Hello from Editor Z. On I-45 between Dallas and Houston, and on I-20 between Fort Worth and El Paso, there are semis running right now with genuinely no one in the driver's seat — not a safety monitor, not a backup, no one. That's the reality of driverless semis in late September 2026, and the number that should worry, or reassure, anyone who cares about it isn't the mileage count. It's the fleet math. Aurora told CNBC on September 22 that it's running roughly 20 fully driverless trucks today and plans to have 200 by the end of this year. Kodiak just named that same Dallas-Houston lane its own long-haul driverless launch route. And California just handed both companies permits to start testing on its highways. This isn't a demo anymore. It's a fleet decision three companies are making at once.
Two Corridors, Zero Drivers
Aurora's Sun Belt Network
Aurora Innovation launched commercial driverless service in Texas in May 2025, and it has since built out a network of Sun Belt routes that now includes Dallas-Houston, Fort Worth-El Paso, El Paso-Phoenix, and Dallas-Laredo. The company told CNBC it's currently running about 20 trucks with nobody aboard and is targeting 200 driverless trucks by the end of 2026 — a tenfold jump in roughly three months. Before any of that, Aurora had spent close to five years running test vehicles with safety drivers along I-10 out of Houston, which is the kind of quiet runway that rarely makes headlines but is exactly what has to happen before a company lets a truck run a real freight route alone.
Kodiak Picks the Same Lane
Kodiak AI, meanwhile, has taken a different route to the same corridor. The company built its driverless track record hauling freight with no one in the cab for Atlas Energy Solutions in the Permian Basin starting in December 2024 — off-highway, low-speed, oilfield roads. In August 2026 it began a driverless long-haul launch program, and it has now named Dallas-Houston as its long-haul driverless launch lane, saying its safety observers aren't touching the wheel anymore on that route, including on surface streets. Kodiak is targeting an unsupervised, fully driverless highway service on that lane by the end of 2026, which would put two separate companies running unmanned trucks on the exact same stretch of I-45 within months of each other.
The Hardware That Makes the Scale-Up Possible
Built for a Million Miles
Aurora's fleet jump isn't just a bigger order — it's riding on new hardware. The company launched its second-generation driverless truck on July 22, 2026, built around a sensor and compute kit designed to last a million miles at 50% or more lower cost than the previous generation. That cost curve matters more than it sounds: Aurora isn't just trying to prove the technology works, it's trying to prove it can be manufactured and maintained cheaply enough to put on hundreds of trucks without the economics falling apart.
Roush's Production Ramp
The trucks themselves are International Motors LT Series tractors, upfitted with Aurora's driving system by Roush Industries, which is targeting a production rate of 20 trucks a week by the end of 2026. Twenty trucks a week is the kind of number that only makes sense if you actually believe you're going to need 200 of them by December — it's a manufacturing commitment, not a pilot-program order.
The Money Behind Aurora's Milestone
A Real Path to Breakeven, Not Just a Mileage Count
CNBC's September 22 report is really a story about Aurora trying to convince investors this is now a business, not a science project. Aurora is guiding to $14 million to $16 million in revenue for 2026, with an exit revenue run rate around $80 million — a sign that most of this year's growth is loading into the back half. The company is targeting gross-margin breakeven on a run-rate basis in the first half of 2027, at around 500 driverless trucks, and it's targeting positive free cash flow on a run-rate basis by the time it exits 2028.
The 2030 Bet
Further out, Aurora is targeting more than $5 billion in annual revenue and roughly 60% gross margins by 2030 — SaaS-like margins for a trucking company, which is the entire pitch: a driverless truck, Aurora argues, can move freight more cheaply than one with a driver behind the wheel once the fleet is big enough to spread the technology cost across enough miles. Whether that holds is precisely what the 200-truck target by December is meant to start proving.
