Own It, Rent It, or Bid Against It: A Contractor’s Guide to the $2 Billion Autonomy Land Grab
In about five weeks, more than 2 billion dollars flowed into companies building heavy machinery that drives itself. But “autonomy” hides five very different business models, and the one you buy decides whether self-driving kit becomes your capability, your subcontractor, or your competition. Here is how to tell them apart before a vendor gets you excited by a stat you can’t verify.
The machines that move dirt are learning to drive themselves, and the money has arrived in a hurry. In roughly five weeks this summer, well over 2 billion dollars went into companies building self-driving heavy equipment. Some of that capital, like Travis Kalanick’s headline raise, spans mining and transport as well as construction, so treat the number as a signal of appetite rather than a pure construction figure. The appetite is real either way.
Here is the useful part, and it’s the bit most coverage skips. Look under the bonnet and this is not one market. There are five distinct ways to sell a contractor a machine that runs itself, and they behave nothing alike. Miss the difference and you might sign up to compete with the company you thought you were hiring. We broke it all down on this week’s Executive Weekly Briefing, which you can listen to here.
So what’s in it for you? A clean way to read the pitches hitting your inbox, sharper questions for the demo, and a feel for which corners are genuinely proven versus still finding their feet.
Prefer it in your ears? Here’s the full breakdown from this week’s Executive Weekly Briefing, with every figure and source. Hit play.
A Two Billion Dollar Vote of Confidence
The timeline that turned a maybe into a market
String the announcements together and the shift is hard to miss. Two of them stand out. Travis Kalanick, the man who built Uber, closed a 1.7 billion dollar round for Atoms with an autonomous mining arm inside it. And when a giant like Komatsu partners with a startup to field driverless dozers, that’s an old-guard OEM telling you where it thinks the puck is going.
- TerraFirma, ~$115M. Series A led by Kleiner Perkins for semi-autonomous earthmoving fleets.
- Atoms, $1.7B. Kalanick’s raise, led by a16z, with mining autonomy folded in.
- Komatsu x AIM. A major OEM partnering to field autonomous dozers and excavators.
- Gravis, $200M. SoftBank backing a retrofit system already deployed across four continents.
In plain terms, investors have decided self-driving heavy equipment is a real business rather than a slide-deck fantasy, and they are paying steep prices to grab a seat before the table fills. That’s the headline. The money you actually need to understand sits one layer down.
Five Business Models Hiding Under One Word
“Autonomy” is a category, not a product
Every company above will happily call itself an autonomy company. Useful for a headline, useless for a purchase order. Underneath, there are five separate plays, and each puts you in a different spot.
The same underlying tech, self-driving heavy kit, arrives through five different doors. Which door you walk through matters more than the sensor spec on the brochure.
Own It, Rent It, or Bid Against It
The model decides the relationship you end up in
Buy a retrofit kit and autonomy becomes your capability. You keep the fleet, the resale value and the control, and you carry the learning curve. Hire a single-task robot by the job and it becomes a subcontract line, handy for a scope you’d rather not staff, with none of the ownership. Those two are easy to reason about.
The awkward one is model two. When an autonomous contractor like TerraFirma bids the earthworks itself, that same clever technology shows up as a rival on the tender list. Kalanick made a related point worth chewing on: if you’re building machines that move through the physical world, you can’t lean on one or two suppliers for the self-driving brain, you have to own that layer yourself. That’s why the smartest players treat autonomy as a platform they build once and point at industry after industry, not a feature they license.
Do you want to own autonomy, rent it, or find yourself bidding against it?The question that decides which vendor conversation you should even be having
Read the room before your competitors do
The deals, the tech and the honest takes on what’s actually working on site, every week.
Join 3000+ Readers
The Honest Part: A Maturity Gap Runs Down the Middle
Proven in mining, still early on the jobsite
Here’s the caveat that keeps you out of trouble. Mining autonomy is genuinely mature. Driverless haul trucks are boringly normal in parts of Australia and Chile, and Caterpillar puts its fleet near 700 machines that have moved billions of tonnes. Construction autonomy is younger, and it arrives one machine type and one workflow at a time. An excavator doing mass dig on an approved site is a long way from a full fleet running your whole job.
There’s a second thing to hold. A good chunk of the numbers flying around, the hours, the tonnes, the productivity gains, come from the companies themselves. Even Gravis notes its up-to-30-percent productivity figures are its own and not independently verified. Treat every vendor stat as a claim until someone neutral has checked it.
