Mach9 Raises $34M Series A After Dumping Robots for AI CAD

Mach9 has closed a $34 million Series A co-led by First Round Capital and Bessemer Venture Partners, capitalizing on a pivot from inspection robots to AI-powered CAD software that's already racked up nearly 100 private-beta customers and more than 2,500 projects.

The round, announced Wednesday, also included New York-based Prudence — an early-stage firm focused on AI software for construction and real estate — and Quiet Capital, which was already on the cap table.

Alex Baikovitz co-founded Mach9 Robotics in 2021 to build faster, cheaper hardware for civil engineers doing visual inspections. The company built it. Then Baikovitz realized hardware wasn't the problem.

"We saw how tedious it was for civil engineering organizations to take data from these amazing laser scanners and image data capture systems and turn that into a usable survey," Baikovitz told TechCrunch, referring to the detailed site maps engineers rely on for design work. "The real pain was in their workflows," he said, and in turning "endless amounts of data" into "information that people can use to make the best design and asset management decisions across the board."

So he dropped "Robotics" from the name and rebuilt the company around AI CAD software.

Why CAD, and why now?

Mach9 is one of several startups attacking the planning-to-building pipeline. In August, CivilGrid raised $26 million for what it calls the "Google Maps for the underground." Blueprints AI, which generates code-compliant, permit-ready architectural blueprints in minutes, made the top 200 at last week's TechCrunch Disrupt Startup Battlefield.

Baikovitz picked CAD because it's the pipeline's slowest, most labor-intensive segment.

"It takes months to be able to deliver a project, and about a week per mile for someone to do manually in the legacy incumbent software that's built by the Autodesk and Bentley Systems of the world," he said.

Think about that pace: at a week per mile, a 20-mile highway corridor eats five months of drafting time before design work even starts. Mach9 claims its software automates enough of that work to run 10 times faster and cut costs dramatically, letting surveying and civil engineering firms take on bigger projects and expand "the reach of how reality-capture data can influence major infrastructure decisions," Baikovitz said.

What does this change for practitioners?

The software doesn't try to replace licensed engineers. It takes survey data and produces a first draft, which professionals then refine. Mach9 gives them "all of the tools to draw, remove, adjust vertices, assess the quality of the AI-generated model, so that they can put effectively their name behind the deliverable," Baikovitz said — a deliberate nod to the liability that hangs over the industry, where licensed professionals must sign off on every deliverable.

Accuracy is the moat. Baikovitz said Mach9 has trained its own in-house AI models since 2024 to hit centimeter-level precision, which construction and infrastructure work demands.

"That's not something that a lot of off-the-shelf models are able to do or deliver at in any capacity," he said. The development focus, he explained, has been building "industry leading models that can turn imagery or these multimodal image and laser scan data sets into survey grade like geometry, like tracing out the curbs, the pavement striping, all of this exactly how our customers want it to be at the accuracies that the industry demands."

The customers who killed the hardware business

Baikovitz credits the hardware detour for the software success. It gave him something concrete to pitch, and customers told him plainly what they didn't want.

"I'm not interested in getting this cheaper, lower-cost hardware," one VP told him. But better software could "improve our efficiency in delivering these big statewide transportation mapping initiatives," the executive said — projects Baikovitz said make states Mach9 has worked with, like Texas, "much more like safe [and] effective."

The $34 million will be split three ways: improving the product, hiring industry experts, and scaling "all parts of our organization to keep up with where the market is taking us," Baikovitz said.

With Autodesk and Bentley Systems still owning the legacy workflows, Mach9's bet is that speed — 10x, at the company's own claim — plus engineer-in-the-loop review is the wedge that finally modernizes infrastructure design.