Serious like "technically level 3" from Mercedes on geofenced highway with a 60 km/h (37 mph) speed limit?
Isn't in-house development their own AI ASICs for driving and training not serious enough?
Serious like not ditching radars, so you don't have to solve some issues with it, at the expense of added safety. Radars are useful even in non-autonomous cars.
Serious like not using customers as QA. I would have expected the issue of Teslas crashing into emergency vehicles to have been caught in testing. or ensuring your driver can not use a simple trick to sleep behind the wheel.
Take a deeper look at what how Waymo has progressed. I am also quite interested in what Honda's lvl3 looks like outside of Japan, as they are far from Tesla's approach of overpromising and underdelivering.
The thing with the German OEMs, if one of them advances, others will follow, as they are intertwined quite a lot through their suppliers.
Regarding the AI ASIC inhouse development - so what? the point is in having (not promising) a quality product. Musk is huge on vertical integration, something automotive industry has evolved away from, spinning off their parts into tier 1 suppliers in order to increase efficiency. Not sure such high level of vertical integration Tesla is aiming for is a good thing. What guaranties success of what NVIDIA will offer? or many other AI ASIC startups popping all over for the past several years?