I don't understand why some perceive the big screen as a "cost-saving touchscreen". It's so much bigger than any other car uses, I would guess other makers perceive it as a "way too expensive touchscreen”.

I do not know the actual manufacturing cost comparison between the Tesla big screen and the corresponding set of buttons and other cars, but I think the key difference is a radically minimalist design aesthetic. I suspect both a bunch of buttons or a big screen are sufficiently inexpensive.

The Tesla decision is more similar to Apple's decision, who was not saving money when they put the largest screen ever used on a cell phone, covering most of the front of the first iPhone. Even though in the long run it probably turns out that the big touchscreen is cheaper than a bunch of buttons which individually wear out and must be warrantied etc.

I've worked on the costing side of things for head units and instrument clusters, so let me weigh in here since there are a couple things at play.

First is that Tesla just straight up doesn't play the traditional head unit provider game. Most OEMs go with one or two hardware suppliers (Denso, Pioneer, Kenwood, etc.) for their head units, pretty much all of which are completely and utterly behind on technology and also ridiculously overpriced for a variety of reasons. They also add expensive OS/software licenses or in-house development for, again, behind the times products to run on the head units and pair with your phone. Tesla, on the other hand, just grabs a good-enough touch screen, powers it with Nvidia hardware, and cranks out some pretty software built on top of Linux that mostly works.

Second is labor cost. Instrument cluster hardware isn't free, but it also isn't terribly expensive for your standard gauges in mass-manufacturing bulk. However, labor is not cheap to actually wire up, place, and test all those gauges. Wiring is one of those things that's hard to automate, so your costs between the hardware and labor boost the BOM cost up quite a bit. Tesla, on the other hand, just runs a single CAN bus connector to their center console display hardware, pulls the info they need to update it, and calls it a day.

The net result is that the total production cost of a single center screen ends up being quite a bit lower than the equivalent head unit + instrument cluster in a traditional car.

I could understand if they just replaced the infotainment with a big screen like in the model S, but the only reason they replaced the gauge cluster with a section in the center screen is to save costs. Comparing for example with the interiors from Audi A3, Mercedes A class, or Volvo XC40 they all look much more finished and developed and practical, where the Tesla model 3 looks like an unfinished afterthought.

Wait–let subcontrators worry about developing state-of-the-art capabilities for them?

FWIW I think they should be way more conservative with how they advertise Autopilot, but the only way something like that is getting shipped at this stage is from in-house.

My point was that there's no reason Tesla needs to be first to market with self driving cars, they already have a unique selling point in their electric cars. The car industry is full of suppliers doing all kinds of advanced stuff. Ferrari uses dampers from MagneRide and Öhlins, brakes from Brembo, transmissions from Graziano, and so on. Rimac makes hybrid drive trains, Bosch makes ABS and ESC, few brands makes their own seats, and so on.

Bosch have their own autonomous driving development. I don't know how far along they are compared to Tesla, partly because Elon likes to boast where others prefer to hold back, but so what if Tesla would've been a few years later with self driving than others?

https://www.bosch-mobility-solutions.com/en/highlights/autom...

> I still think betting on autonomous driving is a strange decision for a small company that have plenty of other stuff to focus on.

Yeah, but how expensive is that team?

Considering that

1) it gets some good PR (although other manufacturers are slowly starting to catch up with their own lane assistance + lane departure + adaptive cruise control features)

2) that PR effect probably generates a good influx of resumes from software engineers well-versed in AI, hardware, data science, etc.

3) those engineers can then be assigned to work on other software features, as shifting priorities demand

If you fired everybody on Autopilot team, you'd probably need to spend most of that saved cash on recruiting funnel.