2026-06-18 — views

Autonomous Trucking — Why Highway Freight May Beat Robotaxi to Profitability

Aurora launched driverless freight on I-45. AV trucking saves 10x more labor per mile, targets a 4x bigger market than ride-hail, and may reach profit first.

Article 77 in the Physical AI Benchmark Series — Autonomous Trucking

Autonomous trucking is the Physical AI application that may reach commercial profitability before robotaxi — not because the technology is easier, but because the economics are clearer. The US trucking industry is an estimated $800B market facing a structural driver shortage of 80,000+ drivers (est.). Aurora has already launched commercial driverless freight on the Dallas-to-Houston I-45 corridor. The unit economics favor trucks over taxis by a wide margin. And the transfer-hub model means AV trucking companies do not need to solve urban driving to deliver commercial value today.

Section 1 — Why Trucking Is Easier Than Robotaxi (and Where It’s Harder)

Highway freight operates on fixed, predictable routes without the pedestrian complexity of urban robotaxi. That asymmetry matters enormously for commercial deployment timelines.

Dimension	Highway trucking	Urban robotaxi
Environment	Highway: predictable, few pedestrians, wide lanes, controlled access	Urban: pedestrians, cyclists, complex intersections, unprotected left turns
Speed	65–75 mph highway cruise	25–35 mph urban stop-and-go
Interaction complexity	Primarily other large vehicles; on-ramps/off-ramps; lane changes	Pedestrians, cyclists, scooters, double-parked vehicles, construction
Route predictability	Fixed origin-destination corridors; same highway every run	Highly variable; any street in the operating zone
Weather exposure	Highway weather can be severe; critical safety consideration	Urban weather more manageable; AV can pull over
Loading/unloading	Human dock workers handle terminal operations; AV does only the highway drive	AV must handle full door-to-door including complex urban navigation
Regulatory environment	Highway AV regulations clearer in TX, AZ, TN; interstate commerce federal oversight	State-by-state; city permits required
Where trucking is harder	High speeds mean less reaction time; large vehicle = more damage in incident; long-haul fatigue simulation	N/A

The critical insight: AV trucking companies do not need to solve the full self-driving problem to generate revenue. They only need to solve the highway segment — which is dramatically simpler than urban driving. Human drivers handle the first and last mile. The AV handles everything in between.

Section 2 — Aurora: The Commercial Launch

Aurora Innovation became the first company to operate fully driverless (no safety driver) commercial freight at scale in the United States.

Metric	Details
Commercial launch	Aurora launched driverless freight operations on I-45 (Dallas-Houston corridor) in April 2024
Partners	Volvo and PACCAR (Kenworth/Peterbilt) are Aurora’s truck OEM partners; FedEx and Uber Freight are early commercial customers
Vehicle	Aurora Driver software running on Volvo VNL and Kenworth T680 Class 8 trucks
Aurora Driver	5th-generation AV system; lidar (Aurora’s own FirstLight lidar), cameras, radar; designed specifically for highway freight
FirstLight Lidar	Aurora’s proprietary FMCW (frequency-modulated continuous wave) lidar — measures velocity directly; longer range than conventional lidar (~400m range, est.)
Route	I-45 Dallas-Houston (~240 miles); one of highest-volume freight corridors in the US
Safety driver status	Driverless — no human in the cab on commercial runs
Nasdaq	Aurora (AUR) is publicly traded; survival as a standalone company depends on reaching commercial scale
Revenue model	Charge per-mile to freight carriers; Aurora drives the truck, carrier handles terminal operations

Aurora’s FirstLight lidar uses frequency-modulated continuous wave (FMCW) technology, which differs from conventional time-of-flight lidar. FMCW lidar measures the velocity of objects directly in addition to their position — a significant advantage for highway applications where closing speeds between vehicles are high. The estimated ~400m range (est.) gives Aurora substantially more reaction distance than conventional lidar systems.

The commercial partnership structure is notable. Aurora is not buying and operating trucks — it is selling miles to FedEx and Uber Freight who provide trucks and handle terminal operations. This asset-light model limits Aurora’s capital requirements but means Aurora’s revenue scales with miles driven, not assets owned.

