year if anyone else were to decide to spend on getting such data. Each year the continue to exist, they get another $B+ bucks worth of data. It won't take long until just the test data they get over the air will be worth a good chunk of their market cap. They are doing some other cool things in terms of data gathering when their cars come in for service, that nobody else is doing, and that gives them a lot of industry unique data as well. Tesla is superficially a car company, just like Walmart is superficially a retail company. Over the years other retailers had to play catch up in terms of their IT and data capture. Same will happen to car companies, but right now Tesla compared to other companies is like Walmart was compared to other grocery retailers in the early days of its data processing push. Very much ahead of the game.