The Rot at Tesla is Deeper Than This Moment

Not to be that guy but I have hated Elon Musk since WAY before it was cool.

When I was in high school my dad was one of the people who helped launch the Tesla Model 3 instrument panel. He didn’t work for Tesla directly but at one of their suppliers. The folks at Tesla all had software backgrounds and didn’t really know how to build cars, it seemed. They didn’t realize that when you build heavy equipment wrong, you can’t just debug it like with code. You have to rebuild them, which can take months and wild amounts of money. My dad had to drive out to where they were trying to build them for like a year, a nearly two-hour commute each way, on top of working the extended hours necessary to clean up the problem. I don’t think he even got paid extra for all the extra work.

This wasn’t even the biggest issue with the Model 3 production, but it was on theme. The car was announced in early 2016 and was meant to turn around the public’s perception of the company’s history of being dogshit at manufacturing. The Model 3 is the cheapest model of Tesla (announced as starting at $35k) and was the first car that Tesla designed while actually considering that it would need to be built at scale eventually. Instead of taking their time or taking advice from existing successful manufacturers, Musk told his team that they would build half a million units before 2018. Again, within less than two years, this car would go from a prototype to half a million existing units. In Q4 2017, Tesla delivered 1,542 Model 3s. That is not quite 500k.

Elon Musk even admitted that the company was overly reliant on “excessive automation.” Automation has been the key to success in the auto industry since Henry Ford and Hitler were pen-pals over their shared love of industrial processes and hatred of Jewish people, but the company took shortcuts instead of advice from people who knew what they were talking about, leading to a year of delays and issues so severe many through the company would not make it.

The big problem with Tesla for such a long time was that there were they could not make enough cars to meet the hype that they whipped up. That is a very different problem from the current one, with sales declining not only due to the sieg heil and all that business, but also due to increased competition. As I go, I’m drawing in part from Ludicrous: The Unvarnished Story of Tesla Motors by Edward Niedermeyer, but this book was published in 2019, so it’ll only take us so far.

Tesla does exist for a reason. Most of the major automakers are like a hundred years old and massive behemoths. They are deeply risk-averse and profit-driven, so innovation tends to come only when forced. Blue-sky R&D includes lighting a whole bunch of money on fire and hoping things turn out okay, which most corporations are unwilling to do. This problem is not unique to the auto industry. Take a look at the biggest movies that came out last year. Almost all of them are sequels or come from existing IP.

Capitalism also tends to lead to a process similar to carcinization. This is the evolutionary process where everything eventually becomes a crab. Because many markets, like the auto industry, have a small number of massive players, these companies tend to prioritize mass appeal. This leads to many vehicles converging to look pretty much the same. This is exacerbated by regulatory loopholes or cost-cutting techniques that make certain designs more profitable.

The auto industry only changes when it has to, and typically off the backs of other’s work. For example, when the Citroën Traction Avant was released in 1934, it was the first car that was mass-produced with front-wheel drive, four-wheel independent suspension, and unibody construction. All this effort being poured into innovation made Citroën go bankrupt the year the Traction Avant was launched. André Citroën, who ran the company, actually died a year later as a broke boy, but the firm went on to be fairly successful for about twenty years after his death. This was until the entire industry adopted most of the innovations and Citroën lost its’ first-mover advantage.

Tesla kinda did this with the electric car. Although EVs existed 100 years ago, their capacity was incredibly limited because we didn’t know how to make good enough batteries. Tesla was able to demonstrate that that was no longer the case. It created proof of concept, nearly went bankrupt just as some of the innovations came through, and then went on to be fairly successful for a while. That is starting to change, and it hasn’t even been as long as it was for Citroën, and there are some obvious reasons why.

Tesla was incorporated as Tesla Motors by Martin Eberhard and Marc Tarpenning in 2003. Eberhard was a hardware guy and Terpenning was a software guy, and the two had previously collaborated on an early Reader, primarily on the battery improvements necessary to make such a device. Driven by a passion for climate and sports cars, the pair then applied this to automobiles.

