Katerra raised more than $2 billion, reached a $4 billion valuation, and set out to vertically integrate construction from design through manufacturing to general contracting, with robotic and CNC-driven factories at the center of the plan. It collapsed in 2021. Blueprint Robotics, a far smaller operation running CNC-driven wall, floor, and roof panel production out of a single 200,000-square-foot factory in Baltimore, is still operating, reporting productivity gains of roughly 2.5 times a conventional framing crew. Same underlying technology. Wildly different outcomes.

Carpenters are rated Moderate risk, with a horizon toward 2045. The risk isn't that a robot learns to frame a house on a job site. It's that more and more of what a carpenter used to build on-site is now cut, assembled, and finished in a factory before it ever reaches the site at all. Katerra's failure and Blueprint's survival tell you something specific about where that shift actually works.

The other force acting on this profession has nothing to do with robots. The residential construction industry needed an estimated 2.17 million additional workers between 2024 and 2026, and the skilled trades workforce has shrunk by roughly 17% since 2006 even as construction activity grew. Whatever automation is doing to this job, it is happening inside an industry that cannot currently find enough people to do the work automation hasn't touched yet.

Key Points

  • Carpenters are rated Moderate risk with a horizon toward 2045, as CNC-driven prefabrication shifts an increasing share of the work from job sites into factories rather than eliminating it outright.
  • Katerra raised more than $2 billion attempting to vertically integrate construction around robotic manufacturing and collapsed in 2021, while smaller, more focused operations like Blueprint Robotics report roughly 2.5 times conventional productivity using similar CNC technology.
  • AI-powered takeoff and bidding tools like Togal.AI report cutting estimating time by more than 75%, but function as tools for estimators and contractors rather than replacements for carpenters.
  • Robotic systems like the SAM100 bricklayer and Canvas drywall finisher handle standardized, high-volume tasks well, but the construction industry's own research consistently finds irregular sites, renovations, and custom work still require a human carpenter's judgment.
  • The residential construction industry needed an estimated 2.17 million additional workers between 2024 and 2026, with the skilled trades workforce down roughly 17% since 2006, a shortage the U.S. Bureau of Labor Statistics' projected 4% growth in carpentry jobs through 2035 won't come close to filling.

What a Carpenter Actually Does

The trade covers framing structural walls and roofs, installing cabinets, doors, and trim, and finishing work that turns a structure into a livable space, plus the judgment to adapt all of it to a real building that rarely matches its drawings exactly, work a building inspector later has to verify against code. Historically, that work happened almost entirely on-site, cut to measure and assembled in place. Prefabrication changes where the cutting and initial assembly happen. It does not, on the evidence so far, change who does the final fitting.

The Factory Floor Is Where the Automation Lives

Blueprint Robotics' factory uses robotic saws, nailers, and framing equipment to produce precision wall, floor, and roof panels, achieving productivity gains its own reporting puts at roughly 2.5 times a standard framing crew, with software from Strucsoft managing the digital-to-physical handoff. Mighty Buildings pursued a more aggressive version of the same idea with large-format 3D printing of building components, reaching UL safety certification and hurricane-resistance ratings, before entering liquidation in 2025, a reminder that even a technically working approach can still fail commercially. In masonry, Construction Robotics' SAM100 lays 2,000 to 3,000 bricks a day against a human mason's 500, though it remains explicitly assistive: a mason still sets up, calibrates, and quality-checks every run.

THE KATERRA LESSON

Katerra's collapse wasn't a story about CNC robots failing to cut lumber accurately. It was a story about a company trying to own design, manufacturing, and general contracting simultaneously, scaling a robotic factory model faster than its business fundamentals could support, and running into cost overruns and accounting problems that had nothing to do with whether the saws worked. Blueprint Robotics does one thing, panelized framing components, well, at a scale it can actually manage. The technology that sank a $4 billion company is quietly profitable at a fraction of the ambition.

Why On-Site Work Resists

Canvas, a robotic drywall finisher, handles taping, mudding, and sanding to a consistent Level 4 or 5 finish and cuts a multi-day process to about two, but it operates in the relatively controlled geometry of a finished wall cavity. MIT's AutoSaw research project demonstrated a workable division of labor rather than a replacement: a robot performs the precision cut, a person handles setup, calibration, and the final fastening and inspection a real, imperfectly built structure demands. Industry surveys back this up directly, roughly two-thirds of builders report shortages specifically in finish carpentry, the renovation, custom-fit, and trim work that resists standardization by definition.

The Bidding Layer

Away from the job site, AI has moved fastest in estimating. Togal.AI applies computer vision to architectural drawings with reported accuracy near 98%, and the company's own data, along with independent university testing, puts the time savings on a takeoff at 75% or more. That compresses a real cost center for contractors and general builders. It is a tool an estimator or project manager uses, not a substitute for the carpenter translating a drawing into a physical structure.

The Labor Market Reality

The U.S. Bureau of Labor Statistics projects 4% growth for carpenters through 2035, in line with the average occupation, with about 62,800 annual openings and a median wage of $60,580. Set against that, the Home Builders Institute estimates the industry needed roughly 2.17 million additional workers between 2024 and 2026 alone, and the skilled trades workforce has fallen from about 8.5 million in 2006 to roughly 7.1 million in 2024 even as construction demand grew. Modular and prefabricated construction, meanwhile, has grown to around $20.5 billion in the U.S. market, expanding at close to 6.5% a year. Put together, this reads less like automation replacing carpenters and more like automation trying, and mostly failing, to close a gap that keeps widening on its own.

How to Use AI as a Carpenter Now

For standardized components: panelized and CNC-cut framing is already cheaper and more consistent than cutting to measure on-site for repeatable structures. Learning to work with factory-produced components, rather than against them, is where the volume of new construction is heading.

For finish and custom work: this remains the least automatable, most in-demand segment of the trade, and the shortage data says it will stay that way for years. Specializing here is a defensible long-term position.

For bidding and takeoffs: tools like Togal.AI are worth adopting on the estimating side of a business, whether you use them directly or work with contractors who do. The time they save is real and doesn't come at a carpenter's expense.

What I Think

The 2045 horizon feels right, and the contrast between Katerra and Blueprint Robotics is why. The technology to automate a meaningful share of carpentry, in a factory, at scale, already exists and already works. What has consistently failed is the ambition to replace the trade wholesale rather than augment a specific, well-bounded piece of it. That is a pattern this series keeps finding: automation that tries to do everything collapses, and automation that does one thing well survives quietly.

What I'd actually worry about for this profession isn't the robot. A trade already short millions of workers, with a workforce older and smaller than it was two decades ago, doesn't have much room to lose people to automation before it loses the capacity to build anything at all. If prefabrication genuinely draws young workers into factory-based carpentry roles that are safer and more predictable than a job site, it might do more to solve this trade's real problem than any amount of on-site robotics ever will.

"The $4 billion company tried to automate the whole trade and collapsed. The factory doing one thing well, quietly, is still open. That's the actual lesson here, not the robot."