The Tesla bull case has been carrying a quiet contradiction for six months, and a Reuters tracking study published Tuesday put a number on it. Reuters reporters spent three weeks tracking Tesla Robotaxi wait times in Austin and Dallas. They found Austin wait times exceeded 15 minutes about half of the time, with one in four checks running longer than 25 minutes. Twenty-seven percent of the time, no car was available at all. In Dallas, one reporter spent nearly two hours waiting for a Robotaxi to complete what should have been a 20-minute, five-mile run from Southern Methodist University to Dallas City Hall. The “AI-powered ride-hailing network” narrative is currently rate-limited by something more boring than tech: safety validation Tesla is choosing to throttle itself.
What The Reuters Data Actually Says
The methodology matters because it strips out the talking points. Reuters ran wait-time checks eight times a day for three weeks in Austin, the city Tesla used for its June 2025 commercial launch and where the company has had the longest opportunity to scale supply to demand. The data set is large enough to filter out outliers and time-of-day effects. The conclusion is brutal in its consistency. More than half the requests took longer than 15 minutes. More than a quarter took longer than 25 minutes. More than a quarter returned no car at all.
Tesla’s Dallas rollout last month was framed by management and amplified by sell-side bulls as evidence the network was scaling. The Reuters reporter’s two-hour Dallas wait, on a clear weekday afternoon for a freeway run with no construction and no traffic, is the kind of data point that scales in the wrong direction. Either Dallas is supply-constrained because Tesla has not deployed enough vehicles, or it is service-area-constrained because the validated geofence is narrower than the marketing implies. Both of those answers are bad for the AI-network story.
The number that should be in every Tesla model right now is Tesla’s own self-reported crash count to NHTSA. Per the Reuters report, Tesla has reported 15 crashes in Austin to the federal regulator since August, the minimum disclosure standard for autonomous vehicle operators. None have produced citations from Austin Police, and Lieutenant William White, who oversees AV safety for the city, told Reuters the company has had no major crashes. That is the good news. It is also exactly why the company is choking the supply side.
Safety Validation Is The Binding Constraint, And Musk Said So
On the Q1 2026 earnings call, Musk acknowledged the issue in language that quietly walked back his earlier guidance of Robotaxis “in roughly a dozen states” by year-end 2026. The specific phrase was that safety validation had become “the constraint” on the pace of expansion. That admission was buried under the $25 billion AI capex announcement that dominated the Q1 conversation, which we covered in our breakdown of Tesla’s Q1 earnings and the AI spending spree that spooked Wall Street. Six weeks later, the Reuters tracking confirms what the call hinted at: the company is choosing wait times over crashes, and at the current cadence the math on the year-end city count does not work.
The mechanics of Tesla’s validation problem are different from Waymo’s, which is why a head-to-head comparison flatters Waymo. Waymo’s validation regime is HD-mapped, geofenced, and supervised by a remote operator pool that intervenes on the order of once every several hundred miles. Tesla’s validation regime is camera-only, vision-stack-trained, and supervised by safety monitors who sit in the passenger seat on the current Austin fleet. The Tesla approach is theoretically more scalable. The empirical evidence, however, is that the validation cycle on the camera-only stack is producing slower per-city ramp times than Waymo achieved in the same Sun Belt geographies two years ago.
Why The Reuters Data Is A Valuation Story
The pricing implication is where this becomes a markets story, not a tech story. Piper Sandler published a refreshed DCF on Tesla earlier this month that broke the company into 17 product lines, including vehicles, energy storage, supercharging, FSD subscriptions, insurance, and the Robotaxi business, and arrived at a 20-year sum-of-the-parts value of $400 per share. With Tesla trading near $445, the implication is that the entire $45 per-share premium is being paid for Optimus, inference-as-a-service, and any other AI optionality not captured in the conventional product lines. The Reuters wait-time data does not touch the Optimus narrative directly, but it changes the discount rate the market is willing to apply to the parts of the DCF that include Robotaxi growth.
Here is the trade. Tesla bulls have been priced for a story where the Robotaxi network is on a scaling curve similar to Uber’s 2010 to 2014 ramp, with deployments doubling or tripling each quarter and unit economics improving as scale arrives. Reuters’ data is consistent with a different curve, one closer to Cruise’s 2022 to 2023 ramp where safety bottlenecks produced linear rather than exponential expansion before the rollout pulled back entirely. The price-to-sales multiple the market is paying for Robotaxi exposure is calibrated to the first curve. If the data continues to point at the second, the gap closes through a multiple compression rather than a fundamental miss.
The Optimus side of the bull case has the same problem with extra steps. Wall Street is paying roughly $45 of Tesla’s $445 share price for a robot that has not shipped to a single paying customer, on a timeline that assumes the same validation engineering that is currently throttling Robotaxi also delivers a humanoid platform inside the same calendar year. That is a stack-of-assumptions trade, not a fundamentals trade. The Reuters data is the first hard signal that the validation stack itself is the rate limiter, which means the Optimus timeline implied in the share price is materially aggressive.
Waymo Is The Comparison That Matters, And It’s Not Close Right Now
Waymo’s commercial rollout in Phoenix, San Francisco, Los Angeles, and Austin is currently running at roughly 250,000 paid rides per week, per Alphabet’s Q1 disclosures. Tesla has not published a paid-ride number for Robotaxi. The Reuters tracking implies the Austin fleet, the most mature deployment, is producing per-vehicle-day ride counts well below the 15 to 20 daily rides per vehicle that Waymo achieves in mature service areas. The fleet-utilization gap is the headline number Tesla bulls have to explain away, and the Reuters methodology is going to make that conversation harder.
The competitive picture is not static. Waymo’s Phoenix expansion now covers 315 square miles, the largest commercial AV service area in the world. Cruise’s relaunch in Texas under the new GM autonomous strategy is targeting a 2027 commercial Houston launch. Zoox is operating in Las Vegas. Tesla’s first-mover advantage in passenger-vehicle autonomy is real on the hardware side and disappearing on the operational-validation side. The Reuters data is a snapshot of why.
What To Watch Through Q2 Earnings
Three signals matter through Tesla’s Q2 print in late July. First, whether Musk discloses a paid-ride count for Robotaxi for the first time. Round numbers, even unflattering ones, are evidence of internal confidence. Continued silence is evidence the data is bad. Second, whether the company has expanded the validated service area in Austin and Dallas, or has only added vehicles to the existing geofences. Geofence expansion is the real progress signal. Third, whether the language about “states by year-end” survives the call. Musk has rarely walked back a public number publicly, but the Reuters data leaves no room for the rhetoric without an asterisk.
The bull case for Tesla is a faith in execution. The Robotaxi rollout is the first place where execution can be measured by a third party with a stopwatch. Reuters just published the stopwatch. The trade is no longer “is Tesla the AI play of the decade?” It is “does the AI play deliver more than Waymo by the end of 2026?” The Reuters tracking says the answer is currently no, and the $45 per share investors are paying for Optimus is the price of a story that needs the timeline to hold.