Musk hopped on the company during series A funding about a year later. By 2007/2008, Musk pushed the pair out. After, Musk used his money and power to make it so he was legally allowed to be called a co-founder. Musk’s big claim to being a founder was coming up with the strategy to make an electric sports car, and then use the profits to make progressively more accessible vehicles. Funnily enough, Eberhard and Tarpenning published a white paper with that idea years prior.

Roadster development began largely without Musk, as this was during a period when he was primarily focused on SpaceX. The idea came from combining technologies from the experimental, handmade AC Propulsion tzero vehicle with Eberhard’s lithium-ion batteries and putting it all in a Lotus sports car body.

The tzero was functionally a glorified go-kart, but it could hit zero to sixty in four seconds. Eberhard personally invested in the company and suggested they switch from lead–acid batteries to lithium-ion batteries. Even with the battery upgrades, the vehicles were still basically handmade go-karts, just with longer range, so to improve safety, comfortability, and manufacturability, the Tesla folks teamed up with existing sports car company Lotus.

Lotus made small, lightweight sports cars, which would help make up for the additional weight of the batteries in the new electric vehicle. As the prototype process began, many issues arose. The battery cells were too large for the existing frame, so the chassis needed to be elongated and stiffened, causing the entire thing to be redesigned. By the end of the redesign, only 7% of the original vehicle remained. Not exactly a simple conversion. Musk’s primary inputs on the design were to redesign the door sills, add custom headlights, retool seats for comfort, add electronic door handles, and add carbon-fiber-reinforced plastic. These changes added to the already high manufacturing difficulties and costs. Generally speaking, early Tesla was not concerned with costs, and Niedermeyer refers to this issue as ‘elegance creep.’ The company did not have a CFO or centralized enterprise management software to keep track of expenses. They did, however, have plenty of money from Musk to burn.

After finishing a prototype, Tesla started doing launch events and demonstrations, both so that they could take preorders and prove the product could sell to investors. At these demonstrations, the prototypes had to be taken behind a curtain and pumped with ice water to cool the battery between uses, and barely survived through one day of use. Despite this, the press coverage was overwhelmingly positive and the first deposits of $100k started rolling in.

This press coverage, however, focused primarily on Eberhard and credited him with the success. Musk also spoke at these events but was much less articulate, and the fact that the press wasn’t giving him enough attention drove him bananas. Musk dedicated himself to getting the attention he so dearly craved. He needed to get rid of Eberhard so that he wouldn’t need to share the spotlight. Musk published his ‘Master Plan’ that largely resembles the Eberhard and Tarpening whitepaper as a blog post, taking credit for the company. Shortly after, Eberhard and Tarpening exit the story. Elon making grand announcements for attention will become a theme.

Now with Musk at the helm, Tesla was making big promises on both timeline, cost, and innovation, but the endless redesigns and additions not only burned through cash but also pushed timelines back. This will also become a theme. The first roadster was eventually delivered in February 2008.

Tesla says that they are not a car company, they are a technology company. This is part of their ethos. Because of this, and the small-scale, boutique style of manufacturing at the start, the quality of Teslas has always been bad. By quality, I don’t mean how ‘good’ or cool the car is, but quality in a manufacturing sense: how consistently are you able to make the product? Since the company has a software ethos, they are not as focused on process. As I talked about in my little intro, in software engineering, if something breaks, you can fix it at little expense and typically quite quickly.

In the early years, Tesla focused on seeming dynamic and innovative. The American automakers were like this in the early 1900s but learned how to prioritize quality after getting their asses kicked by the Japanese in the 1960s. Although much of the foundational ideas are from an American guy named Ed Demming, the Toyota Production System (TPS) made Japanese automakers a dominant force. TPS focuses on careful planning and constant improvements to make manufacturing as efficient, consistent, and waste-reducing as possible. Because of this, the Japanese automakers smoked American companies until they incorporated TPS in the latter half of the 20th century. Read more about that from MIT and Yale, but we will come back to the idea of an Asian auto invasion later.

