How do you start a CNC machining shop business in 2027?

The garage-to-shop moment most founders get wrong

Picture a machinist with eighteen years on aerospace floors who finally puts $90,000 down on a used Haas VF-3, signs a three-year lease on 2,800 square feet in an industrial park, and prints business cards that say "precision CNC machining — any material, any quantity." Within four weeks the phone rings. Within four months the machine is running most days. Within eighteen months the shop is doing roughly $290,000 a year, the owner is working seventy-hour weeks, and after debt service, rent, insurance, tooling consumption, and material, the owner's take-home is somewhere near what a senior machinist earns as an employee — with none of the benefits and all of the risk.

That is the modal outcome, and nothing about it involves bad machining. The parts are good. The problem is that "any material, any quantity" is a business model with no pricing power. Every quote lands on a buyer's desk next to quotes from shops with fully depreciated iron, shops in lower-cost regions, and increasingly an instant quote from a CAD-to-quote platform that priced the same STEP file in eleven seconds. The generalist becomes the residual bidder — the shop that wins when it is cheapest and loses when it is not.

The structural context matters here, because it cuts both ways. The US machine-shop industry (NAICS 332710) is genuinely large — tens of billions in annual revenue — and deeply fragmented across roughly 180,000–220,000 establishments, the great majority of them under twenty employees. No incumbent can crush a new entrant. But that same fragmentation exists precisely because the work is hard to differentiate and hard to scale. Meanwhile the tailwinds are real: reshoring activity tracked by the Reshoring Initiative has run at historically high levels since 2022, defense and aerospace budgets have expanded, and the Manufacturing Institute and Deloitte have repeatedly projected large-scale unfilled manufacturing roles through 2030, with machinists squarely in the gap. Capacity is scarce. Skilled labor is scarcer.

So the founding question is not "can I get work" — a competent shop with an open spindle can almost always find work in 2027. The question is whether that work carries margin. The escape is counterintuitive for a craftsperson: you must say no to most of it. Pick a wedge — a material (titanium, Inconel, PEEK, Ultem), an industry (medical device, semiconductor equipment, optics, defense subassemblies), a part family (manifolds, housings, shafts, optical mounts), or a capability (5-axis simultaneous, micro-machining, large-envelope work) — and build the equipment list, fixture library, quoting templates, and marketing around it. Specialization in a market this fragmented is essentially free: a niche narrow enough that you're one of thirty credible US shops for it is still a multi-hundred-million-dollar addressable market. Generalists fight 200,000 competitors. Specialists fight thirty.

How the shop actually converts an RFQ into margin

The operating mechanism of a machine shop is a loop, and the profitability of the whole business is decided by where in that loop the work repeats. Understanding this is what separates the shop that clears 45% gross margin from the one grinding at 20%.

RFQ intake and triage. A part request arrives — from a platform, a direct customer, or an approved-vendor-list RFQ. Triage happens in minutes, not days: is this in the niche and within capability, and is this customer worth quoting? Rejecting fast is a profit activity. Generalist shops quote eight to fifteen jobs to win one to three, and quoting is unpaid engineering labor that can consume 30–40% of the owner's week.

Quoting. Review the model for manufacturability, estimate setup plus cycle plus material plus tooling plus inspection, apply lead-time tiering, and return the quote in hours for prototype work and a day or two for production. Quote speed is itself a competitive weapon — for a hardware startup burning cash, a shop that answers in four hours beats one that answers in four days even at a higher price.

Order entry and planning. PO received, job traveler created, material ordered, slot booked on the schedule. Capacity scheduling is where shops drown; a simple visual schedule used consistently beats sophisticated software used badly.

Programming and setup. CAM programming — reusing prior programs wherever the part family allows — fixturing pulled from the library, tooling kitted, first article cut and inspected before any production run. Always. Skipping first-article discipline to save an hour is how a shop scraps a lot of titanium.

