McCarthy Building Cos. just told the construction software market something blunt: it would rather build than rent. The St. Louis contractor has signed a multiyear, multimillion-dollar agreement with Palantir to power its internal construction AI, and the structure of the deal matters as much as the dollars.
McCarthy isn’t licensing a finished product. It’s using Palantir’s Artificial Intelligence Platform to develop what the firm calls a connected AI operating system, with Palantir engineers embedded alongside McCarthy’s own applications team. The goal is to wire together estimating, contracts and field execution through a single data model.
What Pulse actually does
The center of the partnership is Pulse, McCarthy’s AI-native system for the field. It feeds superintendents, project managers and field operators real-time insight, scenario planning and risk analysis so they can weigh options on the jobsite instead of waiting for an answer from the office. Pulse runs on Palantir’s Ontology, the same segment Palantir describes as a “tool factory” for building and deploying applications.
Palantir is an unusual partner for a builder. The firm made its name in defense and intelligence through its Gotham platform, then expanded into retail, food manufacturing and finance. Construction is newer ground. McCarthy’s chief digital officer, Justin McFarland, credited Palantir’s engineering talent with turning “complex operational concepts into scalable solutions faster than we thought possible.”
The build-versus-buy line keeps moving
Here’s the part the industry should sit with. As contractors get more tech-literate, several have decided they understand their own workflows better than any outside vendor does. That logic pushes them past buying off-the-shelf software and toward building their own, often with a later-stage partner who can tailor the work.
Turner Construction, the largest U.S. contractor by revenue, has questioned the economics of paying a startup for a solution it could build itself, and last fall it signed a companywide deal with OpenAI. Suffolk, meanwhile, is now embedding AI engineers directly on project teams. McCarthy’s Palantir tie-up sits squarely in that pattern.
The risk is obvious. Building enterprise software in-house is expensive, hard to maintain and easy to get wrong, and construction is littered with proprietary tools that never outlived the executive who championed them. A general contractor is not a software company, however many engineers it embeds. The bet McCarthy is making is that owning the data model, and the decisions that flow from it, is worth more than the convenience of a subscription.
For the startups selling construction AI, the message is uncomfortable: their biggest potential customers are starting to look like competitors. Exchange has covered the money chasing this space, from AI preconstruction software funding to Autodesk’s $3.6B MaintainX buy. The McCarthy deal was reported by Construction Dive.