Tesla never really got into TPS in the same way, and it certainly didn’t care much about process in the early days. Early Teslas were made quickly and inconsistently and tended to have a lot of technical problems and frequent breakdowns. This was seen as acceptable since the early clientele were rich people with multiple cars that did not depend on the reliability of an individual car to exist in society. Plus, Tesla offered a fancy, white-glove service that would come to you and fix anything that broke down. Very cool, very silicon valley. Not good at scale.

Because of the high expenses and small volume of products, Tesla nearly went under around the point of the roadster launch. Niedermeyer reports that by the end of 2008, Musk began to dip into roadster deposits to pay for operational expenses. He laid off a quarter of the company to try to cut costs. Just as things seemed especially dire, Tesla announced that 40 million in funding was secured.

The silly thing is, evidence now shows that this funding was not official until two months later. This is not the last time the 'funding secured’ line will pop up. About a year later, the company was still in financial trouble, but they were able to secure a $465 million loan from the US Department of Energy. According to Niedermeyer, Musk sent an email out to investors months before the loan was secure, again with the verbiage ‘funding secured’ and whipping up hype. A FOIA request revealed that the first loan application was denied for inadequate evidence as this language was being used. Musk would eventually use this wording most famously with the “Am considering taking Tesla private at $420. Funding secured,” tweet in 2018. Fool me once, shame on you. Fool me twice, shame on me. Fool me three times, get sued by the SEC. This lawsuit would lead to Musk being forced to step down as Tesla’s Chairman for three years and Musk and Tesla each paying a separate $20 million penalty, but Musk was able to avoid further consequences in court.

Tesla’s first profitable month was July 2009, and it claimed it was profitable due to the success of the roadster. In reality, when the paperwork was filed for Tesla to go public, it was revealed that the 1 million dollars in profit for that month likely had more to do with the sale of ZEV credits. ZEV credits, or carbon credits, are an attempt by the government to get auto manufacturers to move towards low-emission vehicles. If a car company does not make enough low-emission vehicles, they must buy credits from a company that does. Since Tesla only makes electric cars, every car sold has the additional pure profit of selling one of these credits to another company. 85% of the company’s gross margin for 2009 came from ZEV credits. Even into the 2020s, up to 60% of Tesla’s profits each quarter come from these credits. Much of what allowed the company to become profitable was functionally allowing other companies to pollute more and escape regulation - not exactly environmentally heroic. The company has received roughly $10.7 billion from this program, which Trump has indicated he plans to end. Classic pulling the ladder up behind you move.

Tesla went public in 2010 and was the first car company to go public in the US since Ford in 1956. The DOE loan in part contributed to a rather muted IPO. This was because the loan included a clause stating that in 2018 the US government would be allowed to buy more than 3 million shares of Tesla stock at $7.54 per share, diluting shareholder value. The DOE loan also barred Musk from selling Tesla, which is why it avoided the fate of nearly all startups at this stage of capitalism.

The model s also had a whole bunch of production problems. It would be redundant for me to describe them, but it involved a bunch of manufacturing issues and technical problems, including the propensity to light on fire. The company was again about to run out of money as it did during the roadster process, so the company changed its reservation/deposit system so that it would get more money upfront.

Originally, the model s was supposed to include a more affordable option with a smaller battery to keep the price at 50k. This was part of the pitch to receive the DOE loan that saved the company. Instead, the company did a scheme where you could only order vehicles in production through most ordering channels, which did not include the cheaper version. The company used the lack of deposits on the cheaper version, which most people did not have a way to register interest in, to scrap the cheaper model.

Due to some likely financial shenanigans and some legitimate increase in interest in the cars, Tesla had its first-ever profitable quarter in Q1 2013. This made the stock shoot upward. The money from this was used to pay the DOE loan off early, allowing Tesla to get out of the stock deal that was included in the loan terms. If the loan had stayed in place, the US government could have seen about a billion in returns. Instead, they got their money back with a little bit of interest.

The hype went crazy and the stocks went along with it, and now with the DOE loan out of the way, nothing was anchoring the stock price. Fanboys started investing in the company and boosting the stock even further in the meme stock fashion that we all now know and love. The company has been known to use non-generally accepted accounting principles to make things seem better than they are. Anyone who looked at the fundamentals of the business and questioned why the stock was so high was shouted down by Elon and his new army of financially interested dick-riders.

