The Information : How Musk Could Tie Together Tesla and xAI

How Musk Could Tie Together Tesla and xAI

The Takeaway
  • xAI could buy chips from Tesla fab Musk teased at annual meeting
  • Adding steering wheel to Cybercab would help with regulators
  • Musk wants to use Optimus robot for package deliveries

At Tesla’s annual meeting earlier this month, shareholders showed little enthusiasm for a proposal that the electric vehicle maker invest in one of CEO Elon Musk’s other companies, xAI. But Musk’s revelation at the meeting that Tesla is considering building its own chip fabrication plant signaled another way his two companies could work together.

Musk had already discussed plans for such a chip fab internally at Tesla months before the shareholder meeting, describing it as a make-or-break project for the company, according to a person with direct knowledge of the comments. If Tesla goes forward with a chip fab, xAI would likely buy chips from it, according to another person familiar with the matter. Musk also hinted earlier this month that SpaceX may one day use Tesla-designed chips to power data centers in space. The possibility of Tesla building a chip fab to supply several of Musk’s companies would make the automaker even more central to his empire.

Musk is also banking on far out ways to make his two most ambitious bets within Tesla—its Cybercab autonomous vehicles and Optimus humanoid robot—work together. He has told teams working on Optimus that he wants the robot to be able to get into and out of the Cybercab to make package deliveries, according to a person with direct knowledge of the Optimus program, similar to Amazon’s goals of one day using humanoid robots for deliveries. That means Tesla would have to update the design of Optimus so it can easily enter and exit the car, the person said.

Tesla shareholders’ overwhelming approval at the meeting of Musk’s $1 trillion compensation package gives Musk a big incentive to pursue his ambitious strategy for the electric vehicle maker. To collect the full $1 trillion, Tesla has to meet a series of ambitious goals including getting one million robotaxis on the road and selling a million Optimus robots within a decade

As he does so, Musk has to carefully balance Tesla’s needs with those of his other companies. XAI, for instance, needs vast amounts of cash and chips for its data centers to compete with the likes of OpenAI and Anthropic. It has raised money from SpaceX already and it could tap Tesla, which had $41.6 billion in cash at the end of September.

But Tesla also has its own cash needs, which will likely grow in the coming years as its robotaxi and Optimus robot development scales up. Musk said at the meeting that Tesla will need to spend “tens of billions” of dollars on training AI for its Optimus humanoid robot alone.

Tesla also needs chips. Musk has said that Tesla’s goal is to create a chip that works well inside consumer products like cars and the Optimus robot, as well as in data centers. That would be a big shift from how Tesla now operates. It now uses chips it designs in its cars, manufactured by TSMC and Samsung, and relies on Nvidia and others for data center chips. Musk said at the meeting that Tesla might explore a deal with Intel in addition to its own fab because he expects its need for chips to skyrocket in the coming years.

Yet Musk’s projections for Tesla’s chip needs are based on the premise that the company will grow at a breakneck pace in the coming years, which is far from guaranteed and would require Optimus and Cybercab, an upcoming two-seater vehicle built for autonomous ride-hailing, to be major hits. That would be a significant reversal from Tesla’s current state—this year, the company is on track to sell fewer vehicles than it did in 2024, which itself was a decline from 2023. And Tesla faces significant technical and regulatory challenges as it tries to get both Optimus and Cybercab off the ground.

Analysts expect Tesla’s free cash flow to decline slightly in 2026 before accelerating quickly in the following years, according to data from S&P Capital IQ, but that’s dependent in part on its self-driving software helping to make up for sagging electric vehicle sales and increasing competition.

Tesla and xAI did not respond to requests for comment.

Delivery Robots

Optimus is Tesla’s biggest long-term bet. Musk has said there will eventually be more humanoid robots than cars in the world, and that Optimus will one day be responsible for about 80% of Tesla’s market capitalization. Inside Tesla, he’s pushed the Optimus team to find ways to use the robot in tandem with another big, nearer-term bet: the Cybercab, according to a person with direct knowledge.

That includes Musk’s desire to have the Optimus robot sit in the Cybercab so it can deliver packages. That should be possible: newer versions of the Optimus robot are capable of consistently lifting and moving around with roughly 25-pound objects for three to four hours on a 30 minute charge, another person with direct knowledge said.

