Caterpillar plans to spend $100M over the next five years training its workforce in AI, autonomy and robotics, chief technology officer Jaime Mineart told TechCrunch in an interview published on 30 August. TechCrunch puts the relevant headcount at 118,000 employees. The money sits under the company's Building the Future Workforce Initiative, a five-year programme launched to mark its centenary, which has so far rolled out state by state in the US with allocations in Indiana, Texas, Illinois and Arkansas.
Why should anyone outside Peoria care? Simple because Caterpillar is, by its own description, the world's leading manufacturer of construction and mining equipment, with 2025 sales of $67.6bn. When the company that builds most of the world's yellow iron defines the operator of the future, that decision affects every market it serves. The US programme is a case study, not the whole story. And the decision it has made is instructive: the operator does not go away. The job changes, and Caterpillar is paying to change it.
That is a pointed contrast with the challengers we covered last week. Bedrock Robotics, founded by Waymo veterans and launched with $80M from Eclipse and 8VC, takes a different approach. It retrofits existing excavators with sensors and computing power so they can, in its words, work around the clock. The company has been testing with contractors, including Sundt and Zachry across four US states. Two firms, two theories of the same jobsite.
What Caterpillar actually announced
The claims are worth separating, because they come from two places. The AI framing comes from Mineart's fireside chat at the Ai4 conference in Las Vegas, as reported by TechCrunch: $100M over five years to train the workforce in AI, autonomy and robotics.
Caterpillar's own press releases describe the same initiative in broader terms – training for 'manufacturing and industry technician jobs of the future', outreach to students, parents and educators, and partnerships with community colleges. The Illinois tranche alone is up to $10M, in a state where Caterpillar employs more than 17,000 people.
Read together, this is not purely an internal AI retraining scheme. A good chunk of it is a talent-pipeline investment aimed at communities around Caterpillar's plants, of which AI and autonomy skills are one strand. It arguably makes this more interesting than the initial impression. The company is funding the pipeline outside its own walls, on the logic that it cannot hire people who were never trained.
Thirty years of practice runs in mining
Caterpillar's confidence comes from having done this before, in the least forgiving environment available. Its autonomous haul trucks, drills and dozers have been working remote mining sites for years, alongside a software command centre and fleet management tools. Mineart told TechCrunch the company now has about 1.6 million connected assets globally and more than 16 petabytes of structured data flowing from them.
Now we're in this super exciting time where we can take all of that learning from mining and bring it into much more dynamic environments, jobsites, quarries, and construction sites.
For readers running projects, the operative word is 'dynamic' – a mine is a controlled loop, a construction site is not, and Caterpillar knows the second problem is harder than the first.
The consumer-facing end of this is the Cat AI Assistant, first piloted in a Cat 306 CR mini excavator and built on NVIDIA's Jetson Thor platform, unveiled at CES in January. It lets a technician standing next to a machine use voice commands to pull up repair procedures, troubleshoot faults and identify parts before starting a job. Mineart says it is now in use by customers, operators and technicians – a vendor claim, though a checkable one as it spreads through dealer networks.
Brandon Hootman, Caterpillar's vice president of data and AI, gave TechCrunch the clearest one-line justification for putting the assistant in the cab rather than the office: "Our customers don't live in front of a laptop day in and day out; they live in the dirt."
This design principle is worth stealing for any AEC firm deploying AI: if the tool requires a desk, most of the industry cannot use it.
The hard part is not the machine
The most useful thing in the interview is Mineart's account of where autonomy projects actually stall. It is not perception, or control software, or hardware reliability.
The hard part about autonomy and about physical AI is incorporating that technology into the customer jobsite and into the workflows.
Anyone who has tried to land a digital tool on a live project will recognise the shape of that problem: the technology is the cheap part, and the method statements, supervision models and commercial arrangements around it are the expensive part.
Mineart says Caterpillar leans on experienced operators to help train its AI systems, and expects some operators to shift from controlling one machine to overseeing several from a remote command centre. The operator becomes a supervisor of machines; the seat time becomes screen time.
There is a commercial tailwind behind all this, too. Caterpillar's second-quarter revenue hit an all-time high of $20.5bn, helped by demand for power-generation kit for data centres; power-generation sales rose 72% to $3.10bn, and CEO Joe Creed said 'no one is slowing down' on AI infrastructure demand. Caterpillar is being paid by the AI boom on one side of the business while spending to absorb it on the other.
What it means for the AEC talent pipeline
Set Caterpillar's move next to Bedrock's and you get a fair summary of the argument now running through the industry. Bedrock's retrofit kits imply fewer seats per site and a smaller operator workforce. Caterpillar's training pledge implies the same number of people doing different, more supervisory work. Both may be right on different sites and timescales, but only one of them comes with a funded plan for the people involved.
The uncomfortable backdrop is that the workforce problem exists either way. Construction Digital, reporting on the pledge, cites US Bureau of Labor Statistics figures of roughly 298,000 vacant construction jobs in May 2026, and Deloitte research finding more than 80% of US contractors struggle to fill skilled craft positions. Autonomy is arriving into a labour shortage, not a surplus, which is precisely why the incumbent frames it as augmentation.
The same question lands on professional services. If the machine operator's route is 'fewer levers, more oversight', the junior QS, engineer or planner faces a parallel shift as agents take on the production work that used to train them. Caterpillar's answer is to put real money into structured training, community pipelines and redeployment before the technology fully lands.
Takeaway
Caterpillar has effectively priced the workforce transition: about $20M a year, for five years, across a business of 118,000 people and the communities that feed it. Most contractors and consultancies will never spend at that scale. The ratio is the point: the training budget was announced alongside the technology, not five years later. If your organisation has an AI roadmap and no retraining line item, you have half a plan.
Ask your plant and equipment suppliers what operator training comes with their autonomy and AI features, and get it into procurement conversations now.
Map which roles on your sites and in your teams shift from doing to supervising, and write the training plan for that shift before the tools arrive.
Protect the junior pipeline deliberately: if AI absorbs the tasks that used to build competence, replace them with structured exposure rather than hoping it sorts itself out.
Watch whether the retrofit model (Bedrock) or the incumbent model (Caterpillar) wins on your regional sites – it will tell you which workforce future you are planning for.
If you want to keep track of who is automating the jobsite and what it means for the people on it, that is exactly the beat we cover every week. Subscribe free at projectflux.ai and get the plant, autonomy and workforce stories before they reach your standup.
Links and Stuff
All content reflects our personal views and is not intended as professional advice or to represent any organisation.
