Tesla Overcoming Chip Shortages that Are Crushing Other Carmakers

Auto chip shortages have prevented carmakers from making over ten million cars. The microcontroller shortage after the Renesas fire in Japan and the COVID lockdowns in Southeast Asia caused most of the problems in 2021. Toyota is reducing its 2022 car production target to 8.5 million cars. This Toyota target is down from 9 million in January 2022 and down from 10.7 million in 2019, 9.2 million in 2020

Tesla can use more sources for its chips because Tesla programmers can write firmware in 2 weeks to let them use different suppliers for different chips.

There are several major categories of auto chips. There are microcontrollers (MCUs), analog, sensors, power chips, logic and memory.

Microcontrollers for powertrain, chassis & safety, infotainment, body & convenience, ADAS/autonomy
(15 to 100 per car)

Analog for power management, BUS transceivers, RF, radar, motor drivers, display drivers, LED drivers, etc. (hundreds per car)
Power discrete for xEV, chassis, powertrain (tens per car)

High end chips for LIDAR and self driving and driver assist AI.

Tesla created a unified computer architecture used in its cars. This reduced the number of MCUs needed from 80 to about 20.

Tesla has their own design for their self driving chip.

Analog chips shortages could last into 2024. The chip shortages in 2022 and 2023 could mean another 15 million cars are not able to be built. There was 71 million instead of 80 million cars build in 2021 and this could be the same in 2022 and another 5 million cars are not able to be built in 2023.

SOURCES= Evannex
Written by Brian Wang, Nextbigfuture.com

7 thoughts on “Tesla Overcoming Chip Shortages that Are Crushing Other Carmakers”

  1. It would be illegal. You need chips to meet legal requirements like ABS, emissions, airbags, reversing cameras etc.

  2. The economics of Space seem to be as counter intuitive as orbital mechanics sometimes. Bringing stuff from Space to Earth in particular is one of the few freebies out there. Thick atmos does most of the work. My comment was both about the general problems of everything being on Earth, whether all humans or all chip factories, and also micr0g opps. These are sure to make "better" stuff because unique is by definition better than nothing. Things like stability and lack of vibration in addition to any process that requires micr0g make "light and valuable" work because the launch is cheap for light raw materials. A different market from O2 or structural metals. Don't launch them.

    The "better or cheaper" part is the key. Even more critical, getting started is hard. Musk rockets are already good enuf to get going, much more on the way. The classic O'Neill plan is centered on avoiding launch, so looks at material that is cheap/easy and heavy to get from Moon, etc. But the main O'Neill question always is there. DARPA is looking at launch of material for "light and valuable" product that *must* be made in Space, thus Space is the right place. This launch would stop eventually as material is collected in Space, but there is no reason to wait, as the launch of light stuff is not out of reach. This launched resource ISM then grows fast.

  3. If you read stories about their production line, it appears one of the first things they install the chassis is the main computer, and they keep it hooked up while in production, using it for a number of jobs. One interesting mention is flashing firmware to various MCU's and chips throughout the various buses (with a wry note the chips don't come with recent firmware, and the updates often don' t take and have to be repeated…). They also use the main computer for various checkout procedures rather than using external testers. This computer first approach, along with using linux, and having access to silicon valley programmer talent, has helped them considerably, though much of it was due to a decent foundation laid by early employees (who have unfortunately quit).

    With the proper foundation, chip agility is much more straightforward, especially if you treat it like the plug-and-play architecture it truly is. In that sense, having a closer-to-general-purpose computer helps enormously over highly bespoke setups. Though to be fair, with modern MCU's and other chips increasing containing full ARM cores (which in some ways is overkill for the job at hand, but the added functionality allows updating and modification of behavior), it was riding the technology wave at the right time that also helped Tesla.

    In theory any new car startup, such as Rivian, could do the same, since they are not beholden to legacy cruft.

  4. And aside from using them in space, they're probably light and valuable enough to be returned from space, if you can make them better or cheaper there.

  5. Most have said in the past that computer chips would not be made in Space at first because they are so light and valuable that they can be launched. Looks like surface of Earth may not be "the right place" for even these. A curious story on Gateway seems to miss a main reason for Gateway. Micr0g experimentation and ISMRU with lunar rego. Micr0g ISMRU of computer chips. How could they miss that? You cannot do it on Moon or Mars? Where should the main lunar base be, Gateway or lunar surface? These are important, currently relevant questions. If you understand O'Neill, you will already know the answer. The comments in the cite are painful to read.

    https://www.thespacereview.com/article/4343/1

Comments are closed.