Tesla’s Full Self-Driving software is under active development.
Tesla iterates its software through regular updates to improve product capabilities and address issues and edge cases.
The goal has always been to take this to the point where the car is self-driving, and we’ve heard several public statements from the CEO that the fleet will have the option to switch to robotaxi.
For more than 5 years, Tesla has been selling vehicles equipped with Hardware 3 (HW3), a combination of computing and cameras that has been interpreted as the hardware needed to achieve this self-driving feature.
We’ve seen previews of a future Tesla mobile app that would let you register your car with the Tesla Robotaxi fleet. How many owners actually do this, in order to have a chance to monetize their car, remains an open question, but we do know that Tesla will make its own custom Robotaxi that also depends on the completion of FSD.
Tesla will hold a Robotaxi announcement on October 10, 2024 to unveil the robotaxi which will then be sent into production with the first units expected to roll off the production line sometime in 2025.
For current owners of approximately 2 million HW3-equipped Tesla vehicles, a subset of them have purchased the Full Self-Driving software package. This applies to vehicles worldwide, while vehicles in the U.S. and Canada can also purchase it as a subscription and opt-in/out monthly.
On the Tesla website, and in your purchase agreement, Tesla has committed to offering “Autosteer on city streets,” the element of FSD Capability that remains distinct to those who have not opted for FSD Supervised, or live in international markets. Autosteer on City Streets is also known as Full Self Driving (Supervised) which is expected to become Full Self Driving (Unsupervised) in future releases, which will actually be part of the software that will enable automated taxis.
What we saw this week was Tesla’s move to launch the latest version of FSD V12 for cars with HW4 first, and a statement from Musk that HW3 vehicles will take a little longer as they optimize the version for the computing capabilities available in HW3.
As we’ve seen with larger language models, things get much easier with more hardware, but over time, the performance of these models can be maintained or even improved as the models improve.
Given the volume of affected customers who own HW3 vehicles, there will likely be a lot of angry customers who will demand refunds or file lawsuits, if Tesla ever says it can’t/won’t ship FSD to HW3 vehicles.
Option 1: Solve FSD using Homework 3
The most obvious scenario, and the best outcome for everyone, is that Tesla continues to iterate on models to make FSD work on HW3 vehicles, and we may just need to wait a little longer than HW4+ vehicles.
Naturally, this will not sit well with customers who feel they have invested in the company by purchasing a Model 3 at a time when the company’s future is in question and others who have bought at prices that mean they are underwater and cannot economically upgrade to a HW4+, even if they have access to the FSD transfer to the new car.
If Tesla decides that making FSD work on HW3 is not possible or economical, there are a few additional scenarios that could occur.
Option 2: Refund to customers
A number of Tesla customers bought their cars and FSD software because they believed their cars could achieve self-driving capability that would make them money.
These customers still enjoy the benefits of owning a car, and 90% of the program features, and therefore will never be entitled to a full refund.
The price of FSD has changed over time, so the price for each customer will be different, and customers also benefit from other aspects of FSD functionality, even if the Autosteer system on city streets is not complete.
Some other FSD features (Autopilot navigation, smart summons, traffic light and stop sign control) may still be in beta, but they are features that are present in the car, so it’s hard to prove they qualify for compensation.
So let’s say Tesla offered a potential refund for an Autosteer on city streets. The owner would see potential revenue as part of their calculations that wouldn’t materialize in this scenario.
Imagine you paid $10,000 for FSD, let’s imagine Tesla refunds HW3 customers 1/7 of the cost, considering Autosteer on city streets is one of 7 features listed in Full Self-Driver Capability on Tesla’s website.
Refund Possibility
There are a lot of potential variables when it comes to calculating any potential payback, here’s one scenario, but feel free to play with the numbers and make your own assumptions in a comment below.
- Number of HW3 vehicles: ~2 million
- FSD take rate ~30%: 600000
- 1/7 FSD Feature Set Cost: ~85,714
- Estimated recovery value ($85,714 x $10K FSD cost): $850,714
Based on these assumptions, Tesla may be able to “reimburse” affected customers who purchased FSD for less than $1 million, a largely insignificant figure for a company with about $30 billion in the bank.
Option 3: Update your devices
Tesla has already made it clear that it is not possible to directly swap the HW3 for the HW4 and above. There are physical differences in size, so there is a big hurdle here, which may cause the development of the HW4b board to support a potential hardware upgrade.
Account only
Imagine that Tesla could shrink the size of the HW4 board to fit into the physical space, and it also needed to support the existing camera inputs and connectors and it would make sense.
If Tesla manufactures the HW4b panel to solve this problem, it will likely require millions of dollars of development to achieve mass development and production in partnership with suppliers.
The HW3b manufacturing upgrades could reduce the capacity of the HW4+ panels produced for new vehicles and could limit the compute ramp for dedicated robotaxis. This is a significant obstacle for Tesla to move in this direction as everything about robotaxis points to high volumes and with each potential vehicle it will lose future rideshare revenue.
The practical reality is that Tesla Service will need to install any additional hardware. Even Tesla will admit that it is already under-resourced at its service centers, so adding this burden would be unwelcome and would attract additional costs in wages and opportunities.
Computing + Cameras
If Tesla requires additional camera quality (increased pixel count) to match the HW4 cameras, most of the cameras could be replaced.
- The front camera array is accessible from the mirror assembly on the windshield. Once it is on, you will need to recalibrate the cameras, but replacing the cameras will also require this.
- The fender flares on the front quarter panels and the indicator cluster should be an easy upgrade.
- The B-pillar cameras are also replaceable, with the section between the front and rear windows being able to be removed to access and replace the cameras.
- Access to the rear camera located in the trunk lid is also fairly easy.
If the physical connection (power/data) is different between the HW3/HW4+ cameras, you either need to splicing the cables or replace the entire cable.
Obviously the biggest question mark is the wiring. If the wiring from the camera location to the computer can’t handle the data/power required by the updated cameras, there’s no point in upgrading the hardware.
The car’s wiring harness runs through trim pieces, but is usually mounted on the body rails, and replacing it would require dismantling almost the entire interior, which at this point doesn’t make sense.
If so, the service center costs are real and very high in this scenario.
To me, it’s clear from thinking through these scenarios that updating Tesla vehicles with HW3 is not practical.
