In my last post, I summarized a “brief” (ish) history of Tesla’s “Full Self-Driving” offering. In doing so, I discussed my main criticisms with how they’ve handled the offering over the years. I also promised that I would follow that with my suggestions for what I think Tesla should do going forward to improve the situation.
First, a quick recap of my view of “FSD”:
I think Tesla was right to invest heavily into autonomy and their Advanced Driver Assistance System
I think they were not alone in being overconfident and overly aggressive in predicting a timeline for delivering autonomous capabilities
I think Autopilot is a great name for an ADAS and acceptable branding for L2 partial automation modes like Autosteer, Navigate On Autopilot, and the current “FSD Beta” solution - assuming the marketing and product truth are clear that it is not full autonomy
I think “Full Self-Driving” was okay as a pre-order for future functionality still in development, and I think it was clearly articulated as such until 2019
Further, I thought that pre-2019, having “Full Self-Driving” as a separate pre-order thing in fact helped clarify, by way of contrast, that Autopilot was not an autonomous solution
I think the 2019 retcon to describe certain L2 functionality as “FSD” and to sell some L2 features exclusively via the FSD Capability package (with the removal of Enhanced Autopilot) was a huge mistake
It could have led some buyers to believe the system was more capable than it was/is
It makes it seem like they’ve delivered at least partly on the FSD promise when they have not
I don’t describe anything Tesla has done as fraud
More specifically: I don’t think they ever articulated an ambition or promise which they didn’t intend and make reasonable attempts to deliver on
I do think they failed to deliver on the FSD promise in a timely fashion
Tesla never promised a specific timeline for delivery of FSD. However, I believe there was an implicit promise that it would be delivered within a reasonable timeframe, and I think the 4 year limited warranty period of the vehicle is a good upper bound on what is reasonable (and even then, that’s pushing it)
I also am convinced that they will never be able to deliver on the FSD promise for vehicles equipped with “Hardware 3.0”, which accounts for all vehicles except the very latest 2023 S and X models
I think a fair argument could be made for the promised functionality to require at least an SAE Level 4 capability
However, I think they could probably get away with delivering a Level 3 mode with a reasonably wide Operational Design Domain and the majority of customers would feel okay about it
I think their new Hardware 4.0 suite is the absolute minimum with which they’ll ever deliver an L3 mode, and ever delivering L4 capabilities is a long shot (possible, but very far from a sure thing) even there
What to do now
So, with all of that established, what do I think they should do?
Let’s think step-by-step.
Step 1: Give up on FSD on Hardware 3.x
A wise Klingon leader once said, “ending a battle to save an empire is no defeat”.
I think it’s time for Tesla to admit to themselves and their customers that the FSD promise is never going to happen for HW3 vehicles. Arguably, they should have done this long ago, but I think in some ways it could be easier for them to do so now. That’s because they now have unique L2 capabilities that some people want and are willing to pay for. Which brings me to step 2…
Step 2: Replace the “FSD” package with “City Autopilot”
Through the FSD Beta program, Tesla has been testing an impressive L2 solution which they’ve now made available to hundreds of thousands of FSD buyers. I believe many new and existing customers would be happy to pay for this functionality, even knowing it will always be an L2 mode (at least on HW3 vehicles) and thus will always require driver supervision.
Step 3: Convert all current FSD purchases into City Autopilot purchases
All of us who have purchased FSD would be notified that it is being replaced, and that our prior purchases are being converted to City Autopilot purchases. Nothing would change about the functionality we have today. Any “FSD” branding present in the vehicle (which I believe is only there for owners in the beta program) would be changed to just Autopilot or City Autopilot, as appropriate.
Step 4: Give FSD owners a partial refund
Every current FSD owner (not subscribers) would be offered a partial refund - coming from the deferred FSD revenue Tesla has not yet recognized. The amount could vary depending on what they paid. Perhaps a 50% refund. Perhaps less. But it should be a non-trivial amount ($1,000 at the absolute minimum, in my opinion).
It would be acceptable to offer a selection as to the form of this refund. Cash, a supercharging credit (or unlimited supercharging for the life of the vehicle), a Tesla store credit, etc. Cash refund should be one of the options, but I’d be okay with some of the other options having a higher cash value (e.g., giving the option of $2,000 cash or $2,500 of supercharging).
Step 5: Offer a full refund in exchange for giving up City Autopilot
Because the offering is fundamentally changing, current owners should have the option to return their purchase and get a full refund, in exchange for giving up the features unique to the City Autopilot package. Potentially this could include an option to downgrade to “Enhanced (Highway) Autopilot”, even for those whose FSD purchase occurred during the timeframe when Enhanced Autopilot wasn’t available - meaning the refund would be the difference in cost between what they paid for FSD and the current price for Enhanced AP.
An offer to trade-in your current vehicle for a new HW4-equipped model with the FSD City Autopilot purchase transferring to that vehicle would be a nice option to give as well, and might be a preferrable choice for Tesla’s bottom line.
