Mach9's story follows a pattern that repeats itself in AI startups, but with a twist that reflects something deeper about the current state of engineering in the United States: the hardware works well enough, what was blocking progress was the inability to transform raw data into actionable information.
Founded in 2021 by four Carnegie Mellon University researchers under the name Mach9 Robotics, the company launched with an ambitious mission: build a vehicle-mounted scanning system for civil infrastructure visual inspections that was better, faster, and cheaper than what was available. The hardware was indeed built. And it was precisely that hardware that led CEO and co-founder Alex Baikovitz to a surprising conclusion: innovation wasn't in data capture, but in data interpretation.
"The hardware wasn't the bottleneck", Baikovitz told TechCrunch. "Engineering teams already had the scanners. What they didn't have were efficient ways to process what those scanners produced."
The CAD bottleneck
American civil engineering operates on a scale that technologists might find alarming. There are 4.2 million miles of public roads in the United States, not counting electrical transmission lines, fiber optic networks, railroad tracks, and the entire underground infrastructure that supports modern life. Most of that infrastructure was built in a single generation, designed 70 years ago to last 50 years.
Here's the data most articles miss: between 2011 and 2025, the number of licensed surveyors in the U.S. fell by 24%. Fewer people working to inspect infrastructure that keeps growing. It's what economists call structural supply tension — and it's not something solved with more hardware investment.
The standard technology for capturing this reality is LiDAR (light detection and ranging), which fires millions of laser pulses per second and produces three-dimensional point clouds with centimeter-level accuracy. The problem is that a point cloud, no matter how accurate, is not a drawing. To transform that cloud into an actionable engineering plan — with curbs, signs, poles, lane markings, and wires drawn in CAD — engineers still needed to manually click on every vertex, line, and curve. A process that, according to Baikovitz, took about a week per mile of road in traditional CAD software from companies like Autodesk and Bentley Systems.
The software pivot
The decision to walk away from hardware in 2024 and focus exclusively on software wasn't romantic. It was pragmatic. Mach9 discovered, through direct customer feedback, that engineering teams didn't want more hardware. They wanted software that worked with the hardware they already had.
The resulting product, Digital Surveyor 2.0, launched in March 2026 as a web-based platform, uses AI models trained internally to automatically extract features from LiDAR point clouds. Instead of an engineer clicking on every object that matters in the scan, the model identifies and draws curbs, guardrails, wires, poles, and lane markings with centimeter accuracy — the level of precision that the construction industry demands, Baikovitz said.
"It's not an off-the-shelf model that delivers this", the CEO emphasized. "We've spent since 2024 developing industry-leading models that transform imagery and multimodal scan data into survey-grade geometry."
The numbers are impressive: the company processed over 100,000 miles of mapped infrastructure, distributed across more than 2,500 projects in all 50 U.S. states. The customer list includes engineering firms Langan, Olsson, HNTB, and WSP, along with several state Departments of Transportation.
Why now? The right investment at the right time
The $34 million Series A round, co-led by First Round Capital and Bessemer Venture Partners, with participation from Prudence Capital and continued participation from Quiet Capital, brings Mach9's total funding to $49 million. Prior to this round, the company already had around 100 customers in private beta and had applied its software to 2,500 projects.
For investors, the thesis is clear. "There are 4.2 million miles of public roads in the United States", said Bill Trenchard, partner at First Round Capital, "before you count power lines, fiber networks, and other infrastructure". In his view, "Mach9 is building the software layer that will transform our physical world into information engineers can understand and act on. It will become a fundamental tool for how the country's infrastructure is built and maintained."
Caty Rea, Vice President at Bessemer Venture Partners, was even more specific: "AI is creating the opportunity for a new generation of CAD software. Just as software transformed construction, AI-powered CAD will transform how infrastructure is surveyed, designed, built, and operated."
The question is whether Mach9 will be ahead of a tsunami of demand or whether the transformation it predicts is slower than investors hope. Civil engineering is a historically slow sector for adopting new technologies — but demographic pressure and the push for infrastructure reconstruction are creating a window that may not open again.
Sources: TechCrunch, CryptoBriefing, Mach9 Official
✓ Independent sources cross-checked and verified before publishing
