How Much Does FSD Impact Tesla’s Valuation? (NASDAQ:TSLA)

Tesla (NASDAQ:TSLA) is currently valued at almost a trillion dollars. So, the trillion-dollar question is how much of that valuation comes from Tesla’s Full Self-Drive software and how much more could it increase Tesla’s valuation?

According to Elon Musk, as quoted in Business Insider on June 16th, from an interview on Tesla Owners Silicon Valley, Tesla’s valuation is determined by FSD:

“But the overwhelming focus is on solving full self-driving. That’s essential. It’s really the difference between Tesla being worth a lot of money or worth basically zero.”

Musk has also predicted that Tesla will have a million FSD beta subscribers by the end of this year (electrek 7/20/22). Do that math! One million subscribers at $12,000 equal $12 billion of pure profit. At Tesla’s current P/E of 100X, that would add $1.2 trillion in value.

Let’s look at the value and utility of FSD. In addition to doing strategic analysis, I’m a beta tester for Tesla Full Self-Drive and will share my observations.

Tesla’s Levels Of Autonomous Driving

Tesla offers three autonomous driving packages:

Autopilot

Autopilot, which comes standard with all Teslas, as it does with many competitive models, includes the following functionality and features:

• Traffic-Aware Cruise Control: Matches the car’s speed to the surrounding traffic.
• Autosteer: Assists in steering within clearly marked lanes and uses traffic-aware cruise control.

Enhanced Autopilot

In addition to the functionality and features of Autopilot, Enhanced Autopilot also includes:

• Navigate on Autopilot: Actively guides the car from a highway’s on-ramp to off-ramp, including suggesting lane changes, navigating interchanges, automatically engaging the turn signal, and taking the correct exit.
• Auto Lane Change: Assists in moving to an adjacent lane on the highway when Autosteer is engaged.
• Autopark: Helps automatically parallel or perpendicularly park the car with a single touch.
• Summon: Moves the car in and out of a tight space using the mobile app or key.
• Smart Summon: The car will navigate more complex environments and parking spaces, maneuvering around objects as necessary to find the driver in a parking lot.

Full Self-Driving Capability

In addition to the functionality and features of Autopilot and Enhanced Autopilot, the Full Self-Driving Capability also includes:

• Traffic and Stop Sign Control (Beta): Identifies stop signs and traffic lights and automatically slows the car to a stop upon approach, with active supervision.
• Recently Added: Autosteer on city streets.

Tesla makes it clear on its website (but not in the names it uses) that the currently enabled Autopilot, Enhanced Autopilot, and Full Self-Driving features require active driver supervision and do not make the vehicle autonomous. It states that full autonomy will depend on achieving reliability far above human drivers, as demonstrated by billions of miles of experience, as well as regulatory approval. As Tesla’s Autopilot, Enhanced Autopilot, and Full Self-Driving capabilities evolve, cars will continuously be upgraded through over-the-air software updates. There is still an open question about cars needing future hardware and sensor upgrades.

Tesla charges an additional fee of $6,000 for Enhanced Autopilot and $12,000 for Full Self-Driving. There is also an option to pay $199 monthly for FSD. FSD is only available as a beta test version, and the city streets capability was just recently added.

How Well Does FSD Work?

Tesla FSD is the first vehicle available at retail (not including AV fleets like Cruise and Waymo) that can autonomously turn corners and stop and go at traffic signals. It’s impressive, but nevertheless, it isn’t full self-drive in the definition of that term. And there are still many flaws in the way it operates.

There has been a lot of criticism about FSD lately, with articles about tests showing a Tesla with FSD running over a dummy child, followed by rebuttals from Tesla fans who tested it on their own (or borrowed) children to show that it does stop. Even Ralph Nader came out of retirement (since the 1960’s Unsafe At Any Speed) to call on the National Highway Traffic Safety Administration to recall FSD to “prevent the growing deaths and injuries.”

I’m not going to repeat these discussions. There is plenty to read about these. Instead, I will share some of my real experiences beta testing FSD. I’ve been driving a Tesla Model 3 for more than three years. I’ve driven it thousands of miles in Enhanced Autopilot mode. Recently, I’ve been driving FSD with autosteer on city streets. (Maybe that opportunity will be removed after this article?)

