Tesla is always out to sell cars, but this month it’s really out to sell cars: In both the U.S. and China, customers can take a five-year, Tesla-financed loan on one of its electric vehicles at 0% interest. They can get three months of free charging at its Supercharger stations and three months of free use of its autonomous driving software, not to mention an additional $2,000 discount when the purchase comes via a referral from an existing owner.

The condition: The sale has to close by the end of the year.

There’s a big reason Tesla is on an unusual selling binge: For a decade, the company has achieved an unbroken tear of vehicle sales, posting jumps of as much as 130%; its smallest sales increase—in 2020—was 36%, a number almost any other automaker would die for.

Though CEO Elon Musk says Tesla is no longer a carmaker but an artificial intelligence company, developing Robotaxis and humanoid robots, 80% of the company’s revenue through the first nine months of this year stemmed from auto sales. But this year, there is a real chance that Tesla’s EV sales may shrink for the first time since 2010. That would blemish the company’s reputation—and Musk’s superman persona.

To beat its record 2023 sales of 1.8 million vehicles, Tesla will have to sell roughly 515,000 cars this quarter, 30,000 more than its peak number—the 485,000 deliveries it notched in last year’s fourth quarter.

So Tesla is on a drive to make that happen. In October, Musk told analysts Tesla would not shrink. It would not grow as much as before, he said, but it would beat last year’s sales figure.

That won’t be easy: Troy Teslike, the handle of a widely followed Tesla analyst, estimates the company will deliver 488,000 cars in the fourth quarter and 1.78 million for the year, 24,000 units below last year’s number. The company is struggling especially in China, its second largest market, where its sales fell year on year in October and November.

“It is clear that Tesla continues to struggle with demand problems in the U.S. and Europe despite many promotions to increase sales,” Teslike told me via text.

EV sales growth has slowed, leading legacy carmakers like Ford, General Motors and Volkswagen to curtail their plans for making EVs. But the industry is not quite in crisis: For three straight months, carmakers have sold record numbers of EVs, including a 32% year-on-year jump in November, industry research firm Rho Motion said Friday. For the first 11 months of the year, EV sales were up 25% globally.

But China has accounted for 64% of global EV sales this year, according to Rho Motion, and there Tesla faces stiff competition. Byd, China’s largest EV maker, is on track to sell well over 4 million cars globally this year.

A big difference between the rival automakers: Byd makes dozens of models, almost all of them in both fully electric and plug-in hybrid versions. Tesla took China by storm when it began selling cars in the country in 2019, but it hasn’t released a new EV there since the Model Y in 2021. (Tesla does not sell the Cybertruck in China.)

But the company’s more fundamental problem stems from Musk’s decision to put carmaking on the back burner: The world may in fact go to Robotaxis and humanoid robots, but Tesla isn’t producing those products yet.

The Original Sin: Killing the Model 2

Two decades ago, Musk said he wanted to replace the global fleet of combustion cars with EVs as part of a necessary move away from a “mine and burn hydrocarbon economy.” To get there, he said Tesla would systematically make cheaper and cheaper EV cars until almost anyone could afford one.

In 2020, after producing a succession of relatively expensive EVs, Musk unveiled a plan to produce Tesla’s mainstream vehicle. Tesla would charge $25,000 for it, and it would be for sale in roughly three years. He didn’t give it a name, but a lot of people called it the Model 2.

However, Musk killed the car in April, though, as we reported, it was virtually ready—a small version of the Model Y SUV that was on track for delivery to customers in June 2025. Instead, he said Tesla would make a Robotaxi, a driverless car with no steering wheel or pedals. If buyers loaned them out for fares, they would be cheaper than the $25,000 car he talked about in 2020, and could end up essentially free.

Meanwhile, Musk went into what looked like crisis mode: He fired more than 10% of Tesla’s staff, including the team in charge of its vaunted Supercharger network, as we reported. When some Tesla investors and company observers criticized the move, Musk said he would make everything clear in a public event.

In October, he held that event—a glittering nighttime affair where Musk unveiled a fleet of Robotaxis for an invitation-only audience. Attendees could order drinks from a staff of humanoid robots called Optimus. It seemed impressive—until later, when it turned out that the robots were remotely operated, with hidden Tesla staff puppeteering their movements and speech. As for the two-seat Robotaxis, they were OK as fleet vehicles. Taxi companies might order them for cities; Tesla itself could operate a slew of them. But would they sell in the hundreds of thousands or more to consumers and make Tesla a $7 trillion company, as Musk claimed?

There is a big difference between the Tesla of the 2010s and the company now: When Tesla released the Models S, 3 and Y, it had little competition in EVs. It made terrifically designed vehicles that people loved. The Model 2 had a waiting, salivating market.

The future Robotaxi and the Optimus robots have glitz—but also much competition.

Last week, GM shut down Cruise, its autonomous car division, taking itself out of the competition to field driverless taxis on U.S. roads after pumping some $10 billion into the effort. But Alphabet continues to fund Waymo, which takes paid rides in Austin, Texas, Los Angeles, Phoenix and San Francisco. Zoox, owned by Amazon, operates in six cities, including Las Vegas, Miami and Seattle.

In a Dec. 6 note to clients, Deutsche Bank analyst Edison Yu said Travis Axelrod, Tesla’s head of investor relations, had told Yu’s clients that Tesla would start producing the Robotaxi—which some call the Cybercab—in 2026. At high-volume production, he said, it would cost less than $30,000 to manufacture.

