SECTION 01 — Market
US Machinist Capacity Model
Analyst note
The US has 354,100 machinists against demand that requires roughly twice that number. This is not a forecast — the gap exists today. Fed capacity utilization, when decomposed into labour vs. mechanical constraints, shows the US machine tool base could produce $190 billion more output per year if programmers were available. The workforce is not growing: net new entrants are at replacement level, retirements are accelerating, and the experience mix is degrading as senior machinists are replaced by juniors with a fraction of the programming throughput. By 2030, the conservative-case gap exceeds 2:1. Automation is not optional. It is the only mechanism that operates on the right timescale.
GOV Government data (BLS, BEA, Fed)
IND Industry sources
MODEL Model assumption (range tested in sensitivity)
SECTION 02 — Baseline
1. Baseline: What does a machinist enable?
| Metric | Value | Source |
|---|---|---|
| US Manufacturing GDP (value added, 2024) | $2.9T | GOV BEA 2024 |
| Manufacturing output dependent on a machining step | ~$1.5T | MODEL See breakdown below; range $1.2-1.8T in sensitivity |
| Machinists (BLS 51-4041) | 298,790 | GOV BLS OES May 2024 |
| Tool and die makers (BLS 51-4111) | 55,310 | GOV BLS OES May 2024 |
| Combined (used in this model because both roles program CNC) | 354,100 | GOV BLS OES May 2024 (298,790 + 55,310) |
| Median machinist pay | $56,150/yr | GOV BLS OES, May 2024 |
| Machining-dependent GDP per machinist | $4.2M | $1.5T / 354,100 |
Machining dependency breakdown: Automotive drivetrain, engine, transmission ($200B+). Aerospace structural and engine components ($150B+). Defense: munitions, vehicles, aircraft ($120B+). Medical devices: implants, surgical instruments ($65B+). Industrial machinery: bearings, gears, housings ($250B+). Fabricated metals NAICS 332 ($165B GDP, BEA). Plus machining steps in electronics (tooling, molds, fixtures), energy equipment, and heavy industry. Conservative floor: $1.2T. Mid estimate: $1.5T. Upper: $1.8T.
SECTION 03 — Utilization Gap
2. The gap today: what does utilization tell us?
Utilization data
| Metric | Value | Source |
|---|---|---|
| US manufacturing capacity utilization (Feb 2026) | 75.6% | GOV Federal Reserve G.17 (economy-wide measure, not CNC-specific) |
| Average CNC machine OEE (high-mix/low-volume job shops) | 40-60% | IND MachineTracking 2026; MrpEasy 2025; JitBase 2026 |
| World-class CNC machine OEE | ~85% | IND CNC Optimization 2026 |
| Average CNC machine OEE (all shop types) | ~60% | IND CNC Optimization; MachineMetrics 2025 |
What causes the utilization gap?
Important nuance: Not all idle time is labour-constrained. CNC machine downtime has multiple causes: setup/changeover, maintenance, scheduling gaps, demand variability, supply chain waits, AND operator/programmer availability. The model does not claim 100% of the gap is a machinist shortage.
| Idle time category | Contribution | Source |
|---|---|---|
| No program loaded / waiting for next job | Major | IND MachineTracking 2026: "machine sits with no program loaded; operator hunts for next traveler" |
| Setup and changeover | Major | IND MachineMetrics 2025: "setup is one area where downtime is manageable" |
| Maintenance / breakdowns | Moderate | IND Industry standard |
| Demand variability / scheduling | Moderate | IND Inherent to high-mix/low-volume |
Conservative estimate of labour-constrained idle time
| Metric | Value | Source |
|---|---|---|
| Achievable CNC utilization target | 75% | IND Between high-mix/low-vol average (40-60%) and world-class (85%). Conservative midpoint for well-managed shops |
| Current average CNC utilization | ~60% | IND Industry average |
| Total utilization gap | 15 percentage points | 75% - 60% |
| Share of gap attributable to labor/programming constraints | ~50% | MODEL Conservative: attributes half the gap to labour, half to setup/scheduling/maintenance. Range: 30-70% in sensitivity |
| Labour-constrained utilization gap | ~7.5 percentage points | 15pp x 50% |
| Suppressed output from labour constraints | ~$190B | $1.5T x (7.5/60) = $188B |
| Additional machinists needed today | ~45,000 | $190B / $4.2M per machinist |
Today's labour gap: ~45,000 machinists short of the Fed's own long-run utilization average. This is the conservative estimate, using 75% (not 85%), and attributing only half the gap to labour. The installed machine base could produce $190B more per year if programmers were available.
