Last price
$5.27
Autonomous long-haul trucking at commercial launch — the most literal services-as-software deal in the index: a truck driver sold per mile.
Live quote sourced from Yahoo Finance. Prices cited in narrative below reflect the original memo date and may be stale.
A literal services-as-software company: a human driver is replaced by an AI system and the customer pays by the mile. The TAM is the $900B US trucking market; the wedge is the ~$200B long-haul-driver wage bill. Aurora is vertically integrated — perception stack, vehicle platform via OEM integration, operations via Horizon, and customer relationship via carrier contracts. The thesis win condition is a combination of safety-case maturity, lane coverage, and a reasonable per-mile take on the freight invoice. The business is early-revenue, pre-profitability, capital-intensive by software standards — but intellectually the most direct public-market expression of the thesis.
Aurora sits at the narrow intersection of two bets: (1) autonomous trucking is easier than autonomous robotaxis (fewer edge cases, geofenced lanes, professional fleet buyers) and is therefore the likeliest path to driver-out commercial operation in the next 36 months; (2) vertical integration of the Driver + Horizon + OEM relationships is the correct winning structure, more defensible than tier-1 licensing or pure-software models. If both are right, Aurora becomes the rails of a trillion-dollar labour-substitution trade. If either is wrong — if robotaxi transfer learning arrives faster, or if carriers commoditise the Driver layer — the moat is thinner and margin structure looks more like licensing than outcomes.
US long-haul has a chronic driver shortage (est. 80K+ unfilled seats), wages compounded ~6%/yr pre-pandemic, and Hours-of-Service rules structurally cap human productivity at ~11 hrs/day. An AI driver doesn't sleep, doesn't strike, and doesn't take the 34-hour restart. Carriers get more utilisation per truck, shippers get faster lanes, and the per-mile cost equation improves materially even at moderate take rates. The partner list (PACCAR, Volvo, Uber Freight, Ryder, Werner, Hirschbach, Schneider) implies the industry is buying the integration model over tier-1 or pure-software alternatives.
Capex and opex to run an autonomous fleet at commercial scale are enormous; Aurora's $700M+ annual burn means dilution or debt remain live risks even after 2025 raise. Tesla's FSD stack, Waymo's commercial operations, Kodiak, and Gatik are all running somewhat adjacent programmes; technology transfer from robotaxi could compress Aurora's head-start. A single high-profile Class 8 autonomous crash could trigger regulatory reversal, public insurance-market withdrawal, or moratoria, any of which would threaten the commercial trajectory.
| Lane / product | Status | Thesis fit | Revenue shape |
|---|---|---|---|
| Dallas–Houston (TX-45) | Commercial driverless since Apr-2024 | Core | Per-mile outcome contract |
| Fort Worth–El Paso | Ramping 2025-26 | Core | Per-mile outcome contract |
| Phoenix metro + AZ–TX | Early lanes 2026 | Core | Per-mile outcome contract |
| OEM-integrated hardware kit (PACCAR/Volvo) | Shipping | Supporting | Per-truck license + per-mile run-rate |
| Horizon terminals + ops services | Live | Supporting | Operational services fees |
Aurora crossed the commercial driverless line in 2024 and expanded in 2025. Miles, disengagement rate, and accident statistics are now empirical rather than projected. That's a step-function in credibility that only Waymo has previously cleared in the US.
Competing tier-1 vendors must build the stack without carrier relationships; pure-software rivals must build OEM and operations. Aurora has OEM silicon integration, Horizon terminal operations, and freight partnerships all simultaneously. Replication cost is multi-year and multi-billion-dollar.
Per-mile pricing on $200B of driver labour is a fundamentally larger pool than any vertical-SaaS seat budget. Even single-digit market share represents billions in outcome revenue — with software-like incremental economics once hardware integration is done.
PACCAR and Volvo shipping AV-ready platforms with Aurora integration means competitors must either win OEM slots of their own or retrofit, both of which are expensive and slow. The window of OEM exclusivity is a real structural advantage.
At ~$700-800M annual burn and small revenue, Aurora depends on equity markets staying open. Any prolonged valuation compression forces dilutive raises, debt, or strategic partnership at a discount. The thesis is right only if Aurora survives to scale.
If Tesla's consumer FSD dataset transfers credibly into Class-8 operations, Aurora's per-mile data advantage compresses. That is the most-often-cited and most-disputed scenario among credible analysts; it is neither confirmed nor disproven at this writing.
The political economy of AV trucking is more fragile than software. NHTSA, state DOTs, insurers, and public opinion respond asymmetrically to Class-8 incidents. A single high-visibility accident could freeze lane expansion for months and trigger insurance withdrawal.
Early contracts may reflect pilot-economics; steady-state per-mile take rates are not yet public. Carriers are sophisticated margin buyers and will compress Aurora's take as lanes mature. The thesis depends on maintaining healthy unit economics at scale.
Aurora is the most literal implementation of the thesis in public markets: customers pay per unit of delivered outcome (a mile driven) and zero humans are in the loop. On thesis *purity* it may be higher than anything else in the set. The reason the verdict is Positive rather than Highly Positive is execution risk: Aurora is not yet cash-flow positive, capital intensity is high, and the competitive response from Tesla and Waymo is not fully knowable. The thesis-fit story is clearer than the financial story.
Most literal services-as-software story in public markets. Own as a leveraged long-duration call on labour-substitution in trucking; size accordingly for burn and competitive risk.