Practical upshot: when a rep pitches you autonomy, ask two questions. Which of the five models are you, and which of these numbers has been verified by someone who isn’t you? The answer to the second tells you more than the demo ever will. The machines are here and live. Being clear-eyed is how you benefit.
The Five Models at a Glance
A quick map before your next vendor call
One table, five plays, and the question each one really asks of your business.
| Model | What you’re buying | Example players | Maturity | What it means for you |
|---|---|---|---|---|
| Retrofit platform | A kit that drives your existing machine | Bedrock, Gravis, AIM | Early, live on real sites | Autonomy as your own capability |
| Robot as a service | A single task done by the job | Built Robotics, Crewline | Production in narrow scopes | Autonomy as a subcontract |
| Autonomous contractor | A delivered earthworks package | TerraFirma | Early, contract-backed | Autonomy as your competition |
| Mining autonomy | Fleet-scale driverless haulage | Caterpillar, EACON, Pronto | Mature, at industrial scale | The proof the model works |
| Teleoperation | Remote human control of machines | Teleo, Hive Autonomy | Deploying now | A half-step, human stays in the loop |
A lot, and fast. In roughly five weeks, TerraFirma raised about 115 million dollars led by Kleiner Perkins (Source), Travis Kalanick’s Atoms closed 1.7 billion dollars led by a16z (Source), and Gravis Robotics took 200 million dollars from SoftBank (Source). Worth noting that Atoms spans food and transport as well as mining, so it isn’t a pure construction number.
A retrofit kit fits to a machine you already own, turning your fleet autonomous while you keep the asset and its resale value. Robot as a service is the opposite arrangement: you hire the machine and the outcome by the job, with the vendor owning the kit. Retrofit makes autonomy a capability you build; service makes it a task you outsource. Our full breakdown of the workflows where this is already paying off is here (Source).
In narrow scopes, yes. Built Robotics offers autonomous pile driving for solar farms as a service, and Gravis has run a rental-style route through UK plant hire. Broad, off-the-shelf availability isn’t here yet, but job-by-job access to specific tasks is real today. For the mining and quarry side of this, see our piece on Kalanick’s Pronto deal (Source).
Both, depending where you look. Mining autonomy is proven at industrial scale, with Caterpillar reporting nearly 700 driverless trucks in service. Construction autonomy is earlier and rolling out one machine class at a time, such as Bedrock’s excavators working with no operator in the cab on live sites (Source). Independently verified data on the construction side is still thin, so weigh vendor stats accordingly.
The retrofit pitch is deliberately brand-agnostic. Gravis says its Rack fits equipment from makers including Caterpillar, John Deere, JCB, Hitachi and Volvo, so contractors upgrade what they already own rather than replacing the fleet (Source). That mix-and-match approach is a big part of why the retrofit model is spreading faster than purpose-built autonomous machines.
TerraFirma is an Austin company founded by former SpaceX engineers that raised about 115 million dollars in mid-July, with a 100 million dollar Series A led by Kleiner Perkins (Source). Unlike the retrofit firms, it isn’t selling you software, it’s a vertically integrated construction company that wins the earthworks contract and runs it with its own remotely supervised machines. If you move dirt for a living, that model is a potential competitor, not a supplier.
Two questions cut through most of the noise. First, which of the five models are you, retrofit kit, service, autonomous contractor, mining-style fleet, or teleoperation? That tells you whether you’re buying a capability, a subcontract or a future rival. Second, which of your numbers has been independently verified? Much of the data in this space is self-reported, so the honest answer to that question reveals more than any demo. Gravis, for one, is upfront that its productivity figures are its own (Source).
Related Articles
Bedrock Robotics Says Its Excavators Are Now Digging With Nobody in the Cab
Uber’s Founder Just Bought an Autonomous Mining Company. Here’s Why Construction Should Pay Attention
Construction Robotics in 2026: 4 Workflows Where Robots Are Actually Working
Business Wire, TerraFirma Raises $115M
TechCrunch, Kalanick’s Atoms Raises $1.7B Led by a16z
Construction Dive, Gravis Robotics Raises $200M
Engineering News-Record, Gravis $200M Series A
The Robot Report, TerraFirma’s Robotic Construction Stack
The AI Insider, Gravis Deployed Across Four Continents