Production, inspection, secondary operations. Run the lot with in-process inspection at defined points, deburr and finish, coordinate outside processing (anodize, plate, heat-treat, passivate) with vendors whose lead times you must schedule around.

Final QC, documentation, ship, invoice. Certificates of conformance, material certs, first-article inspection reports where required, packaged and shipped, invoiced the same day.

Log actuals and capture the reorder. Record job-cost actuals against the quote — that feedback loop is the only thing that makes quoting improve — then save the program, fixture, and setup sheet and tag the customer for proactive follow-up. The reorder is the single most profitable job the shop will ever run, because the setup, programming, and fixturing investment is already sunk.

That loop back from reorder capture to "program already exists" is the entire economic engine. A part with a three-hour setup and a four-minute cycle time, run once, loses money at almost any hourly rate. The identical part ordered at quantity 200, or reordered quarterly against saved fixturing, is highly profitable. This is why the niche decision is not a marketing preference — it is the mechanism by which setup amortizes.

The numbers: capital, rates, margins, and a five-year trajectory

Startup capital, itemized. A realistic owner-operator build with one VMC and one lathe in a leased 2,500–3,500 sq ft space:

• Used 3-axis VMC (Haas VF-2/VF-3 class): $35,000–$85,000. New, configured: roughly $70,000–$110,000. A well-maintained five-to-eight-year-old machine delivers most of a new machine's capability at half to sixty percent of the price.
• Used CNC lathe, ideally with live tooling: $30,000–$90,000. Live tooling plus a sub-spindle enables done-in-one turned parts, which is a direct margin multiplier.
• Tooling, holders, collets, vises, initial crib: $25,000–$50,000. This line is chronically under-budgeted.
• Metrology — surface plate, height gauge, quality hand tools, or a used CMM: $8,000–$45,000.
• Support equipment — bandsaw, air compressor sized for the shop, parts washer, deburr bench, pallet jack or forklift: $15,000–$35,000.
• Software year one — CAM, quoting, accounting: $3,000–$15,000.
• Facility deposit, three-phase electrical service and drops, buildout, signage: $15,000–$50,000.
• Insurance, first year: $4,000–$20,000.
• Initial raw material inventory: $5,000–$20,000.
• Entity formation, permits, attorney, CPA setup: $2,000–$8,000.
• Working capital and six-month runway: $40,000–$120,000 — the line everyone shaves to afford another machine, and the reason most failed shops fail.

Total realistic range: $120,000–$320,000, with most disciplined single-owner shops landing at $150,000–$250,000. A bare-bones single-machine start can be done for $60,000–$100,000, with materially worse survival odds.

Funding. Equipment loans and leases are readily available with the machine itself as collateral. SBA 7(a) works for general startup capital and SBA 504 is specifically structured for equipment and real estate — the SBA's own program pages spell out eligibility and structure. Machine tool builders and used-equipment dealers offer in-house financing. Avoid funding working capital on credit cards, and avoid giving away equity unless the partner brings customers or capability you genuinely cannot buy.

Rates and cost structure. A loaded shop-hour cost — machine depreciation, floor space, power, software, insurance, indirect labor, and a share of owner time — typically runs $45–$95 per hour for a small 3-axis shop and higher for 5-axis or specialized equipment. Billing rates that clear that with real margin: $75–$145/hr for 3-axis milling, $90–$160/hr for turning, $110–$210/hr for 4- and 5-axis, with premiums for exotic materials, tight tolerances, and certification-documented work. Material is marked up roughly 15–35%; tooling consumption is built into the rate.

Gross margin by segment. Commodity 3-axis and platform-routed work: 18–30%. Niche fast-turn prototype work: 40–60%. Certified regulated production, net of documentation overhead: 35–55%. Owner's discretionary earnings for a well-run owner-operator shop land around 12–22% of revenue; a scaled shop with managers runs 10–18%.

