NHTSA's safety investigation into Tesla's camera only FSD architecture is why the 14 to 1 Texas fleet gap now reads as a federal question, not a deployment race.
Waymo runs 14 driverless vehicles for every one Tesla runs in Texas. State registration filings analyzed by CNBC in May showed 577 registered driverless Waymos against 42 Teslas in the state, a ratio that has stayed lopsided even as Tesla has begun operating a Robotaxi service in Austin.
Waymo's July 8 announcement opens fully driverless service in Las Vegas, San Diego, Tampa, and Denver. The four-city expansion is the visible event. The federal investigation into Tesla's camera-only approach is what makes the Texas fleet gap the lead.
NHTSA's Engineering Analysis EA26002 covers roughly 3.2 million Tesla vehicles equipped with Full Self-Driving (FSD) software. An Engineering Analysis is the step before a recall recommendation or a finding of no defect, and it puts Tesla's camera-only architecture under federal safety review for the first time.
Tesla's FSD and Robotaxi product runs on cameras alone. The company argues that human drivers navigate on vision, and that high-resolution cameras backed by neural networks can replicate that ability without lidar (a laser sensor that builds a 3D map of the world) or radar. Waymo's driverless service runs a multi-sensor stack: lidar, radar, and cameras in parallel, with the lidar providing depth and distance data that cameras cannot reliably infer at highway speed.
The two designs answer different safety questions. NHTSA's probe is the first time the federal government has treated the choice between them as a safety investigation rather than a marketing argument. TechCrunch independently confirmed Waymo's Texas registration lead on the same day as CNBC's filing analysis, giving the 14-to-1 figure two independent readings.
Waymo's announcement post puts Las Vegas on driverless service immediately, including the Strip, starting with Alphabet employees and opening to a public waitlist in the coming months. San Diego, Tampa, and Denver follow in staged rollouts.
Denver is the test that the other three are not. Las Vegas, San Diego, and Tampa have warm, dry, well-mapped weather. Denver introduces snow, ice, road glare, and low winter sun angles that distort camera inputs and stress every layer of a sensor stack. For Waymo's multi-sensor architecture, lidar returns are largely unaffected by sunlight, which gives the company a designed-in advantage in the conditions Denver will see. For Tesla's camera-only approach, those same conditions are where vision systems have historically struggled.
Waymo published safety performance data in March 2026 showing low crash-severity rates against a human-driver baseline. The numbers are company-published, and the comparator set is one the company selected. They are useful as a self-reported benchmark, not an independent verdict. Waymo's total US driverless service area already exceeds 1,400 square miles, which gives the data set a denominator that earlier-generation autonomous pilots did not have.
Tesla does not publish comparable crash-rate data for FSD, and the NHTSA investigation is partly about closing that information gap.
Three clocks now tick together. The four-city rollout, where Las Vegas going fully driverless on day one is the only opening that reads as immediate. NHTSA EA26002, which can move from Engineering Analysis to a recall recommendation or a finding of no defect on a timeline measured in months, not weeks. And Denver's first sustained winter, which will be the first empirical verdict on whether a multi-sensor stack performs as designed when the weather turns.
The 14-to-1 Texas fleet gap is not a finish line. It is the visible surface of an architectural bet now under federal review, with a cold-climate experiment already on the calendar.