First, FSD can now make left-hand and right-hand turns, stop at traffic signs and lights, start driving on its own when the light changes, and change speed automatically based on speed limits. It’s impressive. It’s fun but stressful to drive/ride. Here are some of the problems I experienced:

• Many of the turns are clumsy. To turn right, it frequently drifts to the left into the other lane before maneuvering the right-hand turn.
• On one left-hand turn, it cut deeply across one of the oncoming lanes on the road it was turning into. It couldn’t tell that three lanes were coming in that direction because its vision was blocked by cars in the first two lanes. It (I) was almost hit by a car coming up in the lane it cut into.
• It drives erratically on roads with no painted lines. To its credit, FSD does estimate where the road ends even without lines, but then it struggles to position itself on the right side of the road. In one case, where there weren’t any painted lines, but there was an apron on the side of the road, it would try to center itself on the apron instead of the road.
• Enhanced Autopilot and FSD are two distinct software platforms. Tesla plans to merge them in an upcoming release. However, for now, the hand-off from one to the other is clumsy. Previously with Enhanced Autopilot, when it exited the highway, it would force me to take control at the end of the exit ramp. Now it just gets confused and comes to an abrupt stop at the end of the ramp because it goes straight where the exit splits to the left and right.
• Once, it tried to turn right with an oncoming car and would have hit it. Not sure why it did that.
• Sometimes it doesn’t read a speed limit sign and continues to use the previous speed limit.
• It gets confused at right-on-red turns. It hesitates, then tries to jump into too small of a gap between oncoming cars. At one intersection, it attempted to turn right-on-red when it didn’t recognize a no-right-turn-on-red sign posted across the intersection.
• A couple of times, it tried to go through a stop sign because it didn’t register the sign in time. These are intersections where the stop sign is far to the right of the turning lane because of a large bicycle lane or extra lane width. It only recognized the sign as it went through the intersection and then stopped abruptly in the middle of the intersection.
• Driving on a rotary/roundabout. Nope. It can’t do it. Don’t try it. There are three unmarked rotary lanes with the protocol to weave inside and outside as you approach your exit. This is too advanced.
• It does identify pedestrians. I wish it would go wider and slower around them, though. Remember, this is a silent EV that people don’t hear. It frequently identifies multiple people as one and golf bags as people, but these are not big problems. One time it stopped abruptly because it thought pedestrians walking on the sidewalk would try to cross the street when they weren’t.
• It automatically drives at the speed limit, which is sometimes too slow and sometimes too fast. For example, on a winding road, the speed limit is 30 MPH. That is too slow for the straight sections, which isn’t much of a problem except for the frustrated trailing cars. But at some curves, 30 MPH is too fast. Experienced drivers intuitively know to slow down approaching those curves, but Tesla FSD tries to navigate the curve at 30 MPH — and it doesn’t. It veers into the oncoming lane and sometimes stops abruptly if there is an oncoming car.
• The visualization system is much better with FSD. It’s cool. It recognizes the edge of the road, not just the painted lines. It doesn’t always recognize objects correctly. The other day I drove right through a tractor-trailer or just the image of one created by the shadow from my garage. It doesn’t recognize some types of emergency vehicles. It interprets the flashing lights as backup lights or turn signals. I wish it would add color to the visualization, but that is too much for the computer’s processing power.
• It stops suddenly when a car coming in the other direction on a curve appears to be approaching you.

Despite all of its flaws, FSD is still impressive technology. It’s a real novelty to take people for a ride in it. The panic on their faces is priceless. Is it safe? Well, maybe, as long as you really pay attention to how it is driving and take control of it before it causes an accident. And it will improve with software upgrades. Using Autopilot for highway driving is more relaxing than driving yourself. You need to pay attention, but very little happens that the car doesn’t control appropriately. FSD, on the other hand, is more stressful than driving yourself. It would be best if you worried at every turn, every curve in the road, every traffic light or stop sign, and every pedestrian you pass. In city driving, this is constant. You need to override FSD frequently.

In addition to the question about safety, two critical strategic questions will determine the success of FSD for Tesla. The first is understanding if the problems like the ones I mentioned can be solved by updated software or are due to a fundamental limitation of Tesla’s technology approach. The second is deciding if FSD is worth $12,000.

