GM and Cruise, did Intel make a mistake

[Arthur Hanson]

GM is getting ready to deploy self driving cars on a ride sharing business model. This seems like a far better way to initially deploy and develop self driving cars, where GM can far more easily work with cars they control, than cars sold to the public. This also allows them to continuously use best of breed, rather than be tied to a single system or platform. This seems a far better way to develop a system. than combine two companies, neither of which makes cars. This is a far more efficient use of resources and talent with for more flexibility to adapt to the "Great Acceleration" we are now living in. It's far better to use strategy than just throw money and resources at a challenge.

Cars are a complete system and this allows GM to develop and test a wider variety of technologies and platforms, than just creating an add on system developed in isolation. Autos already have extensive computer systems and to not fully integrate these from the start needs a single unified strategy that this model provides. Link below.


Forbes Welcome

 

[name99]

This is orthogonal to the Intel-MobilEye merger. It is about the business model for the COMPLETE VEHICLE, but has nothing to do with the TECHNOLOGY that said vehicle uses.

There's nothing, in principle, wrong with MobileEye's position in the food chain --- they can be a parts supplier just like tyre manufacturers or transmission manufacturers. The grounds for criticizing the merger, rather, are

- does MobilEye have the correct technology? Which will win (cameras vs LIDAR)? Or will the final solution use both?

- this is Intel we are talking about, a company that is basically a one-hit wonder (hell of a hit, but still...). They've wrecked everything they've tried that hasn't been able to exploit their x86 advantage, and over the past two or three years they've moved up from incompetent strategy to incompetent execution. Autos look like very much the same sort of field as mobile; in other words there is ZERO advantage to using x86 over any other ISA, there is no installed base that anyone cares about. And mobile, like autos, seems like a field whose significant players want to decide (at a high level) what goes into their particular bespoke SoC rather than just accepting whatever Intel forces on them.

Will Intel accept this reality? Will they accept an ARM-like existence where they sell blocks of IP (that aren't forcibly tied together by pieces of x86 legacy that burn excess power, are harder to design for, and present a larger hacker attack surface) that GM, Mercedes etc, can configure as they wish? Or will they try a re-run of the Atom fiasco?
From my tone, I guess you can tell how I think it will play out.

 

[John Grant (Nine Tiles)]

Which will win (cameras vs LIDAR)? Or will the final solution use both?

Autonomous vehicles must use multiple different kinds of sensing, in the same way that humans both look and listen before crossing the road. Otherwise there will be more acccidents like the one where a Tesla crashed into a trailer which its camera hadn't recognised.