Tesla’s real problem isn’t that its cars are expensive. It’s that they’re unreliable.

Few companies have enjoyed more hype over the past few years than electric carmaker Tesla. And not without reason: Tesla is the most successful automaking startup in decades and has almost singlehandedly made electric cars cool.

Yet the automaker has also been struggling with the quality of its vehicles. On Thursday, the National Highway Traffic Safety Administration revealed it was investigating a possible problem with the suspension of Tesla's flagship Model S sedans. The reported flaw is the latest in a series of quality and reliability problems with Tesla's vehicles.

Industry analyst Edward Niedermeyer, who blogs at the Daily Kanban, argues that Tesla's challenges with vehicle quality are only going to get worse in the coming years. Tesla is preparing to release the Model 3, whose modest $35,000 price tag is designed to appeal to mainstream customers. And Niedermeyer argues that middle-class customers are less forgiving of quality problems than the wealthy customers Tesla has served so far.

Tesla's basic problem, Niedermeyer argues, is culture. Industry leader Toyota conquered the American car market with a rigorous manufacturing process that emphasized quality and reliability above all else. But Tesla has a freewheeling Silicon Valley culture that values innovation and creativity over reliable execution.

We spoke by phone last week. The conversation has been edited for length and clarity.

Timothy B. Lee: You're skeptical that a Silicon Valley company like Tesla can become a major player in the auto business. Why do you think cars are different from software?

Edward Niedermeyer: To boil it down to the most essential issue, it's a question of scale. With software, you have a fixed cost of development that is oftentimes quite high, but once you have a viable product and you pay off that fixed cost, your variable cost to scale beyond that is almost nonexistent. You're literally just copying code.

With automobiles, not only do you have immense fixed costs in research and development, tooling up factories, creating testing, but once you've done all the development work for a car, you still have a process of scaling. Not only are the variable material and labor costs much higher than in software, but you also have a lot of details that can go wrong.

Cars have become so reliable and so easy to use that we think about them less than we ever have in the 100-plus-year history of the automobile. This is one reason we don't appreciate this depth of complexity. Not only are cars different from software in very fundamental ways, they're much more complicated than anything else consumers buy.

Cars use a wide variety of materials, built into components and subassemblies by massive global supply chains. Car companies have to choose and develop the right materials and components, maintain their uniformity and integrity throughout that supply chain, and ensure that they operate reliably in almost every imaginable condition on Earth.

A great example is the problem of mold growing from inside the Model S's roof, particularly in Norwegian cars. Because its large panoramic sunroof is difficult to manufacture and install to a precise specification, Model S roofs often leak. A lot of those leaks are so small that customers might not notice. But because Tesla used an organic-fiber pad at the edge of the sunroof, aggressive molds invade at alarming rates in certain climates. This kind of complex, cascading defect is why automakers value their accumulated institutional knowledge and spend years testing vehicles.

TBL: It seems like Tesla's early cars — the 2008 Roadster and the 2012 Model S — were lauded for their innovative designs and were well-received by customers. But you're skeptical about the Model 3, which is more affordable and aimed at a mass market. What's the difference?

EN: It's a common misperception that the more expensive the car, the more people expect out of it. The opposite is true: the cheaper the car, the more people tend to rely on it, and the more reliability and quality come into play.

The car business is a very risk-averse business by nature. It's capital-intensive and relatively low-margin. The first 50 years or so after the Model T (in 1908) was focused on technology development. People pushed the limits in terms of power, styling, futurism.

Since the 1970s or so, it's reverted to a kind of more pragmatic, utilitarian mode. The market has become more mature. Toyota and Honda have really made their names on quality and reliability, not exciting futuristic values.

In a lot of ways, Tesla is a throwback to an earlier era of the auto industry. They tap into the idea that there is new technological space to be conquered; you get there by focusing on performance, on building a very attractive, appealing car. That's what Ferrari and Lamborghini did between the 1930s and the 1960s.

I think that parallel is worth looking at, because neither Ferrari nor Lamborghini is known for quality. If you operate in the high end of the market, consumers appreciate performance and design. If their Ferrari or Lamborghini breaks down, they have their chauffeur take them in a Mercedes or a Lexus.

It's not the end of the world that Tesla's quality has been bad so far, because they're operating in a luxury space. But as they move down market with the Model 3, reliability and quality are going to be real issues. The level of quality they've achieved in the Model S is not going to be sufficient to succeed in the $30,000-to-$50,000 price range.

TBL: Is it really that hard to improve manufacturing quality? Elon Musk is a smart guy, and he recently put his desk at the end of the Model X assembly line so he can personally keep an eye on the progress there.

EN: Anything is possible. Obviously they've proven doubters wrong before. So I'm not going to say that it's impossible for them to do it again.

But raising quality is very different from the things they've done to gain market position so far.

One parallel that's worth thinking about is General Motors. This was the most successful car company in the world for the better part of a century and the most valuable company for a period of time. Then they got surpassed on quality in the 1970s, and they still haven't caught up.

Photo by Justin Sullivan/Getty Images

So the question is what makes the quality of Japanese companies? I think you can trace that back to Toyota, which developed the Toyota production system and — just as importantly — a broader corporate philosophy called the Toyota way. It systematizes everything about the production of automobiles.

For example, the need to keep plants operating at a high rate meant you'd let defects go down the line and fix them at the end. One of the things Toyota did was when the defect came down the line, you stop the line and you trace the defect back to its root and fix it, then you restart production. This is just one example.

To this day, they're still the leaders in quality. Nobody has caught up with them.

TBL: But GM in the 1970s was a big, bureaucratic organization tied down with a lot of union rules. It seems like it should be easier for a young and nimble company like Tesla to pivot and adopt a more Toyota-like production philosophy.

