When InsideEVs first talked about the MCUv1 issues on the Model S and Model X, some Tesla supporters insisted it was natural for computers to decay. Al Prescott, Tesla’s vice president of legal affairs, recently told NHTSA that is exactly the case. In his words, the computer does not have a defect. It just wears out like “brakes, lights, tires.”
You can read his words in the letter Prescott sent NHTSA. In it he says the company would comply with the agency’s request for a recall of first-gen MCUs, stressing that “Tesla respectfully disagrees that the eMMC wear-out condition constitutes a defect.”
Prescott goes on in the letter to oppose the dangers NHTSA believes a failed MCU could pose were all “caused by a defect.” That would be a condition for NHTSA to demand a recall, considering it is “limited to non-compliance with motor vehicle safety standards and defects that present an unreasonable risk to safety.” He also does not agree the MCU preventing defogging, rearview camera, Autopilot, and turn signals to work properly would pose unreasonable safety risks.
According to the Tesla VP, “components and systems (even those that impact safety, e.g. brakes, lights, tires) are not deemed defective if they fail due to age and wear,” which would have been the case for the MCU.
Prescott expressly says that the eMMC flash memory card was not designed to be “a component or system” that would function “without replacement or repair for the life of the vehicle.” He stresses that by writing, “it is inherently subject to wear, has a finite life (as NHTSA itself acknowledges), and may need replacement during the useful life of the vehicle.”
This is why Tesla would have “significant concerns with the impact of” saying that the MCU “should last at least the useful life of the vehicle, essentially double its expected lifespan.” According to Prescott, it was sure to fail while the vehicle was still able to run.
Although that may be a valid legal strategy to challenge the need for a recall, Tesla is not on record have ever warned customers they would need to replace their MCUv1 after around four to five years of use. Tesla claims it can hold up for five or six years, many owners have had to replace theirs in much less time.
If MCUv1 were a wear part, Tesla might have an inventory of such computers for immediate replacement when they started to fail. The owner’s manual might estimate its lifespan and inform recommended replacement times. It might also be easy to replace. Finally, it might have a predictable discard policy that would prevent these computers from ending up on eBay.
Regardless of the implications a computer as a wear part can have, Prescott makes it clear that Tesla “disputes that every safety risk is caused by a defect and that every defect creates an unreasonable risk to safety, especially when the condition does not surprise the driver while driving and the vehicle can continue to be safely operated.” That is how he believes Tesla customers deal with the MCU chronic issue.
“In the interests of efficiently resolving this matter and providing a better experience for the customer,” Prescott informs in the letter that Tesla filed “the requested Part 573 report.” That document, which you can see here, is the one that says the recall should apply to 134,951 units of the Model S and Model X.
Tesla does not say what happens to the other 23,049 cars that would complete the 158,000 cars in the US supposed to have the MCUv1. While some could have crashed, been converted to V8s in Rich Benoit’s hands, or anything similar, the explanation may be on the eMMC Warranty Extension.
The proposed remedy for Part 573 is not only related to the MCU, but also the OTA (over-the-air) updates 2020.24.6.11 and 2020.48.12. Tesla claims they have already addressed any safety concerns that failing MCUs would bring.
The 2020.24.6.11 update would keep defrosting/defogging instructions in case the MCU failed and would set the cabin temperature to 22ºC (71.6ºF) in every new trip after that. It would also “initiate windshield defrosting/defogging to ensure sufficient windshield visibility.”
The newer 2020.48.12 update would also take care of defogging and make the rearview camera show its images on the center display when the MCU failed. Apart from that, the car would warn the driver to schedule the MCU's replacement one to six months before it died. BMW already does that with CBS (Condition Based Service) when the car in question needs an oil change, new filters, or brake pads.
Prescott says, “NHTSA’s anachronistic regulations are unfit for situations where there is no safety defect, but nevertheless the manufacturer immediately can improve vehicle performance, including safety performance, without the cumbersome need for physical repair.”
Still, the defective MCUs will have to be replaced. And these OTA updates were only released in August and November 2020, according to Teslascope, while NHTSA launched its investigation of the problem on June 22, 2020.
