Is eliminating human safety drivers the turning point that could reshape the trajectory of Tesla’s valuation? Last weekend, Tesla commenced its test operations for its robo-taxis without human supervision in Austin, Texas, making it the first time that its vehicles would be driven on public roads with no human “chaperone” on board. This announcement by Elon Musk, its CEO, and its VP for AI software, Ashok Elluswamy, pushed the company’s stock price up 3.6% on Monday to $475.31, just short of its all-time high of $479.86.
For Wall Street optimists, the announcement marks a major shift towards realizing Musk’s vision of a national network of autonomous ride-hailing services, which could potentially be worth trillions of dollars. Vijay Rakesh of Mizuho raised his price target for the firm to $530, terming the elimination of the safety directors “a key catalyst for earnings momentum.” Another analyst, Dan Ives of Wedbush, even set a price target of $2 trillion over the next nine months, driven by accelerated production of the Cybercab and deployment across the U.S. market in 2026. The Cybercab was first shown at a launch in October of last year and is a two-person car that does not require steering or pedals to operate and so is totally autonomous from the point of its conception.
Nevertheless, the way to expansion is anything but smooth. Barclays analyst Dan Levy warned: The US is a “patchwork” country, at least at the state level, which could slow the rollouts. Texas already allows Autonomous vehicles for test drives without licenses, but Senate Bill 2807 makes it necessary that the DMV approve such services after May 2026. The California DMV and Public Utilities Commission also stated that Tesla has not applied for the licenses necessary for the operation of its self-driving services.
Safety metrics are also being closely watched. Tesla’s Austin, TX, fleet averaged one crash per 40,000 miles in controlled closed courses. But data submitted by the company to the National Highway Traffic Safety Administration showed seven collisions from June through the middle of October within a pool of cars numbering less than 30, all of which were operating Tesla’s most advanced autonomous systems. Though these were not serious, Carnegie Mellon’s Philip Koopman, an expert in robotics, called attention to the high frequency within this human-supervised group. Tesla has not made available the text descriptions of these crashes.
A comparison between Tesla and market leader Waymo reveals how much work Tesla has cut out for it. Waymo is working in five major U.S. cities and providing more than 450,000 paid rides per week, and their safety statistics reveal accidents a tenth of those of conventional vehicles. Waymo’s plan for 2026 includes entering 20 new cities, more extreme-weather cities such as Denver and Detroit, and entering international markets in London and Tokyo. Waymo’s capacity to deal with complex scenarios of both urban and freeway routes, along with approvals in numerous U.S. states, reflects a very high standard of performance.
From an innovation perspective, Tesla’s autonomy drive is driven by its Full Self-Driving (FSD) AI pipeline, with its AI networking and machine learning processes trained and honed from data generated by actual customer car use, amounting to over billions of miles of data collected from actual customer use cases. The latest version of the FSD (Supervised) system has been noted to have improved handling and lower engagement levels, prompting automotive reviewer Jason Cammisa to call it “door-to-door amazing,” even as Barron’s columnist Al Root admits to preferring the experience of driving an FSD-equipped car “most of the time” when operated under supervision by human beings. Nonetheless, according to data released by Tesla, even when combined with human supervision, the improvement from purely human-driven cars is not significant coming short of the 2 to 3 times performance improvements demanded by Elon Musk for autonomous systems deployment.
The computational needs for the scaled FSD are quite challenging. This is due to the nature of Tesla’s machine learning tasks involving the high-throughput data import from the company’s global fleet, simulation tasks, and hardware-specific inference optimizations for functionality within the vehicles. To attain parity with or better the disengagement rates offered by the Waymo system with the ability to run for tens of thousands of miles without human intervention, it will take both algorithm improvements and hardware solutions.
Although American EV sales are down 41% in November due to the loss of the $7,500 federal tax credit, investors are now focusing from the “fragile” retail story to autonomy as the next growth catalyst. The Tesla Robotaxi app, which was released in September, already has 529,000 downloads so far, with an average of 2,790 downloads per day in the last month. Although this pales in comparison to Waymo’s downloads, the Austin completely untethered tests have recently given Tesla’s autonomy story a push, with investors now monitoring filings, fleet growth rates, and safety data releases as key indicators of the next wave of the stock market rally.