Tesla’s Autonomous Driving Technology Platform

There are several alternative approaches to autonomous driving, each using a different technology platform. These autonomous driving platforms have different costs and limitations. I go into these in detail in my books on autonomous driving, but I’ll try to summarize the two primary ones.

Camera-Only Based System

The first autonomous driving approach uses a system with cameras to identify what is around the vehicle and where it is within lane markings. This is the approach used by Tesla with eight monochrome cameras of different focal lengths around the vehicle. The way this system works is that the cameras identify what is ahead and around the vehicle. It uses lane markings to position the vehicle on the road. The cameras also identify other vehicles, stationary objects, and pedestrians. The Tesla onboard computer characterizes these objects based on shape, size, and motion. It then determines the vehicle’s path, acceleration, and stopping.

This autonomous driving approach is used in Tesla’s Enhanced Autopilot, as it is in other cars for semi-autonomous highway driving. These work very well on straight highways and roads where there is no need to turn corners. The Mercedes Distronic and GM Super Cruise are similar systems. I’ve driven with the Mercedes system for thousands of miles in addition to the Tesla, and I found the Mercedes system to be smoother.

The big technical question is: Is the camera-only approach sufficient for city and rural driving, where it needs to turn corners, stop at traffic signals, and interact with pedestrians? That is what Tesla is trying to do with FSD.

Camera System Combined with High-Definition Maps

The second approach uses a similar camera-based system with other sensors. But, it importantly adds another system with a detailed high-definition map and lidar to precisely position the vehicle within that map.

HD maps (illustrated in the photo above) provide three primary sets of information. The first is the actual 3D representation of the road and its related features like stop signs, traffic signals, lane markings, crosswalks, curb heights, etc. These features are measured with centimeter-level accuracy. The second interprets what each such sign, light, and marking means. And the third outlines the optimal drive paths the AV can follow smoothly. Sensors on the AV, especially lidar, identify where the AV is in the real world and matches it precisely to the HD map. It can then use the HD map information to guide it.

In addition, to the HD map system, these vehicles also have cameras and other sensors to identify all dynamic objects around them, such as vehicles, pedestrians, etc.

Pros and cons of the two approaches

The camera-only approach has two significant advantages. It is cheaper and not restricted to areas defined in the HD map. The disadvantage is that it might not work for city and rural autonomous driving, where it needs to turn corners, merge with traffic, and follow traffic signals. So it might not be capable of fully driverless operation.

Let’s consider some of my experiences with Tesla FSD discussed earlier. Many of these can be improved with software changes but not all of them.

• For example, the left-hand turn where it could not visually identify that there were three lanes, so it assumed two lanes because it could see cars stopped in those lanes. It then turned into the third lane because it thought it was the correct lane, but there was an oncoming car. The correct lane for the direction it was turning was the next. A system with an HD map would have identified the three lanes and provided the correct turning radius.
• The examples where the stop sign was not visible to the camera would not have been a problem for the HD map system because it would be identified on the map. FSD just didn’t see it in its field of view.
• The problem of excessive speed on upcoming curves might be able to be addressed with longer-distance cameras but might not be reliable in all cases. The curves would be fully identified with an HD map, and the AV would be instructed to reduce speed as it approaches the curve.

These limitations could keep Tesla FSD from being truly driverless and unable to dispatch a Tesla vehicle to drive on its own with no driver to take control. This would limit its value, as we will discuss below.

The other approach, the HD map system, can be fully autonomous but only within the mapped area, generally referred to as geofenced. That is why it’s the system of choice for what I refer to as sufficiently autonomous driving. An AV can reliably drive from one point to another along a predefined route. The HD map system will be used for fleets autonomously driving specific routes. The fleets will test their AVs on the routes over and over to make sure they function safely.

These will be the first three implementations of autonomous driving, with significant launches expected in the next two years:

• Autonomous Ride-Hailing. Waymo (GOOG) is already operating in Scottsdale, and Cruise (GM) is operational in San Francisco. To better understand this market and technology, see: Advantages GM Cruise Has Over Uber.
• Autonomous Long-Haul Trucking. See Autonomous Long-Haul Trucking: 3 Pure Play Investment Opportunities.
• Autonomous Home Delivery. There are several pilot projects around the country, and Nuro is developing a special AV delivery vehicle.

How Valuable Is Tesla FSD?