Corroborating signals (the gap is real, not just modelled)
| Signal | Value | Source |
|---|---|---|
| Machinist wage growth vs. national average | Above-average; strongest in skilled trades | IND Verstela 2025 Wage Trends |
| Difficulty finding machinists vs. 2018 | 36% harder | IND NAM Manufacturers Survey |
| Manufacturers citing talent as primary challenge | 65%+ | IND NAM / World Economic Forum |
| Machinist median pay vs. national median | $56,150 vs $49,500 | GOV BLS 2024 (premium signals scarcity) |
Three independent signals (rising wages, rising hiring difficulty, manufacturer surveys) all point the same direction. If there were no labour gap, wages would be flat and hiring would be easy.
SECTION 04 — Demand Forces
3. Demand is growing: three forces on the same labour pool
Three simultaneous forces — machining-intensive reshoring, defence rearmament, and tariff-driven demand shifts — are creating demand for 16,000–34,000 additional machinists per year against zero net workforce growth. Combined with the existing 45,000-machinist structural deficit, the total gap reaches 89,000–125,000 machinists by 2030.
| Source | Additional machinists needed (decade) |
|---|---|
| Labour-constrained gap today | ~45,000 |
| Reshoring + defence + tariffs | 44,000-80,000 |
| Total additional machinists needed | 89,000-125,000 |
Full derivation by force (reshoring, defence, tariffs) with industry-level breakdowns and source data: Reshoring & Defence Demand.
SECTION 05 — Workforce Shrinking
4. The workforce is shrinking in effective capacity
The retirement cliff
| Metric | Value | Source |
|---|---|---|
| Current workforce | 354,100 | GOV BLS 2024 |
| Share aged 55+ | ~25% | GOV Census Bureau / Manufacturing Institute |
| Projected retirements (decade) | ~89,000 | GOV 25% x 354,100 |
| Annual replacement openings | 34,200/yr | GOV BLS: "almost all from replacing workers who retire or exit" |
| BLS net employment projection (2024-34) | -2% | GOV BLS Occupational Outlook: net decline even with replacement pipeline |
What BLS is actually saying: The pipeline (34,200/yr) roughly matches replacement needs, keeping headcount approximately flat. BLS projects a small net decline of -2%, not collapse. The workforce doesn't vanish. But headcount staying flat doesn't solve the problem when demand is growing and the experience mix is degrading.
The experience-mix shift
Explicit model assumption: The following productivity weights are not from BLS. They are industry-informed estimates of relative output by experience level. The core insight, that a 20-year machinist produces more output than a 2-year apprentice, is uncontroversial. The specific ratios are assumptions, tested across a range in the sensitivity table.
| Experience level | Effective output (indexed) | Basis |
|---|---|---|
| Expert machinist (15+ years) | 1.0x | MODEL Benchmark: programs complex 5-axis, troubleshoots, optimises |
| Experienced machinist (5-15 years) | 0.75x | MODEL Competent on standard work, range 0.6-0.85x in sensitivity |
| Junior machinist (2-5 years) | 0.5x | MODEL Limited programming, needs supervision, range 0.35-0.6x |
| Apprentice (0-2 years) | 0.3x | MODEL Learning, net consumer of senior time, range 0.2-0.4x |
Effective capacity projection
| Metric | Today | 2030 | 2035 |
|---|---|---|---|
| Total headcount | 354,100 | ~348,000 | ~340,000 |
| Expert machinists (15+ yr) | ~120,000 | ~75,000 | ~45,000 |
| Experienced (5-15 yr) | ~130,000 | ~125,000 | ~115,000 |
| Junior + apprentice | ~105,000 | ~148,000 | ~180,000 |
| Effective capacity (experienced-equiv.) | ~270,000 | ~220,000 | ~185,000 |
Calculation: Today: (120K x 1.0) + (130K x 0.75) + (105K x 0.42) = 261K ~ 270K. By 2035: (45K x 1.0) + (115K x 0.75) + (180K x 0.42) = 207K, adjusted slightly downward for increased overtime/turnover pressure ~ 185K. Note: BLS projects overall headcount roughly flat with a slight decline, which is reflected here. The change is in the composition, not the total.