This is the point in the story where we reach an inflection point. For most of Tesla’s early years, the primary issue it had was that it had high demand for a product it was unable to produce. Now, we are starting to see that Tesla can make cars much more consistently, and for a while things were going well.

The Tesla model 3 and model y account for the vast majority of Tesla sales worldwide. By vast majority, I really mean it (see the chart I made below). These cars have been very popular and were very successful during the pandemic when many other automakers were struggling. Tesla is more vertically integrated and has less complicated supply chains than the major automakers, meaning they were less impacted. Because of this, they saw huge profits for a few years. But instead of reinvesting the money into improving the existing products or developing new ones, Tesla paid its executives $2.5 billion during its successful period.

Tesla’s product line is getting dated in an industry that is rapidly expanding and increasingly competitive. By the time of writing, the cost for each base model is $34,990 for the model 3, $52,490 for the model y, $79,990 for the model s, $84,990 for the model x, and $99,990 for the cybertruck. The two cheapest are by far the best sellers, but the plan to expand downmarket is getting scrapped to pursue non-core products that we will discuss later. Tesla has been claiming that it will eventually produce a model 2, an even more affordable version of the model 3, for years now, but those plans have constantly gotten pushed back. On top of this, the model s and x have not been seriously updated in over a decade. The only refresh planned for this year is a slight design update to the model y.

The increased competition has caused Tesla to enter price competition that it had never faced previously and has caused profits to fall. Tesla is losing its first-mover advantage, and it risks facing the same fate as Citroën even without all that other stuff.

Getting into the truck market as an automaker makes so much sense. It is one of the most profitable sections of the market. New electric vehicle startups like Rivian have started with trucks because of this. Tesla making a new electric truck coming off of their most successful years very well could have cemented the company into a pretty secure state. The profits from the new truck could fund the innovation necessary to create the new, cheaper model 2. This would lead to larger sales volumes, and the smaller profit margin on the cheaper car would be fine because of the massive increase in ZEV/carbon credits the company would earn. Great situation, great plan! This is not what happened.

The cybertruck is a great example of what happens when you are not allowed to tell someone no. Elon came up with the concept and a bunch of engineers hated it for a whole variety of reasons. It still got made the way Elon wanted, and now we are where we are. It is objectively poorly performing. Musk stated that he thought that Tesla would sell 250,000-500,000 cybertrucks per year. Instead, the company sold less than 40k in 2024, or about 8-16% of Musk’s projections. On top of the low sales, the product is also of poor quality. The infamous cybertruck demonstration fail was only a preview of the nonsense to come. The cybertruck has been on the market since Q4 2023 and has already had eight recalls, with the most recent one including virtually every cybertruck ever sold because the stainless steel panels keep falling off the truck.

One of the big claims of the cybertruck is that it would be incredibly safe, because “if you’re ever in an argument with another car, you will win,” according to Musk. This is a general problem in the auto industry, where vehicles keep getting bigger in a sort of arms race that leads to more fatalities overall. Personally, I think that some regulation needs to come in and force some arms control, but let's not get too sidetracked. The cybertruck has caused some pretty gnarly deaths due to its ‘indestructibility.’ Infamous Kentucky Senator Mitch McConnell’s billionaire sister-in-law died after accidentally reversing her cybertruck into water and a few college kids burned to death in a cybertruck with doors that wouldn’t open. Not exactly good press.

On top of this, the sales of the cybertruck seemed to be almost directly negated by decreases in sales of the model s and model x. You can see this in the charts above, with the ‘all other model’ lines not increasing as the cybertruck is introduced. This suggests that the high-end of Tesla’s market is probably limited, and creating a new expensive model just canceled out the purchases of the other expensive models. It could also be because the low quality of the cybertruck has damaged Tesla’s high-end reputation, and people no longer feel that their products are worth the high sticker price.

