ANALYSIS: Musk Launches the Tesla Robotaxi — What If This Is the Most Dangerous Gamble in Automotive History?

Marco, 34, has been driving for Uber for four years in Austin, Texas

Marco Reyes, 34, has been driving for Uber since 2022. He works about nine hours a day, six days a week, in a 2019 Toyota Camry he bought on credit. He earns about $2,400 a month after taxes—enough to rent a two-bedroom apartment in Round Rock, a northern suburb of Austin, with his wife and two sons, ages 6 and 9. He saw Musk’s post on his phone between rides, while parked outside a Whataburger on Route 183. He put his phone down. He looked out the windshield. He said nothing.

Marco isn’t a metaphor. He is one of 1.5 million U.S. ride-hailing drivers whose primary platform is Uber or Lyft. According to a study by the Economic Policy Institute published in March 2025, 73% of these drivers have no other primary source of income. The majority are men between the ages of 25 and 50, from African American, Hispanic, or recent immigrant communities. The Cybercab, at 20 cents per kilometer, isn’t going to phase them out gradually. It’s going to shut them down just as you’d cancel a subscription.

The argument I’m already hearing: “Technology has always created more jobs than it has destroyed.” That’s true over the course of a century. It’s not true over eighteen months. And Marco can’t wait a hundred years.

1.5 million drivers—the number that needs a face

The U.S. Bureau of Labor Statistics lists 1,507,400 taxi, ride-hailing, and shuttle drivers employed in the United States in 2024. This number does not include home delivery drivers, private school bus drivers, or long-distance ride-sharing drivers. If we include these related categories directly threatened by full automation, the Institute for Automation Research at the University of Massachusetts estimates that up to 3.2 million professional driving jobs could disappear by 2035 in the United States alone. Not transform. Disappear.

And yet, in his April 24 announcement, Musk spoke of job creation. Engineers. Maintenance technicians. Fleet managers. These are real jobs. They require at least two years of post-secondary education. Marco Reyes doesn’t have two years of post-secondary education. He has a driver’s license and a car loan to pay off until November 2027.

13 versions, 14 years, and still “supervised”

Tesla began selling Full Self-Driving as an option in 2016 for $3,000. It is now called FSD Supervised, version 13. The word “Supervised” was added following accidents, after NHTSA investigations, and after drivers were filmed falling asleep on the highway while letting the system drive. The NHTSA opened nine formal investigations into FSD between 2021 and 2025, two of which resulted in recalls affecting more than two million vehicles.

The Cybercab physically removes the possibility of human supervision. There is no steering wheel to grip. No brake pedal to press. If the system malfunctions, there is no human failsafe on board. Tesla claims that FSD version 13 is mature enough to operate without supervision. That’s what the company says. Not the NHTSA. Not an independent certification body. Not a congressional committee. Tesla.

There is a fundamental principle in the certification of medical devices, airplanes, and nuclear power plants: the burden of proof for safety rests with the manufacturer, and it must be validated by an independent third party before any public deployment. For the Cybercab, this principle seems to have been suspended. I don’t know why. But the regulatory silence surrounding this launch strikes me as the smell of something burning in a room where everyone pretends not to notice.

Documented Accidents Leading Up to the Launch

In October 2023, a Cruise robotaxi—an autonomous competitor to Tesla operated by General Motors—dragged a pedestrian 6 meters in San Francisco after a driver-operated vehicle had initially struck her. The Cruise system decided to “park safely” and dragged the victim under its wheels. The pedestrian survived but sustained serious injuries. Cruise was forced to suspend its operations in California. General Motors finally shut down the program in December 2024 after accumulating losses of $10 billion.

Waymo, a subsidiary of Alphabet, operates robotaxis in Phoenix and San Francisco with a safety record superior to that of human drivers—but within defined geographic areas, with extremely precise mapping and predictable weather conditions. Austin is a city of 978,000 residents with complex intersections, recently redesigned streets that haven’t been mapped, and pedestrians. Extrapolating Waymo’s safety record to the Austin context using a vehicle with different technology isn’t science. It’s optimism.

Tesla Stock: Up 12% in 48 Hours

In the 48 hours following Musk’s announcement, Tesla stock rose 12.3% on the NASDAQ, climbing from $241 to $270. The company’s market capitalization increased by nearly $80 billion in two days. Analysts at Wedbush Securities, led by Dan Ives, maintained their price target of $500, stating that the Cybercab represents “a turning point in Tesla’s long-term valuation.”

That $80 billion does not represent the value of a certified, insured, deployed, and proven product. It represents the value of an announcement. Of a 47-second video. Of a post on X. The financial markets have priced in the success of the Cybercab as a certainty before a single passenger has been transported without incident. This is exactly the mechanism that inflated Theranos’s valuation to $9 billion before its collapse. I’m not saying Tesla is Theranos. I’m saying the mechanism is identical, and that no one in the trading rooms seems to care.