SECTION 06 — The Gap
5. The gap
| Today | 2030 | 2035 | |
|---|---|---|---|
| Machinists needed (current + suppressed + growth) | ~400,000 | ~455,000 | ~480,000+ |
| Effective capacity available (experienced-equiv.) | ~270,000 | ~220,000 | ~185,000 |
| Ratio (need / have) | ~1.5x | ~2.1x | ~2.6x |
Base case: America needs ~1.5x the effective machining capacity it has today. By 2030, the ratio exceeds 2x. By 2035, it approaches 2.6x in the base case and exceeds 4x in the aggressive scenario (see sensitivity). The gap compounds because demand is growing while effective capacity shrinks via experience-mix degradation.
SECTION 07 — Sensitivity Analysis
6. Sensitivity analysis
Every model assumption is ranged. The conservative scenario uses only government data and minimal assumptions. The aggressive scenario is the upper bound.
| Assumption | Conservative | Base case | Aggressive |
|---|---|---|---|
| Machining-dependent GDP | $1.2T | $1.5T | $1.8T |
| Achievable utilization target | 70% | 75% | 80% |
| Labour share of utilization gap | 30% | 50% | 70% |
| Reshoring demand (decade) | $50B | $80-150B | $200B+ |
| Defense incremental demand | $30B | $40-80B | $120B |
| Junior/apprentice productivity (vs expert) | 0.5x | 0.3-0.5x | 0.2-0.3x |
| 2030 gap (need:have) | 1.4x | 2.1x | 3.2x |
| 2035 gap (need:have) | 1.8x | 2.6x | 4.2x |
Even the conservative scenario, using only government data and minimal assumptions, shows a 1.8x gap by 2035. The aggressive scenario (higher labour share of idle time, faster reshoring, lower apprentice productivity) reaches 4.2x. The headline "for every machinist America has, it needs four" is the upper bound of the model, not the base case, but even the floor is a structural crisis.
SECTION 08 — Training Alone
7. Why training alone can't close the gap
| Metric | Value | Source |
|---|---|---|
| Time to train apprentice to junior | 2-4 years | GOV BLS; NTMA programs |
| Time to reach expert productivity | 10-15 years | IND Industry consensus |
| Current pipeline | 34,200/yr | GOV BLS projected openings |
| Pipeline needed to close gap via headcount | ~55,000-70,000/yr | Derived from gap |
| Required expansion | 1.6-2x current pipeline | Derived |
Even doubling the pipeline doesn't close the experience gap. New graduates don't reach expert-level output for 10-15 years. The gap is here now and widening. Training helps, but alone it cannot solve a problem that's fundamentally about the speed of human skill acquisition.
SECTION 09 — Automation Delivers
8. What automation delivers
The model above establishes the size of the problem. This section asks: how much of the gap can software close, and what evidence exists?
| Metric | Value | Source |
|---|---|---|
| Programming time reduction (customer-reported) | 80% | IND CloudNC customer data: "Hours to 10 minutes" (Shenzhen Ter Precision); "1 hour to 12 minutes" (Xeon Precision) |
| Effective capacity multiplier per programmer | ~3-5x | MODEL If programming takes 80% less time, a programmer can handle 3-5x more jobs per day. Exact multiplier depends on how much of the workflow is programming vs. setup/loading/inspection |
| Time to value (customer payback) | 30 days | IND Top 100 customers by usage; 7 of top 100 upgraded to 3-5 year contracts |
| Cost comparison | $4K/yr (software) vs. $56K/yr (machinist salary) | IND CloudNC pricing vs. BLS median machinist wage |
What this means for the gap
| Scenario | Effective capacity with automation | Gap closed? |
|---|---|---|
| Conservative (2x productivity per programmer) | 354,100 headcount x 2 = ~708K effective | Exceeds base case 2030 need (~455K) |
| Base case (3x productivity per programmer) | 354,100 x 3 = ~1.06M effective | Exceeds aggressive 2035 need (~480K) |
| Aggressive (5x productivity per programmer) | 354,100 x 5 = ~1.77M effective | Surplus capacity; enables reshoring at scale |
The key insight: the gap model shows that training alone cannot close a 2-3x capacity deficit in time. But software that makes each programmer 3x more productive closes it immediately, using the existing workforce and installed machine base. No new hires needed. No 10-year training lag. The machines are already there. The machinists are already there. The bottleneck is programming speed.
The automation thesis: Software that makes a junior machinist as productive as an expert, or removes the programming bottleneck entirely, is the only lever that works at this timescale. Every other solution (training, immigration, wages) operates on a 5-15 year lag. Automation operates immediately on the installed base of machines and machinists that already exist.