The trajectory. Year 1 is survival and proof: the owner runs machines fifty to sixty-five hours a week, quotes at night, takes platform work to fill gaps, and lands five to twelve direct customers. Revenue $180K–$420K, owner income often modest at $40K–$90K as cash recycles into tooling. Year 2 brings the first one or two machinist hires and the beginnings of a second shift: $400K–$850K. Year 3 is a real shop with three to five employees, possibly a quality certification in process, and blanket POs from anchor accounts: $650K–$1.4M. Year 4 forces the scale-or-plateau decision: $1M–$2.5M on the scaling path. Year 5 forks — $2M–$5M with eight to twenty employees if scaling, or $700K–$1.5M as a tight high-margin lifestyle shop.

The variance across that range is driven almost entirely by three decisions: whether you specialized, whether you priced on value or cost-plus, and whether you added the second shift, the automation, and the people before it felt comfortable.

Labor costs, since labor is the binding constraint. Regional variation is wide, but 2027-era ranges run roughly: entry operator $18–$26/hr, setup machinist $26–$40/hr, senior machinist/programmer $38–$58/hr, CNC programmer or lead $70K–$110K+, quality manager $70K–$105K. BLS occupational data for machinists and tool-and-die makers is the reference point to localize these. Skilled machinists in hot markets will leave for two dollars an hour, and culture retains them better than small raises do.

Certification costs, which gate customer segments. ISO 9001 runs roughly $8,000–$25,000 to implement and certify over six to twelve months. AS9100 — the aerospace quality system built on ISO 9001 with added traceability, first-article inspection per AS9102, configuration management, and risk requirements — runs $20,000–$60,000 over nine to eighteen months including the audit cycle, plus ongoing surveillance audits. ISO 13485 for medical device work carries a similar profile. ITAR registration with the State Department's DDTC is inexpensive to file relative to the compliance burden of controlled access and technical-data handling. CMMC and NIST SP 800-171 compliance is increasingly non-optional for defense subcontracting and carries real IT investment. Each certification is a capital project with a measurable customer-access return — match it to the niche, and do not chase all of them.

Choosing the wedge: trade-offs across niche, equipment, and channel

Every founding decision in this business is a trade between speed of revenue and durability of margin, and the three that matter most are which customers you serve, what you buy, and how you fill the machines.

Customer segment trade-offs. Hardware startups and R&D buyers — robotics, drones, medical device startups, climate tech, national labs — need three-to-ten-day turns, tight communication, and design-for-manufacturability feedback on quantities of one to fifty. They are not price-sensitive within reason, because a delayed prototype costs them weeks of burn. They are also volatile and occasionally vanish when funding does. Regulated production suppliers in aerospace, medical, and defense want certification, traceability, and documentation; the certification is the moat, the sales cycle runs six to eighteen months to get onto an approved vendor list, and the revenue is repeatable once you are on. Industrial OEMs and contract manufacturers needing second-source or overflow capacity give predictable base-load volume at moderate price sensitivity and reward consistency with blanket POs — excellent ballast, dangerous as your entire book. Platform work through CAD-to-quote marketplaces delivers instant deal flow with zero sales effort and thin 18–28% gross margins, and the platform owns the customer relationship, which is why it belongs capped at 20–35% of capacity as gap-fill rather than as a foundation. Local walk-in and repair work carries high nominal rates but heavy quoting overhead per dollar; take it when it's easy, never build around it.

The pragmatic Year-1 configuration for most founders: fast-turn prototype work as the primary wedge, a couple of local OEMs for base load, platform work filling the gaps, and a regulated certification pursued in Year 2 or 3 once cash flow supports it.

Equipment trade-offs. Used versus new is mostly a question of what your niche demands. Used buys you 95% of the capability at 50–60% of the cost, and Haas dominates the small-shop used market because of parts availability, service network, and resale value — though Brother machines excel at fast small parts, and Doosan/DN Solutions, Mazak, and Okuma each have their arguments. New buys warranty, current controls, and the accuracy some regulated work requires. The upgrade ladder runs: 4th-axis rotary on the VMC ($8K–$20K, the best versatility-per-dollar upgrade in the shop), then a true 5-axis machine ($150K–$400K) if the niche rewards it, then automation — pallet pool, bar feeder, robot tending — which is a labor strategy disguised as a cost strategy. The discipline that matters: one VMC and one lathe can produce $300K–$600K of annual revenue. Do not buy the third machine until the first two run above 70% utilization with a backlog behind them.