Eighty billion dollars in two days. And Marco Reyes, between rides, sitting in front of a Whataburger, putting down his phone and looking out the windshield. These two facts coexist in the same reality. I find that hard to believe.

Tesla’s Debt and Timeline Pressure

Tesla ended the first quarter of 2026 with deliveries of 336,681 vehicles, down 13% from the same quarter in 2025. This marks the third consecutive quarterly decline. Sales of the Model Y have plummeted in Europe and China, a decline attributed in part—according to analyses by Bloomberg Intelligence and Bernstein Research—to the polarization of Elon Musk’s image since he joined the U.S. government in January 2025 as an advisor to the DOGE (Department of Government Efficiency).

The Cybercab is therefore not arriving in a context of quiet confidence. It is arriving at a time when Tesla needs a new narrative, a new wave of enthusiasm, and a new reason for shareholders not to sell. Tesla’s commercial urgency and the Cybercab’s public safety imperative are pulling in opposite directions. One wants to move fast. The other demands that we slow down.

Baidu Apollo, Huawei, and the Silent Race

While Musk announces the Cybercab, Baidu Apollo is already operating 1,000 robotaxis in Wuhan, Chongqing, and Shenzhen, China. The Apollo Go service completed 7 million rides in 2025, according to figures released by Baidu in January 2026. Huawei is developing its own autonomous driving system, integrated into vehicles from partner brands SAIC and Chery. China isn’t waiting for U.S. approval. It’s rolling out the technology.

This reality puts the United States in an uncomfortable position: holding Tesla back in the name of public safety amounts, in Musk’s rhetoric, to “letting China win.” The argument is powerful. It is also intellectually dishonest. China is deploying robotaxis in controlled areas, with precise mapping and under state supervision. This is not comparable to an open commercial launch in an American city without a federal regulatory framework. But the argument will hold sway. It already is.

Technological acceleration and the geopolitical race are becoming the two arguments that shut down any discussion of security. “We can’t slow down—China is moving forward.” It’s the same line we hear about military AI, lethal autonomous drones, and mass surveillance. Each time, it’s true. And each time, it’s an insufficient answer to the question: “What if someone dies?”

Europe is watching—and beginning to ban

In March 2026, the European Union published a draft regulation on highly automated vehicles (Levels 4 and 5 according to the SAE classification) that requires certification by an independent notified body, mandatory insurance coverage by the manufacturer, and a public incident registry. If adopted—the vote is scheduled for September 2026—this framework would make the Cybercab, as launched in the United States, illegal on European roads. The European Commission explicitly cited the Cruise precedent in its explanatory memorandum.

This is not protectionism in disguise. It is the difference between regulations designed to protect citizens and regulations designed to protect innovation. Both are legitimate. They are not equivalent when a child is crossing the street at a red light.

DOGE, Deregulation, and Tesla

Since January 2025, Elon Musk has headed the Department of Government Efficiency (DOGE), the advisory body created by the Trump administration to reduce federal spending and regulations. Among the agencies that DOGE has identified for “regulatory cuts” is the NHTSA—the very agency responsible for certifying autonomous vehicles such as the Cybercab.

This is no coincidence. It is a documented, identifiable, and datable conflict of interest. Elon Musk, CEO of Tesla, has a direct financial interest in weakening the NHTSA. Elon Musk, an advisor to DOGE, has the power to weaken the NHTSA. These two facts coexist within the same biography. No Republican leader in Congress has called for his recusal. No investigative committee has been launched. The institutional silence surrounding this situation is, in itself, newsworthy.

I realize that I have just written a sentence that no one in the mainstream American media has phrased so directly. Not because it is false—it is documented point by point. But because there is a pervasive pressure, in the current media environment, not to name the conflicts of interest surrounding Musk too clearly. I am naming them. That is my role.

The NHTSA Is Being Weakened at the Very Moment It Is Most Needed

In March 2026, the DOGE recommended a 23% cut to the NHTSA’s budget and the elimination of 340 out of 817 positions. These cuts would primarily affect the teams investigating automated driving incidents. In response, the NHTSA issued an internal memo obtained by Reuters warning that these cuts “would severely compromise the agency’s ability to oversee the deployment of autonomous technologies on a safe timeline.”

To be precise: at the very moment Tesla is launching production of the world’s first driverless vehicle intended for commercial use by the general public, the agency responsible for certifying this vehicle is being stripped of its investigative teams by a government body led by Tesla’s CEO. This paragraph needs no commentary. It needs to be read twice.