The Part 573 document also clarifies how the repair will be performed. It does not consist of replacing the whole computer, only the VCM (Visual Compute Module) daughterboard where the eMMC flash memory card is within the MCU. This service can take up to 90 minutes in the Model X, and 75 minutes in the Model S, according to the page Tesla uploaded in its Support section – a lot less than the MCUv2 upgrade that was supposed to fix the issue, which takes about three hours.
In the same section of the document, Tesla promised to “reimburse any affected customers who previously paid to replace an eMMC, Visual Compute Module (“VCM”) daughterboard or MCU that was determined to be caused by accumulated wear of the eMMC.”
As we have covered in previous articles, many customers replaced their MCUv1 units with an MCUv2, believing that was the only solution for the issue. Some others replaced it because of leaky yellow screens that were also a frequent MCU issue. Regardless, all these computers would be subject to the eMMC failure, as Prescott admits, and therefore be eligible for a reimbursement of costs.
Instead of leaving that up in the air, Tesla addressed these situations in a FAQ for the recall. According to the company, these owners may also receive a reimbursement “up to the cost of the 8 Gb eMMC repair available at the time of your upgrade.”
Offered since March 2020, the upgrade cost $2,500 at the time, while a new MCUv1 used to cost $3,000 to $4,000. The MCUv2 upgrade price has recently been lowered to $1,500.
We have never heard of a repair that would replace only the VCM daughterboard and its 8 Gb eMMC card until the eMMC Warranty Extension Tesla presented in November 2020.
As the name suggests, the VCM daughterboard would be replaced under warranty for specific Model S and Model X units. These vehicles would have to be made before March 2018 and to respect two limits: a maximum of eight years or 100,000 miles. If anyone made the upgrade instead of replacing the daughterboard in their MCUv1 for free, the company may argue they are not eligible for reimbursements.
Intriguingly, the eMMC Warranty Extension is basically Tesla’s recall, only without the time or mileage restrictions. Even the link for that eMMC Warranty Extension page became the page for the recall’s FAQ: type https://www.tesla.com/support/warranty-adjustment-program, and you’ll land on the new “8 Gb eMMC Recall Frequently Asked Questions” page.
The last bit of Part 573 worth checking does not answer any questions but raises some more. When Tesla informs NHTSA how it fixed the issue in production vehicles, it says this:
“Tesla introduced all aforementioned OTA FW updates in production at or around when the Company deployed them OTA to the existing customer fleet, such that new vehicles were produced with the FW updates prior to customer delivery. In addition, in March 2018, Tesla discontinued the NVIDIA Tegra 3 processor with a Hynix 8GB eMMC in production and introduced the Intel Apollo Lake processor with a 64GB Micron eMMC.”
By mentioning the MCUv2 as a production correction, Tesla states it replaced the MCUv1 with the MCUv2 to correct the issue. If that were the case, it would be natural that the OTA updates that prevent safety hazards with failing MCUs were released around the same time, not two years later.
The text also suggests that the MCUv2 has 64 Gb since it was introduced, which is not true. The first 64 Gb eMMC cards in that computer would have started to appear by May 2020, according to TeslaTap. When we wrote about the issue for the first time – in October 2019 – it still had the 32 Gb card.
Prescott ends his letter explaining why replacing the MCUv1 with the MCUv2 is not the recall. In his words, it was to “make it available as an alternative customer-pay remedy.”
According to the Tesla VP, “some customers will prefer to upgrade to the newer Intel car computer, rather than receive the remedy repair, because the newer car computer offers enhanced features and performance.”
By stating the recall was not the right measure or that NHTSA could not ask for one since it does not come from a defect, but from normal and predictable wear, Prescott may have raised other legal concerns for Tesla. With the public document now in NHTSA’s hands, customers may decide a reimbursement is not enough for all the trouble they had. They may ask about MCUv2 being a “production correction” they now have to pay for. Another hypothesis is questioning the OTA updates, which were already accused in the past of concealing Tesla issues. The end of the MCU soap opera seems to be just the beginning of another complicated story.