Why the work stays at the bench
Cabinetmaking is a loop of measuring, cutting, fitting and correcting. A drawing says a door should be 23 and 7/8 inches wide. The opening in the kitchen says something else. Someone has to verify dimensions against the real space, shave the stile, and hang the door so it closes square. That correction step is where most of the hours go, and it is physical.
The second reason is material. Wood moves. Grain tears out on one board and not the next, a panel cups after a humid week, a veneer lifts under the sander. A cabinetmaker feels that through the tool and changes feed, grit or clamping pressure before the part is ruined. Software can plan a cut; it cannot notice that this piece of maple is behaving badly today.
Software has still taken a real bite out of the office side of the trade. Drawings, cut lists, nesting layouts and material estimates all move faster with modern design tools. That is task erosion, not a job disappearing. Our can-AI-do-it score sits at 11 out of 100 (higher means more of the task time is already within reach), and you can read how that figure is built on the coverage method page.
What software does, what it assists, what people keep
Start with what AI can carry on its own. Mostly it is paperwork and planning: turning a specification into a cut list, estimating how much sheet goods a run of cabinets needs, and laying out parts on a panel to waste less material. That group holds 0% of task time on this job.
The assisted group is larger in practice. Reading and interpreting blueprints, programming a CNC router, checking parts against the drawing and tracking a job through the shop all go faster with software in the loop, but a person signs off on each one. Those shared tasks account for 20% of the work.
What is left belongs to people. Setting up and running saws, routers, shapers and sanders. Assembling and gluing parts, clamping, squaring the box. Installing hinges, drawer slides and pulls so they run true. Sanding, filling and finishing surfaces by feel. Repairing and refitting existing cabinetry on site, where nothing is plumb. Together that is 80% of task time, which is why the headline score lands at 82 out of 100 (higher is safer).
What the evidence actually shows
No study has yet tested an AI system against a working cabinetmaker on cabinetmaking tasks. Our evidence grade for quality parity is D, and a D means not measured, so we publish no parity number for this job. That is an honest gap, not a verdict.
What would settle it is specific: a timed, side-by-side trial on a full build, covering setup, cut accuracy, assembly tolerance, hardware fit and finish quality, scored by someone who inspects the result rather than the process. Until something like that exists, the fair reading is that software handles the planning layer while people handle the making layer. You can see how we grade evidence on the quality parity page.
The market picture comes from official data. The Bureau of Labor Statistics counts about 77,170 cabinetmakers and bench carpenters in the United States, with median pay of $46,680 a year (BLS, 2025), and projects employment to change by about -2.9% between 2025 and 2035 (BLS, 2025). That slow drift owes more to imported ready-made cabinetry and shop consolidation than to anything a model does.
When the picture could shift
Most likely after 2046 (8 in 10 of our scenarios). The replacement-year method page explains what that window is measuring and how it is produced.
Two things could pull it earlier. Cheaper five-axis CNC cells would push more shaping, boring and joinery off the bench and onto the machine. And flat-pack engineering keeps improving, so more cabinets arrive as pre-cut, pre-drilled kits that need assembly rather than fabrication.
Two things hold it back. The physical share of this job is high enough that our robotics read puts it in fixed automation territory: bolted-down routers and panel saws, not machines that walk into a kitchen and fit a face frame. And the economics of custom work are stubborn. A machine that pays for itself across thousands of identical doors earns nothing in a shop where the next job is a one-off island in reclaimed oak. Install and repair work, done in finished homes, resists the same way.
How to stay needed in this trade
Lean into the parts of the job that are hardest to specify from a drawing. First, on-site fitting and installation: scribing to crooked walls, hanging doors, aligning drawer fronts. Second, finishing: stain matching, grain filling, spray technique and repair of blemishes. Third, custom joinery and one-off problem solving, where the client changes the brief halfway through.
Two skills pay off alongside that. Learn CNC setup and programming, so you run the machine rather than compete with it. And get comfortable with design software and quoting tools, because shops that turn a site visit into an accurate quote the same day keep the work.
What to do: ask your shop which steps already run off a nested cut file, and make sure you are one of the people who can edit it.
Nearby jobs are worth comparing if you are weighing a move. Furniture Finishers share the finishing skills, Model Makers, Wood sit closest on precision one-off work, and Woodworking Machine Setters, Operators, and Tenders show where the machine-side skills lead. You can put any two side by side on the compare tool, browse the rest of the woodworkers family, see how the wider manufacturing sector scores, or scan the jobs that mostly need a person. Our full method is on the methodology page.