According to Yu, Axelrod said the company would start autonomous taxi services next year in California and Texas. The Robotaxi wouldn’t be ready, so Tesla planned to launch the services in existing company-owned Models 3 and Y employing an unsupervised version of the self-driving software, Axelrod said; that assumed regulators would allow the fleets to operate.

Axelrod claimed Tesla could make self-driving cars more cheaply than Waymo because Waymo partnered with automakers such as Jaguar and used expensive lidar sensors, while Tesla would use its own vehicles and rely only on cameras, Yu said. Axelrod also said Tesla’s large fleet of existing cars gave it access to far more data than Waymo collected. Tesla’s Full Self-Driving software was training on data processed by 90,000 graphics processing units that Tesla had bought from Nvidia. (Waymo and most other automakers argue that safe driverless cars will require redundant sensors, including radar, cameras and lidar.)

Even if Tesla conquers Waymo, it will face much tougher competition in China, where virtually every EV is equipped with some version of self-driving software. Tesla’s FSD is not yet legal in the country, putting the company behind rivals. In terms of robotaxis, there is competition from Pony.ai and WeRide, among others.

To get FSD approved by Chinese regulators, Musk appears to be betting on his relationship with President-elect Donald Trump. That may be a smart gamble; as we have reported, Chinese President Xi Jinping may consider it in his interest to do a grand bargain with Trump that includes some sweeteners for Tesla.

Even if FSD is approved, Musk will have to contend with his formidable Chinese competition—most notably Byd.

Optimus

In his appearances with Deutsche Bank, Tesla’s Axelrod said the company expected to start selling its Optimus robots to industrial customers in 2026. The robots would have limited capability, handling simple tasks.

Of the major points Yu cited from Axelrod’s presentation, that seemed the most realistic: As we have reported, the type of AI used in humanoid robots is not close to the capability of the large language models used in ChatGPT. No one has sufficient data to train the large behavioral models that Optimus and other humanoid robots would need to carry out the sort of daily tasks Musk promised at the October event—“be a teacher, babysit your kids, walk your dog, mow your lawn, get the groceries.”

Ken Goldberg, an expert on robots at the University of California, Berkeley, told me that no humanoid robot comes close to the dexterity of a human, nor to operating autonomously. “The data we need—it’s not available on the internet,” he said. “The only way to get it is to actually create it by having people control robots or maybe simulation, but that’s not as good.

“If you think we’re here or we’re almost here,” Goldberg continued, “we’re not.”

Perhaps even more than the Robotaxi, Optimus faces formidable competition: Figure makes the 02 humanoid robot, powered by AI from OpenAI. Physical Intelligence, founded a year ago by DeepMind veteran Karol Hausman, has produced a humanoid robot called 𝝅0 (pi-zero). Naturally there is competition in China as well: EngineAI’s SE01 arguably beats all its American competition in terms of delivering a natural human gait.

Musk thinks Tesla’s work on AI gives it a leg up on the competition. Axelrod said Tesla aims to reduce the cost of materials for Optimus to $30,000, which presumably means the company will sell it for less than $40,000. That may sound cheap to a corporate executive considering the cost of a human salary and benefits, plus breaks and vacation. But it turns out robots can suffer the same physical toll as humans in tough jobs. Just watch this video.

Yu for one isn’t banking on big revenue from Robotaxis or Optimus for roughly a decade. Still, he was sufficiently impressed with what he had seen to make Optimus 31% of his valuation of the company today, according to his note.

Yu inadvertently created a stir reporting another Axelrod statement—that Tesla would release a new vehicle in the first half of next year costing under $30,000 after government subsidies. Yu dubbed the car the Model Q. That twist—naming the vehicle—captured the attention of Chinese websites covering the EV industry. Several picked up Yu’s note, spooning in their own spin (for instance that the car was actually called the Redwood).

Excitement broke out in the U.S. as well. Redwood was an alternate internal name Tesla had used for the Model 2. If Musk was thinking of reversing himself again and releasing the Model 2 next year, that would be a big deal.

Alas, Yu told me in a phone call, Axelrod had announced nothing of the sort: He had simply reiterated what Musk had previously said—that Tesla would release an affordable new model in the first half of 2025—and had given no details about the new car. “I think there’s been some misinformation or some misconstruing of information,” Yu said.

The misguided social media excitement over Yu’s note was telling: Tesla would arguably swamp everyone if it released the Model 2. As for the Robotaxi and Optimus—who knows?

---

Noteworthy

After months of small responses to U.S. restrictions on China’s EVs, battery components and chips, Beijing reacted forcefully: China launched an antitrust investigation of Nvidia and blocked exports of batteries and electronic components to more than a dozen U.S. drone makers. Unbowed, President Joe Biden unveiled new tariffs on Chinese imports of materials for the U.S. solar industry.

Lucid Motors is looking for production and technology partnerships with traditional automakers. CEO Peter Rawlinson said he is in talks about Lucid sharing its next-gen EV technology in exchange for help producing its cars. In June, VW agreed to pay Rivian $5.8 billion for the right to use its software across VW brands.

Huawei Technologies’ EV partner, Seres Group, may launch a second share listing in Hong Kong with a goal of raising some $1 billion. Huawei and Seres, which is already listed in Shanghai, make EVs under the Aito brand. Seres also makes its own EVs.

—-

Note: The Electric will go dark until Jan. 2 while I am on vacation. Have a great holiday and see you next year.

---

Recommended newsletter

AI Agenda by Stephanie Palazzolo separates hype from reality and explains how AI is transforming industries. The four times a week newsletter details the innovation and disruption happening in AI, from the AI startup funding frenzy to major technological breakthroughs that will set the agenda for decades to come. Sign up today.