The fact they seem to be custom manufacturing silicon carbide FET's.... While everyone else is using off the shelf components.

Those custom FET's haven't yet publically been torn down. But I'd guess they enable more efficient motor inverters. More efficiency means smaller battery, and less cooling. Less cooling means the whole heatpump system can be downsized, reducing weight. Reduced weight means smaller battery. Smaller battery means less weight... Which means an even smaller battery... And a smaller battery in a lighter car means more profit margins.

Because of the recursive nature of this, even small efficiency gains on the FET's have a pretty massive impact on profit margins.

Do you really want 'untested' technology hurtling you down a road at nearly 100 mph?

Fets are pretty low risk... Worst case, you lose a motor. They have the explosive fuse to prevent a bunch of shorted fets causing massive braking, and I've never heard of that fuse being activated.

Even with that 5 year lead, I would rather buy an EV from any other "mainstream" automaker. The build quality even on a 2023 manufactured Tesla is just insulting. If I buy a BMW I'm also being ripped off but at least the door panel won't fall off if I slam the door, and BMW will repair faults within days and not months.

Other car companies e.g. Lucid, BMW have better technology in other areas.

And why would anyone bother with custom ML SOC when you can just partner with Nvidia like Rivian, Mercedes etc.

Because maybe there's some advantage to be had by reducing dependencies when you can afford to? And maybe they value SOC expertise, since they do other tech besides cars?

And when they outsource it people will say they're just focusing on core competencies. The great thing is businesses can use a magic 8-ball to make business decisions and someone will laud the results as a good move.

Vertical integration has many benefits. It’s the same reason Apple started making their own chips and ditched intel. The SOCs will be cheaper and better optimized for Tesla’s use case.

Every Nvidia GPU I’m aware of ships with both CUDA cores and Tensor cores. For a pure ML application those CUDA cores are almost useless. Tesla would be paying Nvidia’s premium for transistors that aren’t optimized for ML inference.

A Tesla designed chip can be 100% dedicated to tensor multiplication. You’re paying less per transistor and every transistor is utilized to the fullest extent.

Nvidia isn't using PC GPUs in cars. They have a dedicated SOC for self driving:

https://www.reuters.com/technology/chipmaker-nvidia-launches...

Is this even a lead? Seems like Tesla is just bearing the R&D that any other auto manufacturer can decide to implement if it makes sense for margins. There's no secret sauce when a competitor can get a hold of your product and see how it works. Its why in past wars doing things like scuttling vehicles was important to prevent competition.

5 years headstart means very little advantage in car industry unless you are developing ICEs.

I don't know if people realize how expensive it is to engineer an engine. Mercedes spent 4B $ to develop their diesel OM654. 4 billions of R&D to build an engine!

EVs don't have the same complexity, barrier of entry is much lower which is why Chinese are doing so well in them, even though they didn't with ICEs.

Auto industry will never be one where margins are good for long, doesn't matter who you are, it's a very expensive, risky low margin business if Tesla inverstors think thei company is immune they are in for a very harsh reality.

This is like investing abc.

I've driven other EV cars and Teslas, they do not have a 5 year lead by any stretch.