Software trade-offs. Fusion 360 is the rational default CAM seat for a startup shop at a few hundred to a couple thousand dollars annually, with capable 3-axis work and increasingly credible multi-axis. Mastercam costs several times more per seat but is the lingua franca of experienced machinists and carries deeper capability — worth it if your hires or niche demand it. GibbsCAM, Esprit, and SolidWorks CAM occupy the middle. Quoting software is the highest-ROI and most-skipped category: structured quoting tools let you answer in hours instead of days and defend the number afterward, and spreadsheet quoting loses money both by under-quoting and by being slow. Shop-management ERP can wait until roughly $500K–$800K in revenue — earlier if you're pursuing AS9100, since traceability requirements are far easier to satisfy inside a system built for them.

Channel trade-offs. Ranked by realistic return for a new shop: CAD-to-quote platforms first, because they convert an idle spindle into revenue with no sales cycle. Direct anchor relationships second — identify thirty to sixty target companies in your niche and geography, and get in front of both design engineers (who specify suppliers) and buyers (who approve them), aiming for five to fifteen anchors that supply 60–80% of revenue. Geographic proximity to an aerospace, medical-device, semiconductor-equipment, or defense cluster is itself a channel, because primes prefer nearby suppliers for logistics, just-in-time delivery, and audits — meaning where you sign the lease is a marketing decision. Approved-vendor-list placement is slow but durable. Referrals compound only if you have a category people can name you for. Content and web presence support credibility rather than generating demand. What consistently underperforms: generic directory listings as a primary channel, cold email at scale, paid search on "CNC machining," and trade-show booths bought before you have a niche to sell.

The pitfalls that actually close shops, and how to avoid each

Under-capitalizing working capital. The number-one killer, ahead of bad machining and bad selling. Machining is cash-flow-hostile by construction: you buy material and tooling up front, cut the part, ship it, then wait thirty to ninety days to be paid. A shop can be profitable on paper every month and still die. Mitigation: fund six-plus months of runway as an untouchable line, invoice the same day you ship, enforce terms rather than hoping, take deposits on large first-time jobs, credit-check significant new customers, and secure a line of credit while you still look bankable rather than when you need it.

Cost-plus pricing. Quoting machine hours times a markup guarantees you leave money on every job the customer values highly and lose every bid against a shop with paid-off iron. Value-based quoting prices what the part is worth to the buyer: lead time (a three-day turn is worth multiples of a three-week turn to a company burning runway), tolerance and finish (few shops are genuinely qualified for the tightest work), complexity, certification and documentation, and reliability — a customer burned by a missed delivery will pay a real premium for a shop that always hits the date. Implement it through tiered lead-time pricing with rush at 1.5–2.5x, quantity breaks that reflect setup amortization, minimum lot charges of $150–$500 so tiny jobs are not loss leaders, and explicit non-recurring engineering charges for programming and fixturing instead of burying that investment in the piece price. Then track win rate: above roughly 60–70% you are priced too low; below 20% you are mis-targeted or genuinely uncompetitive.

Over-buying equipment. Three machines on debt before the first has a backlog produces debt service without utilization, which is a spiral rather than a setback. Earn each machine with utilization and backlog.

Customer concentration. One account at 50%+ of revenue is an existential dependency — they get acquired, insource, or have a bad year, and you are gone. Hold the largest customer under roughly 25–30% of revenue and build toward eight to fifteen anchors.

Quote churn. Quoting everything burns the owner's week at a 15% win rate. Triage hard, niche tightly, use real quoting tools, and say no within minutes.

No job costing. Not knowing which jobs and customers actually make money means you keep running the unprofitable ones enthusiastically. Log actuals against quotes from the first job, and review by customer quarterly.