I’m not going to go too in-depth about the Tesla electric semi-truck. Building an electric semi truck does not make sense at this point. The point of semi trucks is shipping, and usually long-haul shipping at that. Electric vehicles don’t have the range for this yet and there is still a lack of charging infrastructure in enough places that this business model is years away from making sense to pursue. They announced this vehicle in 2016. This was obviously just to get attention and they have not gotten past the prototyping and testing stage almost ten years later. Even so, one prototype caught on fire and shut down a California freeway for a while in 2024. Ah, Tesla and their fires. Now to get into some of the Tesla stuff that isn’t cars.

Speaking of charging infrastructure, the initial appeal of Tesla for a lot of people was the environmental impact. This is partially negated by the ZEV credit scheme we previously discussed, but on top of that, Musk made a big deal out of the energy being zero-emission. The problem, though, is that the energy being used has historically not been zero emission. Some folks call this the ‘the long tailpipe,’ where emissions are not really reduced but just moved to a different point in the process. New technologies, like solar panels, are also made using processes that emit pollution, and the level of pollution is even higher in places where they are made more cheaply like China, but developing these technologies and then figuring out how to make them even greener feels like the practical solution to me. Plus, even with the long tailpipe problem, most of these transitions are still a net benefit either way.

Despite this, Tesla claimed that all their new superchargers would be solar-powered. They started to say this in 2015. It isn’t even true now. Again, in my opinion, this isn’t even that big of a deal, but it is wild that they just choose to exaggerate, overpromise, and underdeliver for no reason and everyone decided that that was okay actually. On top of this, Tesla has been investigated by the SEC for the fact that their solar panels keep lighting on fire.

For a while, Tesla was saying that it would offer not only its superchargers, which are legitimately a great infrastructure project but also would offer battery swaps for folks in a rush at these stations. They did a big demonstration to show that it was possible, then claimed to set up a small pilot program for folks in California. The program was not made publicly accessible, and then when people didn’t use the thing they didn’t know existed, the company claimed there was not enough interest to expand the program. In reality, it was never feasible for a whole bunch of reasons. This demonstration that the vehicles were technically capable of ‘fast refueling’ allowed them to claim double the ZEV credits under California’s laws.

As a quick aside, Canada’s electric vehicle rebate program recently lapsed. It will probably get restarted after the elections that are about to happen, but as the government swaps over, electric vehicles are not gaining credits. The weekend before this lapse happened, 4 locations in Canada claimed to sell over 8,000 vehicles in 4 days. That is more than one Tesla sold per location per minute. These Teslas would have generated $43 million in rebates, but due to the suspicious situation, tariffs, and Canada’s recent liberal-nationalist turn, these rebates are being frozen. On top of that, Tesla is being blacklisted from future rebate programs, both at the federal and provincial levels. Canada is getting feisty.

Anyways, Tesla superchargers are one of the best things the company has built and a massive competitive advantage. Despite that, Tesla cut its entire supercharger team in 2024, firing hundreds and hundreds of people who were doing great work. This has since caused a massive slowdown in supercharger expansion for obvious reasons, and Tesla has since started trying to rehire people that they had laid off to at least maintain the superchargers that already exist, especially with the firebombings and such.

Hmmm, firing huge teams of people, realizing it is a mistake, and then trying to rehire at least some of them? Have I seen this in the news recently? Maybe something to do with the government? Certainly, it would have to be someone who didn’t have an obvious scenario like this they could have learned from so as not to repeat it. Also, man, it makes so much sense to treat the government like a business and there are no risks or downsides to doing that, am I right?

Speaking of treating workers like dirt, Tesla is known to be one of the most dangerous employers in the industry. In 2015, serious injuries on the job happened twice as often as the industry average. When the injury rate started to decline, investigations found that Tesla was not properly reporting injuries. Even in the last few years, workers at the Texas factory experienced explosions, concussions, and grisly robot injuries. On top of this, there is a history of sexual harassment claims, multiple instances of alleged retaliation for whistleblowing (example, example), and cases of intense racism and racial harassment.