Section 3 — Waymo Via: The Freight Arm

Waymo Via is Waymo’s trucking division, applying the same sensor suite and software maturity that powers the Waymo One robotaxi fleet to Class 8 freight.

Metric	Details
Name	Waymo Via — Waymo’s trucking division
Status	Testing phase; commercial launch timeline not publicly confirmed (est. 2025–2026)
Technology	Same Waymo Driver sensor suite as passenger vehicles; adapted for Class 8 trucks
Partners	J.B. Hunt (major US trucking carrier) announced partnership; Ryder for fleet management
Advantage	Inherits Waymo’s sensor suite + software maturity from 10+ years of development; proven safety record
Route focus	Sun Belt freight corridors; Texas and Arizona initially
Differentiation	Waymo Via positions as a freight-as-a-service platform — OEMs provide trucks, shippers provide freight, Waymo provides the driver software

Waymo’s structural advantage in Via is its 10+ years of AV development and the largest real-world autonomous miles dataset in the industry (est.). Waymo Driver has accumulated more driverless miles than any competitor across its passenger fleet. Adapting that system to highway freight — a simpler environment than urban San Francisco — is a logical extension of a mature technology platform.

The freight-as-a-service positioning means Waymo does not need to own or lease trucks. J.B. Hunt and Ryder bring the existing fleet relationships, terminal infrastructure, and freight network. Waymo brings the driver software. This is the same platform logic Aurora is using — but Waymo enters with a more mature software stack.

Section 4 — Other AV Trucking Players

The AV trucking landscape includes companies at every stage from early development to driverless commercial operations.

Company	Status	Notes
Kodiak Robotics	Commercial testing; driverless runs in Texas	Spun out of Uber ATG; focus on Texas freight corridors; raised ~$250M (est.)
Gatik	Commercial operations — short-haul B2B middle-mile	Walmart, Kroger customers; fixed routes between distribution centers and stores; driverless in Arkansas; most commercially mature short-haul AV (est.)
Torc Robotics (Daimler)	Development phase; Daimler Truck majority acquisition	Mercedes-Benz Trucks-backed; long-haul highway focus
TuSimple	Effectively ceased AV operations 2023	Cautionary tale — Chinese founders, US security concerns, CFIUS issues, technology sales to China; Nasdaq delisted
Locomation	Convoy (platooning) approach	Truck platoons with human lead driver + autonomous follow trucks; lower cost than fully driverless; less certain safety argument
Plus.ai	Highway AV software; Series B funded	Sells AV software to OEMs; international presence (China + US)

TuSimple is the cautionary tale for the sector. Once one of the most-funded AV trucking startups, TuSimple effectively ceased US AV operations in 2023 following CFIUS concerns about its relationship with Chinese entities and reports of unauthorized technology sharing. Its Nasdaq delisting underscores the national-security overlay that applies to AV technology with dual-use potential. Any AV trucking company with Chinese capital or technology partnerships now faces additional regulatory scrutiny in the US market.

Gatik represents a distinct commercial model. Rather than long-haul highway freight, Gatik operates fully driverless on fixed urban-adjacent routes between warehouses and retail stores — the “middle mile” of B2B logistics. Its customers include Walmart and Kroger. The fixed-route model eliminates open-ended urban complexity while still operating in urban environments. Gatik claims to be the most commercially mature short-haul AV company by miles driven (est.).

Section 5 — Unit Economics: Why Trucking Wins

The economic case for AV trucking is structurally stronger than the case for robotaxi. The labor cost saved per mile is larger, the revenue per mile is higher, and the addressable market is substantially bigger.