Skipping first-article discipline. Saving an hour on a first article and then scrapping a lot of titanium is a trade nobody makes twice. Beyond routine scrap eating margin silently, a quality escape into an aircraft or medical device is a catastrophic-exposure event — which is also why product liability coverage is the insurance line most often under-bought. The full stack is general liability, commercial property, equipment/inland marine, workers' compensation, product liability, business interruption, and increasingly cyber liability for CMMC-track shops.

The owner as permanent bottleneck. If only the owner can quote, program, and run the hard jobs, the shop cannot scale and cannot be sold. Document processes, cross-train every machinist toward programming and inspection, build the fixture and program library deliberately, and hire into the constraint before it is comfortable.

The hiring ceiling. Growth in this business is usually capped by staffing rather than demand. Solve it structurally: apprenticeship pipelines with local community and technical colleges (registered apprenticeship programs can carry grant support), automation that lets fewer people run more spindles, lights-out running that adds a virtual shift off machines you already own, and a shop culture — clean floor, modern equipment, clear advancement — that makes you the employer of choice locally.

Bad contract terms. Accepting a prime's standard purchase-order terms without review can hand over unlimited indemnification, punitive delivery clauses, and ownership of your programs and fixtures. Have a manufacturing-literate attorney review your terms of sale and the first several customer agreements: lead times, payment terms, default tolerances per print, liability caps, program and fixture ownership, and RMA process.

Ignoring the exit math while building. Small generalist shops trade around 2.5–3.5x seller's discretionary earnings because the value leaves with the owner. Focused shops with AS9100 or ISO 13485, recurring blanket-PO revenue, diversified anchors, documented processes, modern equipment, and an owner who is not the critical path trade at 4.0–5.5x and higher. Buyers include private-equity-backed roll-ups assembling certified-shop platforms, strategic acquirers buying capacity or certifications, competitors expanding geography, and individual buyers using SBA acquisition financing. The useful truth is that build-to-sell disciplines and build-to-run-well disciplines are identical — there is no trade-off between a shop that is good to own and one that is good to sell.

One closing note on how to think about the business side, borrowed from software: the same revenue-operations discipline a RevOps team applies — measuring pipeline conversion, tracking win rate by segment, knowing customer acquisition cost and lifetime value per account, and reviewing cost-to-serve by customer — applies directly to a machine shop. Quote-to-order conversion is your pipeline metric. Actual-versus-quoted job cost is your margin telemetry. Anchor-account revenue concentration is your churn risk. Most shops that plateau are not bad at machining; they are running a business without instrumentation. Build that instrumentation from the first month and the second five years look nothing like the first.

Related questions

How much can one CNC machine realistically produce in revenue per year?

A single well-utilized 3-axis VMC in a focused shop generates roughly $150,000–$300,000 in annual revenue, and a VMC plus a live-tool lathe together support $300,000–$600,000. Utilization above 70% and a high repeat-work ratio matter far more than machine count.

Should I buy a used or new CNC machine for my first purchase?

Buy used unless the niche demands new-machine accuracy or warranty coverage. A well-maintained five-to-eight-year-old VMC delivers most of a new machine's capability at 50–60% of the price, freeing capital for tooling, metrology, and the working-capital runway that actually determines survival.

Do I need AS9100 certification to get aerospace machining work?

For most aerospace production work, yes — it is the entry ticket to approved vendor lists. Budget $20,000–$60,000 and nine to eighteen months. Some prototype and non-flight-critical aerospace work is accessible with ISO 9001 or no certification at all.

How long before a new CNC shop becomes profitable?

Most focused shops reach breakeven within six to twelve months and meaningful owner income by Year 2, at $400,000–$850,000 in revenue. Year 1 typically returns modest owner income as cash recycles into tooling, fixturing, and runway rather than distributions.

Is platform work like instant-quote marketplaces worth taking?