Kodiak's Oilfield-First Playbook
From 28 Trucks to 100
Kodiak's approach to scale has been more gradual and more industrial. As of March 31, 2026, Atlas Energy Solutions was running 28 Kodiak-powered driverless trucks across 15 routes in the Permian Basin, logging more than 23,500 driverless hours in the first quarter alone. By June 30, 2026, that fleet had grown to 35 trucks and surpassed 40,000 cumulative hours of paid driverless operations. Atlas and Kodiak now plan to expand that fleet to 100 driverless trucks by mid-2027, with public highway driving across Texas and New Mexico targeted for early 2027 — oilfield roads first, then the open road.
Pulling the Safety Driver
The bigger swing is Kodiak's push to pull safety drivers off its highway routes entirely by the end of 2026, starting with Dallas-Houston. That's a materially different bet than Aurora's approach of building a second-generation truck first and scaling after: Kodiak is trying to prove the unsupervised model works on one lane before expanding it, while Aurora is trying to prove the hardware and manufacturing pipeline can support scale everywhere at once.
California Says Yes, With Conditions
What the New Permits Actually Allow
On April 28, 2026, the California DMV approved updated rules lifting its ban on driverless testing for vehicles over 10,000 pounds — the weight class that covers commercial semis. Both Aurora and Kodiak applied that spring, and by mid-August 2026 both had received permits to test on California public roads, with Kodiak already running a handful of test trucks near its Mountain View office. It's a meaningfully smaller step than what's happening in Texas: California's permits still require a human safety operator behind the wheel at all times, and they bar operation on any road posted at 25 mph or under, unless it's the only direct route between two points.
The Teamsters Pushback
Not everyone welcomed the rule change. Teamsters California filed a lawsuit challenging the DMV's approval, arguing the agency didn't adequately study the economic impact on drivers or run a sufficient safety risk assessment before opening the door to heavy-duty autonomous testing. That lawsuit is still pending, and it's a preview of the fight that will follow both companies into any state where they try to remove the safety driver for real, not just for a supervised test permit.
| Company | Aurora | Kodiak AI |
|---|---|---|
| Current driverless routes | Dallas-Houston (I-45), Fort Worth-El Paso (I-20), El Paso-Phoenix, Dallas-Laredo | Permian Basin oilfield routes; Dallas-Houston highway launch underway |
| Driverless trucks today | ~20 | 35 (Atlas fleet, as of June 30, 2026) |
| Near-term fleet target | 200 by end of 2026 | 100 (Atlas fleet) by mid-2027 |
| Safety driver status | Already removed on Sun Belt routes | Targeting removal on Dallas-Houston highway lane by end of 2026 |
| California status | DMV testing permit (Aug. 2026), safety driver required | DMV testing permit (Aug. 2026), safety driver required |
| 2026 revenue guidance | $14M-$16M, $80M exit run rate | Not disclosed in these sources |
What Truckers Are Actually Saying
On TheTruckersReport's industry forum, reaction to Aurora's second-generation rollout has been a mix of grudging respect and real anxiety. One experienced trucker described watching an Aurora truck handle a shoulder situation the way a good human driver would — moving over for a vehicle parked on the shoulder, signaling properly, holding its lane discipline. But that same thread carried a heavier undercurrent: drivers worried aloud about an industry "creating its own demise" without a clear transition plan, and pushed back hard on the idea that new jobs will simply appear to replace the ones automation removes, asking pointedly what those new jobs would actually be. Opinions on timeline split sharply — some drivers still expect widespread driverless trucking is 10 to 20 years out, while others, watching Aurora's own 200-truck target for this December, think 2 to 3 years is closer to right.
What to Watch Next
The next few months will tell you more than the last year of press releases has. Watch whether Roush actually hits 20 trucks a week and whether Aurora's fleet is really at 200 by December 31 — that number is specific and dated enough to be a real test, not a vague promise. Watch whether Kodiak actually pulls its safety drivers on Dallas-Houston before year-end, which would mean two different companies running fully unmanned trucks on the same Texas highway. And watch the Teamsters California lawsuit, because however that case lands will shape how fast — or how slowly — every other state follows California's lead on heavy-duty autonomous permits. The pilot-program era of driverless trucking is closing. What replaces it is still being negotiated, on the highway and in court.
-EditorZ
Photo by Bhargav Panchal on Unsplash
Post a Comment