Bryant Walker Smith, professor of law at the University of South Carolina

Bryant Walker Smith, a law professor at the University of South Carolina and one of the world’s leading experts on autonomous vehicle law, said in an interview with The Verge on April 25, 2026: “Manufacturing a vehicle is not the same as certifying it. Tesla can build as many Cybercabs as it wants. The question is under what authority it puts them on public roads without a driver.” Smith added that the lack of a clear federal framework creates a legal vacuum that states are filling in disparate ways, opening the door to “non-consensual experimentation on civilian populations.”

Missy Cummings, former director of the Office of Automated Vehicle Safety Research at the NHTSA and now a professor at George Mason University, was even more direct in a LinkedIn post dated April 24, 2026: “Tesla’s FSD has not passed a single independent certification test. Not one. Launching a commercial driverless service on this basis isn’t bold. It’s reckless.”

These two voices exist. They are cited in specialized publications. They were not included in the twenty minutes of enthusiastic coverage by U.S. financial news channels on April 24. The editorial choice of whom to quote is never neutral.

The Actual Safety Record of FSD Version 13

Tesla publishes its own safety data on FSD quarterly. In the fourth quarter of 2025, supervised FSD recorded one incident requiring human intervention every 14.7 km in dense urban conditions, according to Tesla’s internal data. This figure has been challenged by independent researchers at the University of Berkeley, who—based on mandatory incident reports submitted to the NHTSA—calculated an intervention rate six times higher under suboptimal conditions (rain, nighttime, unmapped intersections).

Austin averages 81 days of rain per year. Austin has unmapped intersections because the city is constantly expanding, and new streets are appearing faster than mapping systems can keep up with. Neither of these facts was mentioned in any of Tesla’s press releases.

38,000 deaths per year on U.S. roads—and 94% caused by human error

In 2024, 38,400 people died on U.S. roads. According to the NHTSA itself, 94% of these deaths involved human error as the primary factor—distraction, alcohol, fatigue, or excessive speed. An autonomous driving system that eliminates human error won’t save abstract lives. It will save names. It will save Derek, 22, who died on Interstate 35 near Austin in August 2024 because a driver distracted by his phone crossed the center line at 120 km/h.

And yet, the argument only holds water if the autonomous system is statistically safer than the average human driver under all conditions—not just under optimal conditions. Waymo has demonstrated this superiority in defined areas. Tesla has not yet demonstrated this superiority in dense and varied urban conditions—at night, in the rain, with unpredictable pedestrians. The promise of saving lives does not absolve the obligation to prove that we won’t lose any in the first place.

I want to be honest here. If the Cybercab lives up to its promises—if in ten years it drives better than the average human in 99.9% of situations—then we will have been debating while people were dying on roads that the technology could have made safe. That burden exists as well. I’m not dismissing it. I’m simply saying that it doesn’t justify deploying uncertified technology on unwilling citizens. We can move quickly and still do things right. The two aren’t mutually exclusive.

Cybercab and Accessibility: What Musk Is Right to Advocate For

Older adults who can’t drive, people with mobility impairments, residents of rural areas without public transportation—for these populations, a robotaxi costing 20 cents per kilometer represents an unprecedented revolution in accessibility. The Brookings Institution estimates that 25 million Americans lack access to adequate transportation to hold down a job or access medical care. A reliable, affordable, and 24/7 Cybercab network could structurally transform the lives of these 25 million people.

This argument is valid. It deserves to be heard. It does not address the issues of certification, legal liability, and the fate of professional drivers. But it does mean that the answer to the question “Should Cybercabs be deployed?” is not “no.” It is “not like this, not at this pace, not without this regulatory framework.”

Facebook, Uber, drones—the same pattern for twenty years

Facebook launched its algorithm-driven News Feed in 2006 without conducting an impact assessment on teenagers’ mental health. The first studies documenting negative effects were published in 2017. As of 2026, U.S. regulations have still not been adopted. Uber launched its operations in San Francisco in 2010, deliberately ignoring taxi regulations. It took four years and massive lobbying for California to create a legal framework. Civilian drones were deployed commercially on a large scale before the FAA adopted its first rules governing the use of airspace.

The sequence is always the same: deployment → incident → scandal → poorly calibrated emergency regulation. The Cybercab fits into this sequence with a precision that should be alarming. The difference from Facebook or drones is that a Cybercab weighs 1,400 kilograms and travels at 80 km/h on streets where children walk.

Every time I write about a technological innovation, I see this same pattern. The enthusiastic announcement. The rushed rollout. The downplayed inaugural incident. The regulations that come afterward. And between the announcement and the regulations, real people pay the price for the experiment. I’m not anti-technology. I’m anti-impunity.

