Tesla: Pay Attention To FSD 12
Dec. 04, 2023 11:00 AM ETTesla, Inc. (TSLA)MBGAF, MBGYY, XPEV, XPNGF, LI, LAAOF, GM9 Comments
Ultima Insights profile picture
Ultima Insights
331 Followers
Summary
Tesla, Inc.'s limited release of FSD 12 has the potential to change how drivers view autonomous driving and add billions of dollars in revenue and market capitalization.
The FSD 12 update marks a critical juncture in Tesla's journey towards achieving true full self-driving capabilities, with a shift towards a neural network-based system for driving decisions.
Tesla's supremacy in the autonomous vehicle sector, backed by its unparalleled data collection and extensive real-world testing, gives it a competitive advantage over rivals like Cruise and Waymo.
AI Safety Summit - Day One
Leon Neal/Getty Images News
Investment Thesis
As Tesla, Inc. (NASDAQ:TSLA) continues to blaze trails in the electric vehicle ("EV") and autonomous driving sectors, I think investors should pay closer attention to their recent limited release of FSD (full self-driving) 12, given out to only employees this past week but with industry-wide implications.
While much of the focus has been on the Cybertruck this past week, I think the real focus needs to be on FSD. FSD 12 has the potential to change how drivers view autonomous driving and add billions of dollars in revenue and market capitalization to Tesla. While CEO Elon Musk has missed promises in the past on full self-driving timelines, I think this one is different. In my opinion, the stock is a buy.
Background & FSD 12
Tesla's FSD technology, an integral part of its electric vehicle ecosystem, has recently taken a significant leap with the release of FSD version 12 (v12) to employees this past week. This update marks a critical juncture in Tesla's journey towards achieving true self-driving capabilities. Elon Musk has expressed confidence that with this update, Tesla is closer to realizing "true" self-driving capability, a goal that has been pursued with persistent innovation and technological advancements??. Some users have noted that the autonomous driving technology, even in the most recent 11.4 update, has felt smoother compared to earlier FSD updates. The expectation is that FSD 12’s update will make autonomous driving highly fluid and smooth.
In my opinion, the new FSD12 system is the game changer. The v12 update is particularly notable for its shift from hardcoded vehicle controls to those controlled by neural networks, encompassing everything from computer vision to driving decisions. The neural network-based approach allows for more adaptive and intelligent decision-making, a step closer to the complex task of full autonomy in varied driving conditions.
Tesla has also taken another leap forward recently with its integrated highway and city street programs. These programs have been merged into a singular, more complex system used in all driving scenarios. Tesla’s integration of these parts is a crucial step in enhancing the FSD's capabilities, ensuring consistent performance improvements in various environments??. As I mentioned before, enhancing self-driving technology going forward will be a function of making the technology more smooth so drivers have higher confidence in the software.
Implications Of FSD (For Tesla And For Their Customers)
The release of Tesla's FSD version 12 (v12) update signifies a ChatGPT-level technological leap, marking a shift towards a neural network-based system for driving decisions. With a better autonomous driving system comes more Tesla owners who are willing to adopt the technology.
For context, Tesla had around 285,000 FSD customers in the U.S. as of 2022, with payments ranging from $4,000 to $12,000 for FSD. With Tesla's expansion in vehicle sales and the potential increase in FSD prices (up to $14,000 or $20,000, or via monthly subscriptions), Tesla could add potentially billions in revenue each quarter with this technology.
For Tesla owners, Elon Musk is optimistic about the advancements in Tesla's Full Self-Driving technology, potentially leading to owners leasing their vehicles out into an autonomous fleet for a revenue share. Part of what is driving this confidence is a belief that FSD technology is nearing higher levels of autonomy and is “very close” to becoming safer than a human driver without monitoring.
Creating monetization opportunities from FSD is big. The upfront cost of the software package is incredibly expensive for any consumer (I can’t think of any other software I would buy from a personal perspective that is this expensive), but viewing it as an investment that allows you to rent out your Tesla can psychologically change the game.
Competitive Moat
Tesla's supremacy in the autonomous vehicle sector significantly outweighs its competitors like General Motors' (GM) Cruise and Alphabet's (GOOG) Waymo, primarily due to its unparalleled data collection. Tesla's fleet, numbering in the hundreds of thousands, has gathered over 1.3 billion miles of diverse driving data globally, essential for honing its machine learning algorithms. This volume vastly exceeds the data collected by Waymo and Cruise, estimated at 100 to 1000 times less, giving Tesla a unique edge in training and refining its autonomous driving technology.