Economic factor	AV trucking	AV robotaxi
Driver cost replaced	US truck driver: $70,000–$90,000/yr salary + benefits; ~$1.50–2.00/mile labor cost (est.)	Rideshare driver: variable per ride; lower hourly rate but no long-haul stays
Revenue per mile	Freight rate: ~$2.00–$4.00/mile for full truckload (est.)	~$1.50–$2.50/mile for robotaxi (est.)
Utilization	Highway truck: 10–11 hours/day (HOS regulations limit human; AV not subject to Hours of Service limits)	Robotaxi: 40–60% utilization in mature market (est.)
AV utilization advantage	AV truck can drive 22+ hours/day (charging + loading only constraint) vs human 11-hour HOS limit	Robotaxi utilization improvement vs Uber/Lyft driver is smaller
Vehicle cost	Class 8 truck: $150,000–$200,000 (est.) + AV system ~$50,000–$100,000 (est.)	Waymo custom vehicle est. $150,000+; Cybercab target sub-$30,000
Per-mile AV payback	Replacing a $80K/yr driver on a 100,000 mile/yr truck: ~$0.80/mile savings	Replacing a driver earning ~$0.60/mile: smaller absolute saving
Market size	US trucking ~$800B/yr (est.); long-haul segment ~$200B (est.)	US ride-hail ~$50B (est.)

AV trucking has an estimated 10x the labor cost savings per mile and 4x the addressable market of ride-hail. The reason robotaxi received more press coverage is that consumers interact with it directly. Freight moves invisibly — but at a scale that dwarfs the consumer mobility market.

The Hours of Service (HOS) advantage is underappreciated. Federal regulations cap human truck drivers at 11 hours of driving time per 14-hour on-duty window, with mandatory rest periods. An AV truck is not subject to HOS limits. A Dallas-to-Houston run is approximately 4 hours each way. An AV truck could theoretically complete 5+ round trips in a 24-hour period versus 2 round trips for a human driver. At $2–4 per mile (est.) and 240 miles per run, the utilization advantage compounds into significant revenue per vehicle per day.

Section 6 — The Transfer Hub Model

Most AV trucking companies use a “transfer hub” model to capture the economic value of autonomous highway driving without needing to solve urban navigation.

The model works as follows:

AV truck handles the highway segment only (more than 95% of miles, more than 80% of time)
Human driver handles the urban terminal to highway on-ramp (“first/last mile”)
AV truck pulls into a transfer hub at the city edge; human driver takes over for urban delivery

Why the transfer hub model works:

Eliminates the need for AV to navigate complex urban environments today
Lets AV companies deploy commercially without solving the full urban problem
Creates a new infrastructure category: transfer hubs at interstate highway interchanges
Allows gradual automation — highway automation now, urban automation later as the technology matures

Gatik’s model goes further — fully driverless on fixed urban routes between warehouses and stores, but with controlled, predictable paths that eliminate open-ended urban complexity. Gatik effectively created its own version of the transfer hub model, choosing routes where the urban environment is nearly as predictable as a highway.

The transfer hub model also has implications for real estate and logistics infrastructure. As AV trucking scales, highway interchange locations near major freight corridors will attract transfer hub development. These facilities — where AV trucks hand off to human city drivers — represent a new asset class in industrial real estate (est.).

Section 7 — About This Series

This is article 77 in the Physical AI Benchmark Series. Previous articles have covered the ramp index, the humanoid race, unit economics, global competition, HD mapping, software and OTA, consumer demand, competitive moats, Cybercab versus Model Y, safety data, Waymo Gen 6, Optimus manufacturing, scorecard snapshots, 2030 forecast scenarios, the investor framework, city expansion pipelines, Tesla FSD state approval maps, AV weather and climate constraints, the talent war, regulatory calendars, robotaxi fare pricing, humanoid deployment trackers, supply chain analysis, consumer adoption demand index, valuation and IPO analysis, the Physical AI 2026 mid-year roundup, AV unit economics cost-per-mile breakdown, the AV data flywheel comparison, AV cybersecurity attack surfaces, the Physical AI supply chain, AV fleet operations, AV insurance and liability evolution, the full lifecycle environmental cost of Physical AI, the accessibility layer, the mapping architecture comparison, the China AV race, simulation and synthetic data training, the Physical AI investment landscape, and AV urban planning city impact.

This article adds the freight dimension: why autonomous trucking may reach commercial profitability before robotaxi, the unit economics that favor trucks over taxis, and the transfer hub model that enables commercial deployment today without solving urban driving.

Note: Market size estimates, driver shortage figures, cost per mile, vehicle cost estimates, and utilization projections are labeled “(est.)” and reflect publicly available industry data and analysis where available. This article does not constitute investment advice.