Self-driving cars were the easy part. Now the bulldozers are coming.

Self-driving cars learned from lane markings. Bulldozers don't have that luxury. Every pass of a blade erases the path the machine just mapped, so the autonomy stack has to rebuild its world in real time, without painted lines, without a fixed curb, and without a human standing behind the controls.

That harder problem is what a new wave of rocket-trained engineers is racing to solve, and the category is starting to look crowded. Seattle-based AIM Intelligent Machines, founded in 2021, occupies the former SpaceX building in the city's Fremont neighborhood and has staffed up with former SpaceX engineers, according to GeekWire's archived July 25, 2026 report. The company says it is building what it calls the world's first AI platform dedicated to heavy machinery, covering construction and mining sites, and it announced $50 million in funding to scale the work. The $50 million figure is self-reported in AIM's own press release and has not been independently verified against filings.

AIM is not alone. Austin-based TerraFirma, founded by two more former SpaceX engineers, raised $115 million earlier this month to semi-automate physical construction, the same GeekWire report notes, citing the company's announcement. Two well-funded teams, both with rocket-engineering DNA, hitting the same wall in the same month is what makes the sector move visible. The combined roughly $165 million in July 2026 commitments is small by consumer-AV standards, but it is large enough to fund real machine fleets, real job-site pilots, and real runway tests.

Self-driving cars run on a fixed map: lane lines, stop signs, and curbs the vehicle cannot move. A bulldozer, excavator, or earthmover runs on terrain it is actively reshaping. The autonomy stack has to plan a path, execute the cut, and replan against a landscape that no longer matches the one it measured a minute ago. There are no lane markings to anchor against, no high-definition maps that stay valid, and no human standing behind the controls to take over when the model hesitates. That is closer to a planetary rover problem than a Tesla problem, which is one reason the talent migrating into the category is coming from rockets rather than from consumer AV. Rocket engineers are used to unstructured environments and high consequence-of-failure; consumer-AV engineers optimized for a much friendlier world.

AIM's public work already spans that range. The company holds a United States Air Force contract to develop autonomy for heavy machinery used in airfield runway maintenance and related duties, per GeekWire. No contract number, scope, or Air Force subcommand has been disclosed in the reporting so far, which makes the contract a real-but-unverified line item. The company's longer-horizon roadmap is more ambitious: founder and CEO Adam Sadilek has framed the work as preparation for Mars excavation and exploration missions where human involvement could be compromised. Sadilek spent nine years at Google working on autonomous machines and equipment before starting AIM, he told Unite.ai, and the company's stated vision is "Earth first, Mars later." That nine-year tenure is the founder arc; the SpaceX bench is the engineering bench, and the two are not the same.

That framing is aspirational, not contracted. Mars is the founder's long-term goal, not a customer line item, and the source basis supports treating it as commentary rather than as a near-term capability. The runway work is the nearer test. If AIM and TerraFirma can teach heavy equipment to operate reliably on an airfield, where GPS-denied corners and bad weather are routine, the same autonomy stack can plausibly move into private construction sites and, eventually, into open-pit mining operations where the safety case for autonomy is straightforward. The harder-than-roads point is what the wire misses: cars learned from lane markings, and the next autonomy category is the machines that erase their own. AIM and TerraFirma are both hiring toward that problem, and July's two raises are the first sign the market agrees.