Photo: Tesla/edited by autoevolution
In October 2022, I wrote an editorial to discuss how the idea of “computers on wheels” can be a massive issue for cars. Most people crave over-the-air (OTA) updates and think they are an asset, forgetting about bugs and how often it is necessary to upgrade the hardware to keep up with software improvements. Will cars have upgradeable hardware? That question had two critical developments recently.
The first one was Elon Musk stating that Tesla’s next computer would not fit current cars. This future HW4.0 computer will obviously turn the company’s EVs into robotaxis by whatever deadline the Tesla CEO comes up from the top of his head. That breaks the promise he made that all Tesla vehicles apart from the Roadster would become autonomous with HW2.5 in 2016. Musk reinforced that in 2019 when he stated robotaxis would arrive by 2020 with HW3.0, a computer that could be retrofitted into any modern Tesla product. HW4.0 kills the hope of “appreciating assets” once and for all, and I suspect HW5.0 will do the same when it becomes the missing piece to reach full autonomy at some point in the future.
The second development is a reminder from Philip Koopman of the implications that OTA updates bring. The AV safety expert wrote on LinkedIn that “newer software versions won't run on older hardware.” On top of that, “hardware improves much faster than cars wear out, and at some point, companies won't want to upgrade older hardware.” Tesla has already reached that stage. In fact, it reached it in 2021, when it tried to avoid responsibility for the MCUv1 units that were (and still are) failing because of an eMMC flash memory card. Tesla creates constant logs that wear out the component because of its low capacity (8 Gb).
This “mandatory end of life” becomes the expected outcome regardless of if the vehicle can still run. It may have a battery pack that works fine, motors that can last thousands of miles more, and that will make no difference if the computer hardware these “computers on wheels” have needs to be upgraded.
Just think about it for a moment: the primary purpose of a car is to take you places. When it turns into a “computer on wheels,” that objective gets compromised. It will fail as a computer does instead of just entering a forced retirement when something crucial to its main function breaks, making the vehicle useless. That means you may end up with a car that could be running around perfectly if it was not for a computer. Sadly, Tesla vehicles have already shown their main components may fail before their computers do.
Either battery electric vehicles are conceived to allow for an easy upgrade of their computer hardware, or they will create a much bigger end-of-life problem than combustion-engined cars already do. The expectation for them was that they would last much longer. After all, electric vehicles do not emit pollutants apart from rubber dust from their tires – which is probably inescapable for any car using them.
Theoretically, they could work for decades with no public health concerns if their battery packs lasted that much. We already know current ones don’t, but a battery replacement program or swappable battery packs would make that easy. That will never be the case for computers on wheels as they currently are.
When their hardware fails, they are done. Are you ready to replace your car as often as you replace your computer or electronic gadgets? Are you prepared to have something that costs so much as disposable as things that cost ten times less? This is the future you can expect from computers on wheels if we don’t do anything to tackle the problem right now.
The second development is a reminder from Philip Koopman of the implications that OTA updates bring. The AV safety expert wrote on LinkedIn that “newer software versions won't run on older hardware.” On top of that, “hardware improves much faster than cars wear out, and at some point, companies won't want to upgrade older hardware.” Tesla has already reached that stage. In fact, it reached it in 2021, when it tried to avoid responsibility for the MCUv1 units that were (and still are) failing because of an eMMC flash memory card. Tesla creates constant logs that wear out the component because of its low capacity (8 Gb).
Possible outcomes
Koopman tried to figure out what companies would do when they did not want to upgrade the hardware anymore. One possibility is that “they are going to have to maintain multiple generations of software to keep the fleet moving.” That implicates keeping a team busy with an old product while they could be working on new cars. At some point (if not immediately), that will cease to make sense from a financial perspective, and the second and darkest possibility emerges: “bricking older vehicles or disabling automation features.”This “mandatory end of life” becomes the expected outcome regardless of if the vehicle can still run. It may have a battery pack that works fine, motors that can last thousands of miles more, and that will make no difference if the computer hardware these “computers on wheels” have needs to be upgraded.
Just think about it for a moment: the primary purpose of a car is to take you places. When it turns into a “computer on wheels,” that objective gets compromised. It will fail as a computer does instead of just entering a forced retirement when something crucial to its main function breaks, making the vehicle useless. That means you may end up with a car that could be running around perfectly if it was not for a computer. Sadly, Tesla vehicles have already shown their main components may fail before their computers do.
Hardware failures
The 1-million-mile Tesla Model S is the best example of that. Hänsjorg von Gemmingen used three different battery packs in his car and “11 or 12” rear motors so far. When they were replaced under warranty, that was ok, but what will happen when he has to purchase a new battery pack, costing around $20,000? What about new motors every 100,000 miles or even less than that (83,333 if Gemmingen replaced motors 12 times)? I have never asked Gemmingen how many MCUv1s he had in his car, but he must have replaced it at least once.Either battery electric vehicles are conceived to allow for an easy upgrade of their computer hardware, or they will create a much bigger end-of-life problem than combustion-engined cars already do. The expectation for them was that they would last much longer. After all, electric vehicles do not emit pollutants apart from rubber dust from their tires – which is probably inescapable for any car using them.
Theoretically, they could work for decades with no public health concerns if their battery packs lasted that much. We already know current ones don’t, but a battery replacement program or swappable battery packs would make that easy. That will never be the case for computers on wheels as they currently are.
When their hardware fails, they are done. Are you ready to replace your car as often as you replace your computer or electronic gadgets? Are you prepared to have something that costs so much as disposable as things that cost ten times less? This is the future you can expect from computers on wheels if we don’t do anything to tackle the problem right now.
Google News