Everything comes out of the ground. Our world is made of minerals, concrete, ground shaped for infrastructure: homes, roads, datacenters, supply chains, utilities, defense… Everything from your breakfast today to the device you are reading this on to the copper in your GPU requires a key primitive: molecules of earth have to move for any of those objects to exist. AIM is the physical AI platform that moves those molecules autonomously, at scale, reliably, in brutal conditions, every day. We are on an exponential scale ramp across all levels of the business and are hiring across disciplines: from embodied AI to firmware to COO to hardware eng to account management to forward deploy… As fullstack as it gets! If you find terraforming Earth with physical AI deployed in production more meaningful than building robots doing backflips, folding towels, or even just driving cars – let’s talk. Remember even all those robots had to come out of the ground and many of their molecules were made possible because of the machines we have automated. Built by operators across mining, construction, Google, SpaceX, Waymo, Navy SEALs, Amazon, Tesla, AIM enables scalable earthmoving, turbocharging the global economy’s physical foundation. We are backed by some of the most sophisticated capital in the world, including General Catalyst, Khosla Ventures, Elad Gil, HUMAN CAPITAL, Ironspring Ventures, Mantis Venture Capital, MVP Ventures, Qasar Younis, DCVC.
The side that can autonomously throw up revetments, blast walls, hull‑down positions, and covered routes in hours instead of days controls how and where the fight happens. AIM’s focus on physical AI for earthmoving is about giving commanders that kind of ground‑truth leverage — turning terrain into a dynamic asset instead of a constraint.
The operator list (mining, construction, Google, SpaceX, Waymo, Navy SEALs, Amazon, Tesla) is strong but missing one category: people who actually run the mines AIM serves. KZ, AU, Chile, Peru. Shop-floor extraction experience is rarer than ex-FAANG, and it's the bottleneck on the deployment side. Plans to mix that in?
The bottleneck on physical AI at this scale isn’t perception or planning — it’s the maintenance feedback loop. Heavy machinery in the field loses ~30% of uptime to small mechanical issues humans catch by intuition long before sensors do. The first team to embed reliable wear-detection inside the autonomy stack quietly wins the next five years. Are you working with traditional OEM partners on the hardware loop, or building it all in-house?
"Everything comes out of the ground" : that's the kind of first principles reminder that reframes an entire industry in one sentence. The pick-and-shovel logic applied to physical AI is compelling precisely because it's unglamorous. Everyone's building the thing that needs the minerals. AIM is automating how those minerals get extracted. "More meaningful than robots doing backflips" is a mission statement I didn't expect to find in an earthmoving company. But it landed.
Autonomous earthmoving is the unseen foundation of our entire infrastructure. The fact that AIM is already deployed in production, turbocharging earthmoving at scale, raises the bar for what's possible in construction, mining, and beyond. Moving molecules of earth is the ultimate physical AI challenge – and it looks like AIM is tackling it head-on.
Hot damn, as logan says on succession - aim is for serious people tackling serious issues
One of the most overlooked realities in AI is that the physical economy still dominates the digital one in terms of complexity and capital intensity. Moving atoms at scale has always been harder than moving bits. Applying autonomy to earthmoving and industrial infrastructure could ultimately have a larger economic impact than many consumer-facing AI applications.
I had a conversation with one of the world's biggest mine owners a few weeks ago. I told him, "Everything AI-related works because of your job" I can't agree enough on how this is one of the best use cases of Physical AI Can't wait to see what you build! This is SO cool!
The transition from digital algorithms to physical world applications is the most significant challenge of our generation. Building technology that survives brutal real world environments requires an entirely different level of engineering rigor compared to pure software development. What is the most unexpected hardware failure your team has encountered while deploying these autonomous systems in the field? Adam
Nice work Adam!