Tesla’s driverless robotaxi fleet in Austin has now recorded 14 crashes since its commercial launch, according to federal crash data released under a mandatory reporting order. That growing incident count lands at a moment when a rival bet is taking shape: Uber has committed up to $1.25 billion in Rivian to put as many as 50,000 R2 robotaxis on roads, with initial deployments expected in 2028. The split screen between a live fleet accumulating crash reports and a billion-dollar wager on vehicles that have not yet carried a single passenger raises a direct question for regulators, investors, and riders about how safety records will shape the next phase of autonomous driving.
Why 14 Tesla robotaxi crashes sharpen the regulatory stakes
The National Highway Traffic Safety Administration requires every company operating vehicles with automated driving systems to file incident reports under a standing crash order. Public monthly CSV data releases extend through at least April 15, 2026, giving outside analysts a running tally of each company’s crash events. Tesla’s 14 reported incidents place it among the most scrutinized operators in that dataset, and the agency has separately opened a preliminary evaluation, cataloged as PE24031, focused on incidents involving Tesla’s unsupervised robotaxis in Austin.
That probe matters because NHTSA has the authority to escalate a preliminary evaluation into an engineering analysis and, eventually, a recall order if it finds a safety defect. Each new crash logged in the monthly data adds to the evidentiary base the agency uses to decide whether to tighten operating conditions or demand software changes. For riders and city officials in Austin, the practical effect is straightforward: a rising crash count could trigger restrictions on where and when Tesla’s vehicles are allowed to operate without a human driver.
The timing also shapes how other cities might evaluate Tesla’s expansion requests. Municipal transportation departments often look to federal investigations as a proxy for risk; an open NHTSA probe can translate into stricter permitting terms, narrower operating zones, or requirements for human safety operators that cut against the economics of a fully driverless service. If PE24031 advances, Tesla could face a patchwork of local rules that slow its robotaxi rollout even if the federal process stops short of a recall.
The tension is sharpened by the fact that Uber’s Rivian fleet does not yet exist on public roads. When those R2 vehicles eventually begin carrying passengers, they will enter the same federal reporting pipeline. But because the Rivian–Uber robotaxis are not expected to deploy until 2028, NHTSA will have years of Tesla crash data in hand before a single Rivian incident is filed. That asymmetry could shape how aggressively regulators treat Tesla’s current operations relative to a competitor that arrives with a clean sheet and the benefit of hindsight about what went wrong in earlier deployments.
Uber’s $1.25 billion Rivian deal and the contract details on file
Rivian disclosed the binding terms of its Uber partnership in a quarterly SEC filing for the period ended March 31, 2026. The document lists a Subscription Agreement with Uber, dated March 18, 2026, as Exhibit 10.1. That exhibit reference confirms the deal has moved beyond a press announcement into a formal, audited disclosure subject to securities law and internal controls.
Under the agreement, Uber plans to invest up to $1.25 billion in Rivian, with the capital spread through 2031 and tied to milestones. Uber expects to purchase 10,000 R2 robotaxis from Rivian and holds an option for 40,000 additional units, bringing the potential fleet to 50,000 vehicles. Initial deployments are expected beginning in 2028, according to the same reporting, positioning the R2 as a central pillar of Uber’s long-term shift away from human drivers in at least some markets.
The milestone structure means Uber’s full financial commitment is conditional. If Rivian fails to hit unspecified performance targets along the way, the total investment could land well below $1.25 billion. The full executed text of the Subscription Agreement has not been published beyond the exhibit index in the 10-Q, so the precise triggers that unlock each tranche of capital are not yet public. That gap matters for anyone trying to assess how firm the 50,000-vehicle target really is, and whether Uber can easily slow or shrink the program if technology, regulation, or capital markets turn against large-scale robotaxi bets.
Still, the basic outline is clear enough to influence competitors and regulators. A committed anchor buyer for tens of thousands of vehicles gives Rivian a planning horizon for factory capacity and software development. For Uber, the agreement signals to investors that the company intends to remain a central player in autonomous ride-hailing even if it relies on partners, rather than building its own vehicles. And for policymakers, it underscores that by the time they finish digesting the safety record of Tesla’s Austin fleet, a much larger wave of purpose-built robotaxis could be queuing up behind it.
Crash data gaps and the Rivian timeline leave key questions open
The 14 crashes attributed to Tesla’s robotaxis are drawn from aggregate federal data that does not include per-incident narratives or fault determinations. NHTSA’s monthly CSV files record that an incident occurred and which company’s system was engaged, but they do not tell the public whether the automated vehicle was at fault, whether injuries resulted, or what road conditions were present. Without that detail, comparing Tesla’s safety record to human-driven baselines or to future autonomous fleets requires assumptions the data does not support.
That opacity complicates debates over whether 14 crashes represent an acceptable rate for a new technology. A fleet that drives millions of miles with a handful of mostly minor incidents could be safer than human drivers, while the same count over far fewer miles, or with severe injuries, would be a red flag. Tesla has not published granular exposure metrics specific to its Austin robotaxis, leaving outsiders to infer risk from partial signals: the raw crash tally, the existence of PE24031, and any future enforcement actions or software updates that NHTSA might publicly reference.
NHTSA’s preliminary evaluation PE24031 has been listed on the agency’s recall portal, but updated investigation findings beyond the initial citation have not been released. Whether that probe results in a formal engineering analysis, a recall, or a closure with no action will be one of the clearest signals of how federal regulators judge Tesla’s unsupervised driving technology. A recall could force Tesla to curtail operations or push over-the-air changes that materially alter how the system behaves in city traffic, while a closure might be read as a tentative endorsement of the current safety case.
On the Rivian side, the absence of any real-world operating data creates its own uncertainty. The R2 has not carried a paying passenger, and the specific autonomous driving software stack that will power the Uber fleet has not been vetted through the same federal reporting pipeline that now tracks Tesla. By the time Rivian vehicles begin service in 2028, the regulatory bar may have moved, shaped by whatever NHTSA learns from Tesla and other early deployments.
That lag cuts both ways. Rivian and Uber can study years of Tesla crash data, public investigations, and any resulting software or policy changes before finalizing their own safety cases. They can design around known failure modes, emphasize conservative behavior in edge cases, and preemptively adopt reporting or transparency practices that address current criticisms of opaque crash statistics. At the same time, they will face expectations informed by Tesla’s record: city officials and federal regulators may demand stronger demonstrations of safety before approving large, fully driverless fleets.
For now, the contrast is stark. Tesla is learning in public, with 14 crashes and an open NHTSA evaluation framing debates over whether its Austin robotaxis are ready for prime time. Uber and Rivian, by comparison, are still on paper, backed by a multiyear, milestone-based investment but with no incidents, no miles, and no riders to point to. How those two narratives converge-one grounded in imperfect but real crash data, the other in contracts and projections-will help determine which companies, and which regulatory models, define the next decade of autonomous transportation.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
