Electric car company Tesla held its annual shareholder’s meeting on Tuesday afternoon in Mountain View, Calif., and the event was partly a trip down memory lane, partly a confessional of early missteps, and partly a series of thank you’s to partners and employees that helped build Tesla over its 13-year history.
In an event that spanned over three and a half hours, Tesla CEO Elon Musk and CTO JB Straubel—in addition to a series of Tesla executives—candidly discussed many of the issues and problems that plagued a young, inexperienced, and cash-strapped Tesla in its early days. Musk also talked about more recent problems for Tesla, like challenges with getting the software and doors to work on its Model X, an electric SUV car.
Musk said that he thought Tesla (TSLA) had a 10% chance of succeeding when he started the company, working with engineer and co-founder Straubel. Musk described their early efforts building the company as having “no idea what we are doing,” and “completely clueless.” Musk said he wanted to put in 99% of the Series A funding because he thought the company’s chances were so low that he didn’t want to lose investors’ and friends’ money.
Straubel said the early Tesla employees knew the risks and felt the same way, but that they thought it was “worth taking the risk” if they could make a positive effect on the world.
Musk noted that when the company has made mistakes over the years that it was because they were being “foolish or stupid,” but not because they had nefarious motivations. We’ve always had “the right motivations,” said Musk, adding, “We say the things that we believe even when sometimes those things are delusional.”
The creation of Tesla was based on some false principles, which “turned out to be staggeringly dumb,” Musk detailed, as he regularly referenced a sweeping visual timeline that started in 2003 when he met Straubel and ran to the present. One false principle was that the company’s first car, the Roadster, could be built around core technology from a startup based in Southern California called AC Propulsion, which made drive trains for an electric sports car called the Tzero.
For more on Tesla’s Model X watch our video.
That technology from AC Propulsion could not be reproduced and commercialized, said Musk and Straubel. That was partly because the information system that ran the motor was analog, instead of digital, so engineers had to rewrite all of the code involved. Another reason was because the AC Propulsion battery pack was air-cooled, which didn’t provide enough cooling for batteries that get hot when they discharge to power an electric car. Tesla later converted the Roadster’s battery pack to be liquid-cooled instead.
Another false premise was that Tesla could just use the AC Propulsion tech and stick it in a car body supplied by Lotus Cars, a British car maker. There were major issue with converting the technology to fit it in the small Lotus car body, including that the battery pack was just too big for the car and Tesla had to stretch the car’s body. In addition, the Lotus car couldn’t use the air conditioning system because that relied on the gas engine power, so Tesla had to remake the air conditioning system.
It wasn’t a good idea to simply convert an existing gas-powered vehicle (in Tesla’s case, the Lotus car) into an electric one, said Musk. Changing a car from gas-powered to battery-powered, also invalidated all of the crash tests and safety standards that the Lotus car had already achieved.
In the end, Tesla used 6% to 7% of the original technology from AC Propulsion and the Lotus body union, said Musk. “We needed new all new suspension, all new brakes,” Straubel noted. Musk compared the Roadster development process as similar to wanting to build a house, but instead of building it from scratch, taking an existing house and modifying everything in it but one wall in the basement.
Get Data Sheet, Fortune’s technology newsletter.
The result of this hodge podge of technology, among other things, was that there were serious issues with early Tesla Roadsters. At the end of the shareholder event, an early Roadster owner asked Musk a question, and Musk responded by first apologizing to the customer for technical problems that the owner most likely had experienced.
Musk told an anecdote of how he gave Google (goog) co-founders Larry Page and Sergei Brin a test ride in an early Roadster and divulged that the car wouldn’t drive above 10 miles per hour. “I was like I swear guys it goes way faster than this,” said Musk. The Google founders “were kind enough to put a little money in the company despite the world’s worst demo,” reminisced Musk.
Musk and Straubel launched the Roadster during a launch party in 2006. After giving potential Tesla customers rides all night at the party, “the cars were destroyed” from overuse, remembered Straubel. The team ended up pumping ice water into the cars to cool down the batteries, said Straubel.
One of the biggest issues with the early Roadsters was that the transmission didn’t work, recalled Musk. The car technically passed all regulations, but “was completely unsafe,” it “broke down all the time,” and it “didn’t really work,” said Musk. It was stuck in second gear, he added.
Part of the Roadster technical problems were because of the outsourced and slow production process. In the traditional auto industry, auto suppliers from all over the world will send an automaker parts over many months and the automaker might not know if there’s a problem for maybe 6 months. That means that the automaker might have been selling faulty car tech for a half year.
From 2008 to 2009, Tesla decided to redesign its entire production process, said Musk, rebooting the design of car, the technology, changing out many of its suppliers, and bringing a lot of its tech in house. We had the misguided idea that anything built in Asia was better and cheaper, said Straubel.
Originally Tesla decided to buy its battery packs from a supplier in Thailand that made barbeques, but the technical and organization problems were insurmountable. In 2008, Tesla brought its battery pack production in house to its small operations at the time in San Carlos, Calif. “Manufacturing things in the Peninsula is considered super mad,” said Musk of the move.
Tesla didn’t start selling Roadsters with a positive gross margin until the end of 2009. In essence, Tesla lost money on every car sold before then.
Part of the reason that Musk wanted to air some of Tesla’s missteps was because he said he wants entrepreneurs building new things to know they can make mistakes and start companies in areas in which they don’t have a history.
Tesla is a much more stable company now than it was close to a decade ago. But the company still has faced some major technical issues with its recent cars, in particular with its Model X SUV. The software to control the sweeping doors of the Model X has been very difficult to manage, admitted Musk.
The problems with the Model X likely aren’t behind Tesla just yet. “We’re almost there in making the doors useful,” promised Musk, adding that in a month or so, the doors will likely be better doors than regular doors—but they aren’t right now.
This high-flying lifestyle isn’t actually behind the company. It’s still a core tenant of how Tesla operates.
Tesla is still taking major risks and operating like the nimble startup it once was, despite being a public company with an over $30 billion market cap. Whether that’s still a good idea or not at this stage remains to be seen, but Tesla has managed to survive so far.
Next up for the company: build the world’s largest battery factory, launch a mainstream electric car, and sell batteries to power buildings and the energy grid. No big whoop, right? It’ll likely be some of these lessons learned—adjusting to mistakes, rethinking systems, and the long hours it takes to accomplish those tasks—that could get the company there.