EN: It's certainly more likely because they're not at the point where Ford and GM and Chrysler were when they faced that challenge — they already had tens of thousands of workers and faced much more inertia.

But in the first half of the 20th century, Detroit was the equivalent at the time of Silicon Valley today. Yes, Tesla is a startup culture, but they demonstrate an arrogance that is similar to the arrogance that Detroit demonstrated in the past. When a culture works really well, there's an assumption that it can be universalized. I think that shows in the thinking that Silicon Valley culture will apply to manufacturing.

But what does startup culture emphasize? It emphasizes flexibility, individual effort, and working long hours to reach ambitious goals. What it's not is regimented.

But the only way to make money in cars is at huge scale. And scale creates immense complexity. And as you go up the volume scale, it becomes more challenging. So if your goal is to make your car company a mass-market player, you have to bake in the regimentation and the production system from the get-go.

When I say companies are risk-averse, it's because success in the car business is not about reaching out into the unknown in order to achieve unprecedented things. It's about driving waste, inefficiency, and defects out of your production machine. That is what Toyota's innovations enabled it to do. It systematized every aspect of development and production.

TBL: What do you think about the approach Google has taken to the car business?

EN: Google's strategy is the counterfactual that makes me especially nervous about Tesla. Google's core technology is the autonomous drive capability, and I think they have to be closely watching Tesla and the struggles they've had. So Google has hired some very high-profile people from the car business. They have former Ford CEO Alan Mulally on their board. Lawrence Burns, the former research and development boss for General Motors, is a consultant for them. The head of their autonomous car program is John Krafcik, one of the auto industry's most respected veterans.

It's a dream team of real tier-one automaker experience. With their accumulated knowledge — and looking at Tesla's struggles — they know that building their own car is a fool's mission. They also recognize that Silicon Valley culture is fundamentally different from manufacturing culture.

They realize there are plenty of car companies and car factories in the world. Even before autonomous drive comes out, there's a likelihood that shared mobility will begin to impact demand for cars, leaving spare production capacity for autonomous vehicles.

What fundamentally sets Google apart is that these auto people know how hard building cars is. It is not only an intellectual challenge, it's a discipline challenge. Managing that level of complexity requires a certain amount of accumulated knowledge; building that from scratch is incredibly difficult.

And these systems are already highly automated. It's not like the car guys are doing purchase orders on paper. They already need to be highly software-driven in order to make current levels of complexity work. I'm not sure how much room Silicon Valley has to improve that. If they do, they should develop the capability and sell it to the car companies.

So Tesla is fundamentally an old-school car company. They sell you a desirable, high-performing vehicle that you own. Google is trying to transform mobility without becoming a car company. Their focus is on autonomy. They have the leading position right now in terms of self-driving capability. They're going to continue to build on that.

TBL: It seems like the danger of partnering with existing car companies is that they could be too set in their ways, and too resistant to making the kind of changes that are required for self-driving cars to be really successful.

EN: At first, there was this sense that Google was going to directly take on the car companies and take them out of business. The car companies, without question, have cultural biases that prevent them from wanting to develop autonomous vehicles.

These companies have been around for 100 years. They've only ever sold vehicles to drivers, and the vast majority of their profits came from selling cars to drivers. Because cars are low-margin, you have to find ways to pad that margin, and they do it with things that appeal to a driver, like a more powerful engine or a sport suspension.

Fully autonomous vehicles will be fundamentally different. But Google has singlehandedly pushed autonomy from being a science experiment to something that's going to be viable. They have forced the car industry to accept that things are changing.

So what Google is doing is very pragmatic. Instead of having this existential battle between human drivers versus robot drivers, they've shown that this technology works and argued that they need to work together. It's going to be very disruptive to car companies' business to manage the change from driven vehicles to autonomous vehicles. But they're not fighting it. They're going along with it.

I think this is something that's emerged in the last year. What both sides have realized is that Google can avoid massive investments in very low-margin aspects of the business, while car companies can stay with the times if they work with Google or other startups.

TBL: It seems like we're going to see some big changes in the car industry over the next decade. Which car companies do you see as best positioned to navigate those changes?

EN: You see two responses. You see General Motors and Nissan initially trying to get on this wave of excitement and saying, We're going to set aggressive timelines for autonomous capability. GM made the biggest investment by buying Cruise. I've done a bit of research on Cruise and I'm not super convinced by them. I'm not convinced that GM didn't massively overpay.

Toyota has a much different approach. They are basically making a long-term investment in the research capabilities for autonomous driving. Their deployment strategy is super conservative. They are deploying in bits and pieces. They're starting off by putting low-level semi-autonomous safety functions in all of their vehicles.

That's in part due to their culture and in part due to their experience of the unintended acceleration scandal of 2010, where they were basically accused of having self-driving cars. As far as I can tell — and I spent a lot of time covering it — it was basically bullshit and kind of a witch hunt.

But the legal liability risks are very high. It's very easy for people to make mistakes and blame the car for their mistakes. So in some ways that's an incentive to go full autonomy. But even for a company like Toyota with $80 billion in the bank, there could be liability issues that could challenge the fate of the company. So they are incredibly conservative about deployment, and they will not deploy anything unless it works in 99.9 percent of use cases.

That contrasts with what Tesla is doing, a public beta test. They say they are. They admit it. Frankly, that is an incredibly risky proposition. I think Tesla has a halo right now. I think we have a celebrity news cycle where we tend to build things up and break them down. Once Tesla reaches a critical point in that hype cycle, the public beta test of autopilot software could be part of what destroys them as a company.

I personally tend to like Toyota's approach, because they accept and own the conservative nature of the business. So they don't fool themselves that they're going to do this leapfrog approach.

---

When carmakers taunted horses