Musk has made a bunch of wild claims about how bad unions are. Which, I mean, of course he hates unions; they directly threaten his power to do whatever he wants. Tesla has such a bad record on union busting that it literally has its own wikipedia page titled “Tesla and trade unions,” which is separate from a different wikipedia page, “Elon Musk and trade unions.” The company has had 63 cases brought against it by the National Labor Relations Board (NLRB) according to their database. One of the biggest was the unionization push in 2017 that Musk fought back against. Similar battles are being fought at Tesla factories in Germany and recently in Sweden.

Although I care about this less, it is also important to note that working directly with Elon seems to be a particular nightmare. Executive turnover at Tesla in 2019 was nearly 5x the rate of comparable companies, and this trend has continued into 2024. In general, Tesla has a much slimmer C-suite, with only a CFO, CTO, and CAO. The leadership jumps from CEO almost immediately to vice president roles. In comparison to a company like Ford, this gives Musk way more power over Tesla.

It is absolutely wild that Tesla has no COO and the CEO has like seven other ‘jobs’ he is doing all the time. The company does take autopilot seriously, I guess.

In 2013, Google almost bought Tesla for $6 billion. This was in part driven by Google’s interest in self-driving technology. It had started a project to develop autonomous vehicles in 2009, which was being driven by Sebastian Thrun, a professor from Carnegie Mellon and then Stanford.

In the end, the deal fell through probably for a bunch of reasons, but as far as the self-driving car project, Musk decided not to work with Google on this project because he disagrees with their use of lidar and radar sensors. Musk wants to go for full self-driving (FSD) with just cameras. The Google folks, who are now the Waymo folks, use extra sensors, including lidar and radar sensors that allow for data to be collected in a process similar to echolocation that adds a layer of safety. See a good visualization here.

There are three reasons why Musk doesn’t want to use the extra sensors. First, when he first committed to this take, they were really expensive. Like, in the tens of thousands of dollars range. This is no longer a very valid reason because the cost has gone down by about 90-95% in the last decade or so as the technology has been developed. The price is still expected to go down even further to the point the cost will be nearly negligible.

The second reason is that Musk keeps claiming that FSD is only a software problem. He kept saying that once the perfect software has been completed, it will only take a software update for every Tesla sold since 2016 with their FSD package to become autonomous. This means that everyone who already owns a Tesla can get FSD without having to get any tech replaced or updated on their car. Musk has admitted these claims are untrue, and Tesla is currently getting hit with class-action lawsuits claiming fraud.

Because of this appeal, Musk has been saying that FSD is coming ‘later this year’ like every year since 2016. Someone even made a fun timeline with all his quotes on the subject from 2014-2022. Such a great example of how this man just keeps making insane claims for attention and not following through.

The last reason that Tesla folks claim they don’t want to use sensors is that it makes the computational task more complex. Adding lidar and radar sensors means that you have to process and understand more data and noise within the computer ‘brain’ driving the autonomous vehicle. This feels silly to me. Increased complexity seems like an obvious short-term engineering problem that will go away over time like the price issue. Truly innovative tech takes work, and it will take more work to make the more complicated system work, but the ceiling is so much higher. This is especially true in terms of safety, and when people’s lives are on the line, it is absolutely better safe than sorry.

This claim is likely made because Tesla’s current competitive advantage with Waymo is how many miles of data they have from their current fleet. According to Tesla, about 5 million Teslas are on the roads with FSD hardware. This generates training data that can hypothetically be used to train new AI models for FSD. The problem is that it isn’t clear how useful this data is. On top of that, it is possible to train these models on ‘simulated miles,’ which reduces the need for this kind of data. Even Tesla uses simulations to train their vehicles on top of the data that they have. In the web 2.0 and especially the AI era of tech, data has become a primary cash/power source, but in this case, it doesn’t seem to be as clear of an advantage.

The first ever fatal crash including a fully autonomous vehicle happened in January of this year. It was a Waymo. The goofy thing is that the Waymo was empty and not moving in a line of standstill traffic. A person driving a car ran into it and other unmoving vehicles at full speed, killing one woman and her dog. Super sad story but you really cannot blame the Waymo for that one. On the other hand, Tesla’s autopilot has killed an estimated 54 people (more detail). Experts suggest that Tesla’s disinterest in using additional sensors has led to additional crashes and near misses, but on top of that, the company is very okay with taking additional risks and not being clear in their communication about what the real capabilities of the tech are. Over promise, under deliver, even if it kills people.