Step 6: Work on an L3 mode for HW4, give it a unique non-FSD name like “Autopilot Pro”
Ideally, they should eliminate the FSD option entirely, even on HW4 vehicles, at least for now. Instead, sell cars with Basic Autopilot as today, then offer Enhanced Highway Autopilot and City Autopilot upgrades - with the latter at the same or maybe slightly lower price than today’s “FSD” package.
I’d then focus development efforts on getting to an L3 mode. Maybe highway first, but potentially they could explore whether a city L3 mode is viable. An L3 mode allows the driver to stop paying attention to the road in supported conditions, but they must be present and alert (e.g., awake) and ready to take over with reasonably advanced warning when leaving those conditions. This can mean when leaving supported roads (e.g., approaching your highway exit), weather changing, time of day changing, approaching a construction zone, etc. It can also involve routing which avoids complicated maneuvers or troublesome intersections.
If they really need to pre-sell something, pre-sell this. And be clear about what it will and won’t ever be. Give it a name like “Autopilot Pro”. If there is a pre-sale of this, clearly label it as such.
Step 7: Keep working on actually delivering FSD!
Nothing I propose means giving up on L4 Full Self-Driving ambitions. If Tesla can get there, they’ll have no problem selling it. It’s just not something that is going to happen any time soon, even on HW4. With the path this unclear, it’s premature to sell this functionality.
If Musk insists on still selling this, it should only be for HW4+ vehicles, and should be exclusively a pre-order. No additional functionality should be provided to buyers until the actual promised L4 mode exists. No “beta” that’s actually an L2 mode. The presale nature of this should be made perfectly clear to drivers, and the words “pre-order” should be used on the order page.
Retreat from the battle, focus on the war
If Tesla does this right, it will receive approval from owners, investors, and the public alike. They can say they made a valiant effort, but the 3.x Autopilot hardware just isn’t enough for real FSD. But fret not, Tesla faithful - there is new hardware now, which can then be hyped up instead of being a badly kept secret (at least, once it’s widely available on the 3 and Y).
I believe investors don’t care if the current fleet will all become robotaxis. They care that Tesla keeps selling more cars, and that progress continues to be made toward actual autonomy - a story well-supported by the introduction of new, more advanced hardware (including whatever they end up officially naming their “Tesla Radar” that is now complementing “Tesla Vision” with HW4).
This won’t appease all critics of Autopilot and “FSD” (nothing will), but it might actually appease some. And it will dampen criticism from others in the public discourse, while saving face and letting Tesla own the message.
The time is right, Tesla. Do right by your customers and your business. Drop this baggage you’re carrying unnecessarily. Put your focus where it belongs - making the best EVs, leading the industry in ADAS, and being the first to offer a truly useful and scalable L3 solution on every new vehicle.
They can't deliver, won't retrofit and they won't refund. How's that not fraud? Also, don't get me started on the EU.
Tesla would love to be able to take this escape pod and move on... but refunding the cost of FSD won't work. Why? Because I bought a Tesla for FSD. If Tesla admits FSD can't be done, on HW3 then they sold me an entire car under false pretenses. That would be like saying "sorry, you can't use the supercharger network anymore". The autonomy (coming soon™️ 😉) was a differentiating factor. And it wasn't a "maybe" it was Elon saying "not a question, an absolute" multiple years ago. Even last year he said "based on our internal numbers we think FSD will be safer than a human by about the new year."
So Elon has been saying that FSD not only was soon, but an absolute fact. So it wasn't a risk to buy a Tesla and know you'll get FSD.
People bought Volkswagen Diesel cars on the promise that they were "good for the environment". VW wasn't allowed to offer a carbon offset in your name to offset the lifetime emissions. They were required to buy back the car which wasn't what was promised.
Venture capitalists being pitched by an entrepreneur are used to forward looking statements that promise the sky, and then may never take place. Why? Because if I invest $1,000 in Tesla at IPO and they deliver these risky bets I make $100,000. Risk is paired with reward. You're allowed to make optimistically false statements which... might come true when the people who who bare the risk also will stand to gain. A consumer will not get rich if you succeed, they only get what they paid, back in the form of a product equal to the value paid. So Elon's history of selling the moon (or mars) to investors was inappropriate for a consumer product. He talked like he talks to investors.
As a stock holder I think Tesla should try this strategy of suckering as many people as possible into taking the refund on FSD. But I will only accept a refund on the car "with all of the hardware needed for full autonomy".
"I think they were not alone in being overconfident and overly aggressive in predicting a timeline for delivering autonomous capabilities"
But everybody else exclusively took investor's money. Not consumer's money. Tesla sold robotaxis to people that they would be admitting will never be a robotaxi. The difference between "an electric car with above average lane keeping" and "a robotaxi" are two different products. I bought a "robotaxi" that would be unlocked through software later. It would be like buying an iPhone where the modem never gets enabled... You didn't buy an iphone you bought an ipod touch. That's a different product.