The Boeing Precedent—When Commercial Urgency Kills

In October 2018 and March 2019, two Boeing 737 MAX aircraft crashed, killing 346 people. Investigations revealed that Boeing had faced internal pressure to expedite the MAX’s certification in order to compete with Airbus’s A320neo. Internal warnings about the MCAS system had been ignored. The FAA, subject to budget cuts and a culture of deference toward the industry, had delegated part of the certification process to Boeing itself. The parallel with the NHTSA-Tesla-DOGE situation in 2026 can be documented line by line. I am not claiming that the Cybercab will kill 346 people. I am claiming that the institutional conditions that enabled the 737 MAX are being deliberately recreated.

Boeing paid $2.5 billion in fines and lost $20 billion in contracts. The 346 people who died in the crashes cannot be compensated. If Tesla deploys the Cybercab before rigorous independent certification and a fatal incident occurs, the shareholders will survive the fine. The victims will not survive the accident.

Austin, Texas: A Real-World Testing Ground

In six weeks, if Musk sticks to his schedule, Cybercabs will be cruising the streets of Austin. On Congress Avenue, in front of City Hall. On South Lamar Boulevard, where families are leaving restaurants on Friday nights. On the roads around Barton Springs, where joggers run at dawn with earbuds in. These people haven’t signed a consent form. They won’t be told that the vehicle passing them at 60 km/h is driven by an algorithm whose independent certification hasn’t yet been published.

And yet—and this is where I have to be honest with myself—the first month will likely pass without any major incidents. Conditions will be optimal. The routes will be carefully selected. The fleet will be limited. The sensors will have been recently calibrated. The real test will come later, when the fleet expands, when conditions deteriorate, when the system encounters the thousands of unforeseen situations that no training can anticipate.

What I fear isn’t the June 2026 launch. What I fear is the normalization that will follow the first incident-free month. The moment when everyone will say, “See, it was just paranoia.” And when issues of certification, liability, and worker protection will be buried under the weight of a deployment that’s already been completed. The window to have these conversations seriously is now. Before. Not after.

The decisions to be made in the next six weeks

The city of Austin must decide by June 2026 whether to allow driverless vehicles on its public roads without a formal service agreement. The U.S. Congress could—could—fast-track the passage of the AV START Act, a law on autonomous vehicles that has been stalled since 2017 due to a lack of consensus. The NHTSA, despite its budget cuts, can require a supervised public testing period before any commercial deployment. These three decisions are made by people with names, titles, email addresses, and constituencies. They are not inevitable. They are choices.

As of April 25, 2026, Austin Mayor Kirk Watson had not made a public statement regarding the Cybercab. His office did not respond to requests for comment from The Austin American-Statesman. This silence is a decision. Silence in the face of a public safety issue is always a decision.

What We Know for Sure on April 25, 2026

Here are the facts. Tesla has begun production of the Cybercab at Giga Texas. The vehicle is designed to operate without a driver. Musk announced a commercial service in Austin for June 2026. Tesla’s stock surged 12% in 48 hours. The NHTSA is undergoing budget cuts under the direction of the DOGE, led by Musk. No independent certification of FSD version 13 under full-autonomous conditions has been published. 1.5 million U.S. professional drivers have received no communication regarding their retraining.

These facts coexist. They do not cancel each other out. Technological enthusiasm is legitimate. Regulatory caution is legal and mandatory. Tesla’s economic urgency is understandable. It does not constitute authorization for deployment. And Marco Reyes, 34, still drives between Round Rock and Austin six days a week in a Toyota Camry he bought on credit. He has until November 2027 to pay it off. The Cybercab, meanwhile, is set to arrive in June 2026.

I don’t know if the Cybercab will revolutionize transportation or if it will kill someone before it’s properly certified. No one knows. What I do know is that this uncertainty justifies exactly the kind of independent, rigorous, and public oversight that the DOGE is in the process of dismantling. What I do know is that the window to have this conversation seriously is now—before the first passenger, before the first incident, before the answer to the question “Who is responsible?” becomes a decade-long legal battle from which only lawyers will emerge unscathed.

The final image: a road in Austin, one June evening

Congress Avenue, Austin. A June evening in 2026. A silver Cybercab turns right on a green light. Inside, no one. Behind it, a 2019 Toyota Camry waits. At the wheel is Marco Reyes, 34. He watches the empty vehicle disappear into the balmy Texas night. He still has three hours left on his shift before heading home to Round Rock. He has two sons who are already asleep. He has a car loan to pay off until November 2027.

He doesn’t yet know if, in eighteen months, there will still be an app sending him rides. He doesn’t yet know if the city of Austin will have adopted a protective framework for drivers displaced by automation. He doesn’t know if anyone, somewhere in a U.S. congressional chamber, was thinking of him that day. He sets his phone down on the dashboard. The light turns green. He drives on.

By Maxime Marquette, columnist