The National Highway Traffic Safety Administration (NHTSA) has come up with an automation scale from level 0 to level 5 with defined capability categories. Level 5 is full automation, but things like cruise control, lane keep, and automatic braking for accident protection fill in the middle levels and are fairly common in new vehicles. To some extent, the name autopilot itself is meant to make the system seem more impressive than it is. Whereas other companies tend to use terms like “assist,” “sense,” or “team,” Tesla comes up with another name that seems more flashy. A study done after the release of the program showed that users were not at all clear about the level of automation that the product offered. When the system got an update, they started calling it Full Self Driving, but the product is only at level 2 automation. Now Tesla has to call it Full Self-Driving (Supervised) to avoid legal liability, but the name is meant for hype and makes people less safe.

Musk and Tesla only exacerbated this. In a 2016 demonstration, Tesla claimed “The person in the driver’s seat is only there for legal reasons. He is not doing anything. The car is driving itself.” Again, this isn’t even true now. This has led to a problem where people believe that they can rely on autopilot more than is safe, leading to more crashes. To make things even worse for a while, the NHTSA compared Tesla to other companies and found that Tesla’s design choices discourage active driving and Tesla was forced to recall 2 million vehicles in 2023. There are also rumors that Tesla autopilot will disengage if it thinks it is about to crash so that if it does crash, the incident will not count as an autopilot failure. To be clear, any crash within 5 seconds of disengagement of autopilot counts as an autopilot crash, so this may or may not be the case.

Despite all this safety bullshit that is obviously not that important, Tesla is planning on launching a robotaxi service in Austin, Texas in June. Like, in two months, June. This is particularly ominous because Tesla’s head of software engineering and development who has been at the company for 12 years announced he is stepping down as I am writing this.

Even if the robotaxi program works well, I’m not sure that banking everything on it is a good idea. Tesla’s value is very unstable and has been cratering, but the current value of the company is $770 billion. As of 2024, Waymo was worth $45 billion, and it is already a functioning robotaxi company. On top of this, the idea that the robotaxi service is going to be great for Tesla’s stock price feels very hype focused, as Google owns Waymo and its stock does not seem to be impacted that much by its robotaxi success. They just need to feed the hype cycle.

Tesla thinks that they are different from Waymo, however. Tesla’s proposed idea is that if the cars hit level 5 automation, every current Tesla owner will immediately have their own robotaxi. The Tesla owners can then have their cars doing rides for profit while they are at work or sleeping. This has been disproven and the cars will definitely need hardware updates, as previously discussed, but even with the updates, it seems clear that this business plan is not thought through. Most people probably wouldn’t want their very expensive car to be used by random people unsupervised, and the wear and tear on the vehicles would likely be tremendous. People tend to heavily mistreat and even vandalize Waymos since no one is there to stop them. I’m not sure the uber/lyft-type income would be worth it to people based on the risk to the asset, but I could be wrong.

As I’ve said, silicon valley folks tend to be software-minded. On top of this, they also tend to value all-encompassing singular products rather than specialization. You can see this in the race for genAI. The way the Chinese model, deepseek, was able to be so much cheaper and use less energy than the US models is kinda related to this. Deepseek uses something called the Mixture-of-Experts (MoE) model, where a bunch of small AIs are trained to be really good at specific tasks, and then they are combined to create deepseek. This process uses much less energy and requires less input. The US models are just one big model that is very expensive and trying to do everything at once. That is the silicon valley approach to problems.

This is especially true for robotics. silicon valley folks tend to have this idea that the future of robotics is humanoid robots that can do whatever people can do. Most people I know who work on robotics think this is a pretty silly approach. Humans are actually not very well designed for any particular task, and trying to make a robot good enough at everything people can do is not plausible in the short term. The most useful robots currently and in the near term are highly specialized.

Even people who are all in on humanoid robots think it could take decades for the tech to work. Elon Musk says that Tesla will build 10k this year. They have not successfully built one yet, with some researchers in the field referring to the pitch as a "complete and utter scam."