In addition, unlike its rivals, Tesla employs both voluntary and automatic data transmission from its vehicles, ensuring continuous improvements from their owners simply driving their products. This allows Tesla to collect substantial data, estimated at 4 gigabytes per car daily.
First Principles Approach
Elon Musk's decision to rely on cameras over lidar for Tesla's FSD technology is based on his belief in the superiority of vision and a first-principles approach. Musk has expressed confidence in the camera-based system, arguing that vision has the potential to be far superior to radar or lidar. He points out that sensors like radar and lidar provide a bitstream of information, but cameras can deliver several orders of magnitude more bits per second, creating better data throughput, and allowing for better vision processing. Tesla can make radar increasingly redundant as vision technology improves??.
From a first principles perspective, Musk based on the idea on the fact that humans rely solely on vision (no Lidar) for driving. He highlighted this concept back in 2019 during Tesla's Autonomy Day, emphasizing the human-like approach to driving Tesla aims to replicate with its technology. Musk acknowledges the concerns about the safety of using cameras compared to radar, particularly in detecting events several cars ahead, but he remains confident in the capability of vision to perform as effectively as radar??.
The Power of Neural Networks
As touched on before, Tesla’s decision to adopt neural networks for FSD v12 marks a transformative leap in autonomous driving tech, shifting from code-heavy programming and further aligning with a first principles approach. The FSD 12 update is a significant upgrade and slashes over 300,000 lines of code, utilizing neural networks for steering, acceleration, and braking, closely emulating human decision-making.
Leveraging extensive real-world video data, at a level really only Tesla has, FSD v12 enhances its prowess in navigating complex scenarios. Its continuous learning from vast datasets and transition to network-path-based artificial intelligence ("AI") for tackling challenging environments highlight its potential for heightened safety and dependability.
Despite regulatory and safety challenges, I believe this update stands as a pivotal milestone for Tesla, propelling them towards a comprehensive "end-to-end AI" system, aligning AI with nuanced human behavior, and reinforcing their position in the AI-powered transportation sector.
How is The Rest of Auto Responding?
In my opinion, (and in contrast to Tesla's strides in Full Self Driving (FSD) technology), I think legacy automakers are trailing, with a few bright spots. One of Tesla's closest FSD competitors, Mercedes-Benz Group AG (OTCPK:MBGAF), received approval from the California Department of Motor Vehicles before Tesla for its Level 3 DRIVE PILOT system. This allows for conditional automation, but still requires human intervention at times (autopilot levels are 0-5, with 5 being the most advanced).
While this system is set to be available in 2024 S-Class and EQS sedan models in specific regions of the U.S. and Nevada, it's still confined within the bounds of specific use-cases and environments (for example, in California the max speed you can use this technology is set at 40 mph). Mercedes-Benz in a lot of ways is lacking the extensive real-world data and neural network sophistication of Tesla's FSD.
Chinese EV companies, like XPeng (XPEV) and Li Auto (LI), are aggressively pursuing their Navigation on Autopilot (NOA) systems, primarily focused on domestic markets. These systems are currently more driver-assist technologies (vs. full autonomy), requiring human drivers to be ready to take over. Approximately 360,000 cars in China are expected to be produced & equipped with city NOA capabilities in 2023. This compares to the 1.8 million vehicles Tesla is expected to produce this year alone, and all of them possess the hardware for their FSD technology. Furthermore, NOA technologies are still in early stages and are yet to demonstrate global applicability comparable to Tesla's FSD system. The technology is only useable near major cities in China like Beijing, Shenzhen, Shanghai, and Guangzhou.
Finally, while GM's Cruise had a promising start, the last few months have shown a limit to their driverless technology. Cruise has pulled its vehicles off the streets in San Francisco after a series of accidents called into question the safety of their autonomous technology. The Co-Founder of Cruise, Daniel Kan, recently left the company. GM is planning on paring back its spending on Cruise next year in light of these setbacks.
Risks
The potential impact of FSD technology is a pivotal consideration for Tesla shareholders, but full of risks (as is any groundbreaking technology). This is definitely a high risk, high reward play on Tesla.
The elephant in the room is technological challenges and delays. Despite Elon Musk's ambitious vision, the progress towards fully autonomous vehicles has been slower (and rougher) than initially promised, contributing to investor skepticism (Elon has had a history of promising new technology far sooner than it occurs).
Moreover, FSD technology is only available for Tesla vehicles (unlike their charging network which was just opened up to most EVs). Tesla has to keep this technology to their vehicles due to the hardware requirements. What this means is that the limiting factor (until FSD becomes part of the purchasing decision) is the number of Tesla’s on the road. While Tesla is on track to deliver over 1.8 million vehicles this year, this is still a drop in the bucket compared to 70 million vehicles sold annually globally; their market is still relatively small.
