America’s top auto safety regulator is investigating a fatal crash and multiple injuries potentially linked to Tesla’s FSD system.
The auto industry is facing major roadblocks as it shifts from a century of gas-powered cars to electric vehicles. But the usual EV headaches—range anxiety, high costs, and limited charging infrastructure—are starting to feel small compared to the chaos in the autonomous vehicles space.
Every major automaker pushing for higher levels of autonomy is now under federal scrutiny, thanks to crashes, injuries and deaths linked to their systems. Tesla, in particular, can’t seem to shake the heat. A week after showcasing its self-driving Robotaxi, Tesla is back under the federal microscope.
Welcome back to Critical Materials, your daily round-up of news and events shaping up the world of electric cars, software-defined vehicles and autonomous tech. Also on today’s agenda: how General Motors is using artificial intelligence to eliminate its software problems and what Lucid’s massive stock sale means for its business and survival.
30%: Tesla Is Back Under NHTSA Probe
The vibe at Tesla’s Robotaxi reveal last week was a mix of spectacle, skepticism and some cringe fanaticism.
The gold Cybercabs with butterfly doors and no steering wheels and pedals cruised around Warner Bros. Studios, but it all felt choreographed. They didn’t face bad weather, fog, or any real-world challenges—the exact situations where Tesla’s Full-Self Driving (FSD) system struggles.
That’s why FSD, which is not a very accurate name for its capabilities since it’s only a supervised Level 2 driver assistance system, has drawn fresh scrutiny from the National Highway Traffic Safety Administration (NHTSA).
The agency said on Friday that it was opening an investigation into 2.4 million Teslas equipped with FSD after four reported collisions, including a fatal crash. Those include all Teslas model year 2016 onwards equipped with FSD, including the Cybertruck.
Here’s what NHTSA had to say about the incidents:
The Office of Defects Investigation (ODI) has identified four Standing General Order (SGO) reports in which a Tesla vehicle experienced a crash after entering an area of reduced roadway visibility conditions with FSD -Beta or FSD -Supervised (collectively, FSD) engaged.
In these crashes, the reduced roadway visibility arose from conditions such as sun glare, fog, or airborne dust. In one of the crashes, the Tesla vehicle fatally struck a pedestrian. One additional crash in these conditions involved a reported injury.
Tesla’s credibility is on the line here. Sure, it mentions on its website that FSD requires driver supervision and readiness to take control. But the hype and messaging around its FSD system indicate otherwise.
“We’ll move from supervised FSD to unsupervised FSD where you could fall asleep and wake up at your destination,” Tesla CEO Elon Musk said at the We, Robot event last week. “It will save lives and prevent injuries,” Musk added.
Before any of that happens, regulators would need to see millions of miles in testing data without any interventions to analyze FSD’s reliability, according to half a dozen experts and scientists in the autonomous vehicles space that InsideEVs spoke to recently.
Experts said critical questions remain unanswered, such as how the system’s cameras handle poor visibility, what happens after a crash, or if a rider passes out. So all of this uncertainty leaves doubts about whether FSD is truly ready for widespread use, despite the company’s bold claims.
And no, the dozens of FSD beta testing videos online don’t count as proof.
“What matters in safety is not the average day. What matters is the bad day and the bad days are extremely rare,” Phil Koopman, a professor of electrical and computer engineering at Carnegie Mellon University told me last week. “Saying ‘I drove 10 miles without an intervention’ means nothing.”
So unless there’s a change of approach, it seems like solving autonomy is the start of an uphill and extremely uncertain battle for the world’s most valuable car company.
60%: General Motors Taps Into AI To Solve Software Issues
General Motors has added cutting-edge artificial intelligence systems at its software development center in Michigan to detect problems in its vehicles before they enter production.
The AI tools help speed up manufacturing by spotting glitches in a vehicle’s software, something the automaker has struggled with as it attempts to ramp up production of a dozen new electric cars.
GM has developed an automated tool that runs simulations around the clock to spot glitches before the software code is integrated into a vehicle. In addition, cameras, sensors and analytics monitor manufacturing equipment to ensure it works properly and increase plant uptime.
A proprietary tool used in body shops called Spark Eyes inspects and validates weld quality, Clausen said. A separate system of cameras and analytics collects about 165 million images per day to predict manufacturing and conveyor equipment failures before they interrupt vehicle production.
These tools run round the clock, help with testing and validation and evaluate parts for inconsistencies before they’re installed on the vehicle.
And this might actually be working. GM’s EV sales saw record growth in the third quarter thanks to the Chevy Equinox EV which is now the best-selling GM EV. The Blazer EV and the Cadillac Lyriq have also been on a hot streak lately.
90%: Lucid’s Stock Sale Could Be Critical For Its Survival
Lucid has announced a huge share sale ahead of the launch of its Gravity electric SUV. The automaker said on Thursday that it would offer up to 637 million shares which could fetch up to $1.67 billion.
That amount would be used for “general corporate purposes,” including capital expenditures and working capital. Lucid’s biggest shareholder, Ayar Third Investment Company (linked to Saudi Arabia’s Public Investment Fund), is also buying an additional 375 million shares in a private deal to keep its 59% ownership of the company.
The Lucid Air sedan has received rave reviews but it’s a loss-making product. Its lease rates have dropped to $549 per month and the automaker is selling less than 3,000 units per quarter. Now it’s hoping that the Gravity SUV, whose launch is due this fall, could help turn things around.
100%: Where Does This End For Tesla?
Some experts think Tesla is still a decade away from meaningfully deploying any self-driving cars on the road. Musk said he plans to have Robotaxis on the road before 2027, but he admitted that he’s overly optimistic with timeframes.
Given the dozens of unanswered questions about Tesla’s hardware limitations and lack of clarity about how the Robotaxi business would work, that timeline seems unrealistic. When, and how, do you think any of this can be realistically possible? Leave your thoughts in the comments.