So, how valuable is Tesla’s Full Self-Drive? Let’s look at it in several ways.

Elon Musk’s two statements at the beginning of this article frame the value: Tesla is worthless without it, and with it, it will provide $12 billion of incremental profit annually. However, sometimes Elon Musk exaggerates things (ask Manchester United). From a practical analysis, neither of these is most likely correct.

In Tesla’s recent annual meeting (Tesla 8/4/22 Annual Meeting), Elon Musk discussed his view on FSD. He made an important observation that cars today sit idle most of the time, typically used only 12 hours per week. When autonomous, the utility can increase to 50-60 hours per week. The cost is relatively the same. He envisions Tesla vehicles to be fully autonomous through FSD, and those owners will be able to make money by sharing their cars the way Airbnb and Uber work. He believes that assuming this, Tesla will become the most valuable company in the world.

However, this assumes that FSD will be completely driverless.

Tesla recently cited that it had 100,000 people authorized to use FSD and that they drove 10 million miles last month. This is an interesting data point because that’s an average of only 100 miles of FSD per person per month, which equates to a cost of $2-$3 per mile for FSD, depending on how much was paid for FSD. For most people, city driving miles are much lower than highway driving. If FSD is only a convenience feature for a small portion of the driving and not truly driverless, then for many, it’s too expensive an option at $12,000. And this assumes that it works sufficiently.

So, the most significant value of FSD comes if it is fully driverless and Tesla can create a virtual fleet from customer cars that would be competitive with Cruise, Waymo, Argo, Aurora, Motional, and possibly Apple. This isn’t likely in the near future. But maybe there is another alternative.

Recommendation For Tesla

However, there is another way for Tesla to get value from its FSD technology. It doesn’t need to (and can’t) make the jump to driverless right away. Here’s what it could do:

• Rename this release Limited Self-Driving. In addition to enabling the name FSD to be used for a future release, it would diminish many of the FSD complaints.
• Keep it as a beta test program for longer — at least another year. And request feedback more directly from its users.
• Require the driver to be hands-on and have the system turn control back to the driver immediately in certain situations. An example is when it gets to an intersection with a red traffic signal and is turning right, give control back to the driver with the question “turn right on red?” This would enable the driver to decide if it’s safe to turn right or not. It would also avoid the risk I mentioned previously of not seeing a no-turn-on-red sign. There are other similar situations such as turning left on an intersection where the line markings are not clear. This could be implemented similar to Enhanced Autopilot where the vehicle asks drivers if they want to initiate lane changes. Put in many of these cases initially, but gradually eliminate them as the software improves.
• After beta testing, price the Limited Self-Driving feature at $6,000 instead of $12,000. The higher price could be used when true FSD is achieved. This would probably require reducing the Enhanced Autopilot feature price too. It may mean refunding some previously paid fees, but that’s not a big deal.

This approach would provide a safer way forward for Tesla, even though it’s not as dramatic and may not be Elon Musk’s style. It would enable Tesla to still get additional revenue/profit from its technology, with probably more customers buying it at $6,000 than at $12,000. It would mean several years of Limited Self-Drive with FSD deferred until the future. Tesla could still be the first to offer driverless fully-autonomous cars at retail in the future. However, most likely it would be well behind Waymo, Cruise, and others with autonomous ride-hailing. It’s an issue of strategic priorities, having a feasible core strategic vision.

Conclusion

There are several diverse outcomes on the value of FSD.

• It could have no value if authorities shut it down for safety reasons, and/or it proves to be unsafe with too many accidents. It’s possible that this could even have negative value if it damages Tesla’s reputation.
• It could have additional near-term value to Tesla if released as a Limited Self-Driving alternative, along with continued longer-term value if it becomes the first fully autonomous car sold at retail.
• It could have tremendous value if it becomes fully driverless in the immediate future and competes in the autonomous ride-hailing market. But I think this is a long-shot.

Analysts don’t seem to value FSD much in their projections, although Tesla does not break out Autopilot or FSD revenue yet. (Note: Elon Musk criticizes analyst forecasts because they don’t even drive a Tesla.) ARK Invest puts significant value on FSD with autonomous ride-hailing in its forecasts. Many Tesla fans and retail investors are ecstatic about FSD.

In any case, the progress of FSD should be watched closely over the next few months by all Tesla investors.