Autonomous driving works best with more autonomous vehicles on the road. As such, Tesla’s FSD is not and will not be invincible; it's still at risk of crashing the vehicle. Previous accidents related to Tesla self-driving technology have made national news, and the NTSB has investigated previous accidents.
I think a lot of this has to do with the novelty of self-driving technology. It's much like how people fear riding on a plane vs. driving in a car even though riding in a car (no FSD) causes a much higher rate of accidents and injury. In essence, this psychological bias has to deal with control. Getting people to trust FSD technology will take time, but it will be part of the evolution. Already, there are studies that some autonomous technology is already better at driving cars than humans are.
In addition, while there is a risk that Tesla is not the first to reach FSD, I think in this case the downside would be largely mitigated. FSD is currently not a purchasing decision for automotive customers, since no automaker currently offers full self-driving to any customer (Tesla only offers FSD 12 to its employees right now). As such, the customers who purchase FSD from Tesla are people who already have a Tesla. Given they already have the car, I believe they will not make a switch to another vehicle from Mercedes or another automaker just because another automaker gets FSD earlier than Tesla (this was not a purchasing decision when they chose a Tesla EV initially, so why would it be one now)?
So, while there may be a short period where another automaker has FSD before Tesla (and this for the short run could impact future purchasing decisions for new Tesla customers), I believe this would be short lived. In my opinion, there is little evidence to show that Tesla, if it is not the first automaker to come out with FSD, won’t quickly follow and be one of the first. Their technology and datasets are extremely advanced, as I mentioned before. I believe they are a market leader even if they are not the market leader.
Valuation: Assessing Tesla's Market Potential and Growth Trajectory
While Tesla is an expensive stock when measured by normal valuation methods (FWD P/E Non-GAAP ratio of 75.04, compared to the sector median of 14.78), I don’t think investors have ever viewed Tesla as a value stock. Tesla has almost always been viewed as a growth stock, and this is key.
The Fair Value of FSD to Tesla Stock
Globally, the autonomous vehicle total addressable market (TAM) is projected to grow from USD $33.48 billion in 2023 to USD $2,796.33 billion by 2032. It's an exponential acceleration as self-driving technology takes over. I believe Tesla will lead.
Tesla's primary market, the U.S. autonomous car market (their SAM), was valued at $12.27 billion in 2023, with expectations to reach $31.17 billion by 2028. With Tesla's growing FSD Beta user base, I believe this suggests its strong potential to capture substantial market share. I believe they can capture (based on SOM) between $6.23 billion to $9.35 billion in annual revenue by 2028, implying a ~30-50% market share in autonomous technology software in the U.S.
Keep in mind they are likely miles (literally with their training data) ahead of the competition, so I expect them to have a might higher market share of autonomous driving software vs. normal auto sales. Assuming that this annual software licensing revenue has typical EBITDA margins of around 40%, this should imply about $3.2 billion in EBITDA. Using an EV/EBITDA multiple that is just ½ of their current EV/EBITDA multiple of 49.26 (or 24.63), we get a $76.75 billion lift to the company’s enterprise value. Given this software will likely not add any incremental debt to the firm’s balance sheet, this should roll down to their market cap. This would imply about 10% upside to Tesla stock just by Tesla enabling this software for users to buy.
Why I do not think this is priced in
While it’s widely known that Tesla has been working on FSD technology, Elon Musk’s timelines have been long and unpredictable. He previously predicted that FSD would be available in 2020 or even earlier. There is no guarantee that investors are looking at this now as a catalyst. He has previously said that their market cap is tied to solving autonomous driving, but I believe this is not top of mind for many investors.
To be more precise, I am sure that more investors this past week have been looking at the Cybertruck launch vs. the new FSD technology. That’s not to say the Cybertruck can't be immensely beneficial to Tesla (I think it will be), I just think this technology could have an even more massive impact and its not getting the attention I think it deserves, given the potential.
Bottom Line
Tesla stands at the forefront of the electric vehicle and autonomous driving revolution. The company's significant advancements in Full Self-Driving technology and AI are not only groundbreaking but pivotal in shaping the future of automated transportation given their unique approach with neural networks. However, Tesla still faces regulatory pressures, previous bad publicity related to autonomous driving accidents, and likely some resulting consumer hesitation. Nonetheless, Tesla's proven ability to overcome such challenges given their first principles approach presents the company as a strong investment opportunity. This article isn't just an evaluation of Tesla's current status but a forward-looking assessment of its potential to reshape the automotive landscape with their autonomous technology. In my opinion, Tesla stock is a buy.
Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.