Tesla started to see sales start to slump and profit margins shrink in the last few years. This is largely because of increased competition, and that increased competition is coming from all over the world, but is being driven most intensely by China. During Tesla’s successful years, they had a sweetheart deal with the Chinese government. Most foreign automakers are forced to work with a Chinese company to be allowed to sell within the country, with the lone exemption of Tesla. On top of this, Tesla got lower corporate tax rates and subsidy money to the tune of billions from the Chinese government. Tesla was even made an “official Chinese government car” and the company was “one of 16 car makers – the only non-Chinese manufacturer – to sign a four point pledge that included an agreement to adhere to ‘core socialist values.’” Not exactly the image Musk has created of himself in the West, huh? Some folks think this dynamic makes Musk a security threat to the US, but I think he manages that just fine even without pressure from the Chinese.

The Chinese auto market is very complicated. In 1986 the government decided to make the auto industry into a “pillar industry” and began the joint venture programs, which led to the development of the traditional Big Four from China: SAIC, FAW, Dongfeng, and Changan. In the last two decades or so, reforms in the government made it so that there are now not only centrally owned state-run automakers, there are also state and local government automakers, privately owned automakers, and the remaining joint venture foreign cars. There are currently more different companies selling cars in China than in the US, making the market fiercely competitive. This has pushed prices lower, aided by government subsidization in many cases, but has also led to significant improvements in manufacturing quality. Many of the Chinese companies will start to fold or exit as the market becomes more developed and the winners start to emerge. One of the current darlings that is looking like a big winner is BYD.

Build Your Dreams (BYD) was started in 1995 as a battery manufacturer. They started making cars by reverse-engineering foreign vehicles, which has been very common in China. They started developing hybrid and fully electric vehicles in the 2000s, and similar to Tesla, they did amazingly well during the pandemic, experiencing ten-fold growth. This is due to a variety of factors, but in a lot of ways is simply because they make pretty good cars for not a ton of money.

A great example of how a BYD car is different from a Tesla is the battery. BYD uses lithium-iron-phosphate batteries as opposed to Tesla’s nickel-manganese-cobalt batteries, which have shorter driving ranges but are safer (less likely to light on fire), last longer, and are much cheaper.

Car reviewers seem to mostly say that Teslas are a more premium-feeling car. BYD uses cheaper materials on the interior, and many of the more techy aspects are less of a focus, but depending on where you live in the world, BYD is just so much cheaper for a fairly comparable product. The price of a BYD vehicle varies wildly by country depending on tariffs and things like that, but in many countries, it wouldn’t make sense to buy a Tesla over a BYD. The difference in quality does not make up for the price difference, especially in China.

This has led to Tesla trying to cut prices in China to keep up. This turned into a full price war over the last few years and has led to Tesla losing ground in China. The really silly thing about this is that the sales decrease in much of the world seems reasonably tied to Musk being a public menace. The types of Europeans and North Americans who buy electric cars do not want to be associated with the brand for political reasons, and that is where most of the non-Asian market is. That is not the case in China. Folks in China don’t care about Musk being a freak, and the brand's reputation is fine there. They just are buying fewer since they have better options. BYD is outselling Tesla even without any of the noise.

Recently, BYD also announced that they have developed new technology that allows their cars to be fully charged in five minutes, whereas it takes about an hour using a Tesla supercharger. Just another way that BYD is eating Tesla’s lunch.

I have very little faith that any of Tesla’s big gambles are going to play out for all of the reasons that I’ve said. On top of this, folks at Tesla also don’t seem to have much faith that things are going to go well. If you look at the stock trading patterns of the folks at Tesla, you can get a REALLY clear picture of how they are feeling. In the last 12 months, 0 shares of Tesla have been purchased by Tesla insiders. There is typically more stock sold than bought because folks tend to get paid in stock options and then offload, but I’ve added NVIDIA and Ford as references for what these numbers look like at some other companies that people may compare to Tesla. So if my arguments haven’t been enough to convince you that Tesla is in the hole, maybe take their own actions as evidence.

Thanks for reading kristin's substack! This post is public so feel free to share it.