But the connection between the robot’s torso and legs isn’t flexible enough to allow it to seamlessly get in and out of a Cybercab, according to the first person. Tesla would need to redesign the robot to change that or use a different vehicle for deliveries more tailored for Optimus’ shape, that person said.

For now, Tesla is focused on figuring out a design for Optimus that Musk is comfortable scaling up. The company had originally planned to produce 5,000 Optimus bots this year, but abandoned that target and delayed scaled production to 2026 due in part to technical issues with its hands, The Information previously reported. Tesla is now focused on refining a version of the robot Musk has dubbed V3.

At the recent shareholder meeting, Musk said is going to stand up a “million unit production line” for Optimus in Fremont, Calif., followed by a ten million unit production line in Austin, though he did not share a timeline for doing so. One of the milestones in Musk’s pay package is selling a million robots.

In the near term, Tesla expects to deploy at least the first several thousand Optimus bots it produces inside its own facilities like factories and warehouses, one of the people with direct knowledge of the program said. That’s due in part to avoid technology leaks to rivals that could come from placing its robots in other companies’ facilities, the person said. It’s an approach allowed by Tesla’s scale and is a contrast with humanoid robotics rivals like Figure AI, which has deployed its robots in a BMW factory, and 1x, which is putting robots in customers’ homes.

In the longer term, though, Musk envisions Optimus working in people’s homes and offices. That means the robot needs to be able to reliably communicate with humans. To do that, Tesla has been using xAI’s Grok chatbot.

But the companies still have kinks to work out. In September, Salesforce CEO Marc Benioff posted a video from the Tesla office in Palo Alto. While Benioff captioned the video with effusive praise for Musk and Tesla, the video showed the Optimus taking several seconds to respond to Benioff’s voice and required him to repeat himself before it would take an action.

The awkwardness in the video was a result of xAI taking too long to process the audio from the Optimus bot before telling the robot what to do, one of the people with direct knowledge of the program said. After Benioff posted the video, the person said, the Optimus team watched it repeatedly and has been trying to figure out how to make the robot respond to speech faster.


Cybercab Questions

Another big reveal at Tesla’s shareholder meeting was Musk saying that Tesla would start producing the Cybercab at its Austin factory in April.

The Cybercab design shown publicly by the company features no steering wheel or mirrors because Musk says that the self-driving software powering the vehicle will be so capable that a human will never need to take over. But despite the plan to start production soon, Tesla has many hurdles to clear before it can put such a car on the road.

For one, deploying a car without a steering wheel, mirrors or pedals will require Tesla to apply for an exemption from federal safety rules with the Department of Transportation. That process can take many months, and success is far from guaranteed—before General Motors shut down Cruise in 2023, the self-driving car firm had been stuck in limbo for 18 months waiting for the DOT to approve an application for a car without a steering wheel. However the DOT, which is led by Musk nemesis Sean Duffy, said earlier this year that the department would change its rules to streamline the approval process.

Moreover, even if the DOT grants an exemption for the Cybercab, current federal rules allow companies to manufacture and deploy a maximum of just 2,500 vehicles covered by an exemption each year. Unless the DOT were to change its rules, Tesla would be barred from manufacturing Cybercab at the scale Musk wants. One of the targets outlined in Musk’s pay package is to put one million robotaxis on the road by 2035.

One way around this challenge is to manufacture the first versions of the Cybercab with a steering wheel and mirrors. Tesla has already produced versions with those features that have been spotted being driven by humans at the Austin factory and on public roads in Texas and California in recent months.

While Musk has repeatedly pushed back on the idea of putting a steering wheel in Cybercab—and killed plans last year for a $25,000 driveable consumer vehicle built on the same platform as the Cybercab, The Information has reported—Tesla board chair Robyn Denholm dangled the possibility during a press tour ahead of the vote on Musk’s compensation. “If we have to have a steering wheel, it can have a steering wheel and pedals,” Denholm told Bloomberg in October.

Ahead of the Cybercab production, Tesla has launched a limited ride-hailing service, called Robotaxi, in Austin and San Francisco. So far, Robotaxi has been using Model Y vehicles with human operators in the cars who can take over from Tesla’s self-driving software. Musk said in October that Tesla wants to launch the service in 8 to 10 U.S. metro areas by the end of 2025—down from a previous goal of reaching half the U.S. population this year—but Tesla still has to complete key steps with state regulators.