Yes, as gap-fill capped at 20–35% of capacity. It converts idle spindle time into revenue with no sales cycle, but gross margins run 18–28% and the platform owns the customer relationship, so it cannot be the foundation of the business.

FAQ

What is the single biggest mistake new CNC shop owners make?

Opening as a generalist that machines anything for anyone. It feels like maximum market access and is actually minimum pricing power — every quote becomes a price comparison, no part ever repeats so setup never amortizes, nobody can refer you because nobody can describe what you are best at, and the business ends up worth 2.5–3.5x SDE because its value is the owner's hands.

How much three-phase power does a small CNC shop need?

Plan for 200–400 amps of three-phase service for a two-to-four-machine shop, and verify the building's existing service before signing a lease. A VMC in the Haas VF-2/VF-3 class typically needs a dedicated 40–100 amp circuit depending on options and spindle. Upgrading service in a leased building can cost $10,000–$50,000 and take months, so treat available power as a lease-selection criterion rather than a post-signing problem.

Can I start a CNC machining business from a home garage?

Physically yes for light work, and many shops began that way. Practically, zoning restrictions, residential single-phase power, noise and coolant-mist handling, chip and cutting-fluid disposal, and the credibility problem when a customer wants a plant tour before adding you to their vendor list all cap what a garage start can become. It is a viable proving ground for the first few customers; plan the move to commercial space as soon as revenue supports the lease.

What insurance does a machine shop actually need?

General liability, commercial property, equipment or inland marine coverage on the machines themselves, workers' compensation once you have employees, product liability, business interruption, and increasingly cyber liability. Total premiums land roughly $4,000–$20,000 annually at small-shop scale. Product liability is the line most commonly under-bought and the one that matters most when a machined part goes into an aircraft, medical device, or industrial equipment.

How do I compete against instant-quote platforms and offshore shops?

Do not compete for the work they win. Platforms are strong on commodity prototype and low-volume parts priced by algorithm, and weak on complex geometry, tight tolerances, regulated documentation, and relationship-intensive DFM work. Offshore remains cheaper on labor-intensive, non-time-sensitive, non-regulated volume. Target what reshoring is pulling back — certified, complex, fast-turn, and traceability-heavy work — and use the platforms as a customer to fill gaps.

When should I hire my first employee?

Typically at eighteen to twenty-four months, when the owner's quoting and selling time is the binding constraint on growth rather than spindle time. The right first hire is a setup-capable machinist who can run the floor unsupervised, freeing the owner to quote and sell. It is the highest-stakes hire in the shop's history — a bad first machinist can consume enough cash and reputation to end the business.

Sources

• https://www.census.gov/naics/ — NAICS 332710 machine shop industry classification and establishment data
• https://www.bls.gov/ooh/production/machinists-and-tool-and-die-makers.htm — Bureau of Labor Statistics wage and employment outlook for machinists
• https://www.sba.gov/funding-programs/loans/7a-loans — SBA 7(a) loan program terms and eligibility
• https://www.sba.gov/funding-programs/loans/504-loans — SBA 504 loan program for equipment and real estate financing
• https://reshorenow.org/ — Reshoring Initiative data on US manufacturing job reshoring trends
• https://www.nam.org/ — National Association of Manufacturers industry research and policy data
• https://www.themanufacturinginstitute.org/ — Manufacturing Institute workforce shortage research
• https://www.sae.org/standards/content/as9100d/ — AS9100 aerospace quality management system standard
• https://www.iso.org/standard/59752.html — ISO 13485 medical device quality management system standard
• https://www.pmmi.gov/ddtc — US State Department Directorate of Defense Trade Controls, ITAR registration
• https://www.osha.gov/machine-guarding — OSHA machine guarding requirements for machine shops

Related on PULSE

• How do you start a CNC machining business in 2027?
• How do you start a woodworking shop business in 2027?
• How do you start an electronics repair shop business in 2027?
• How do you start an auto wrap shop business in 2027?
• How do you start a welding and fabrication business in 2027?
• How do you start a 3D printing service business in 2027?