More orders in the pipeline, yet unused equipment in smaller shops

US manufacturers appear to face two very different capacity conditions. In Xometry's 2027 U.S. Manufacturing Outlook, published October 1, 2026, 77% of responding manufacturing executives reported production backlogs larger than a year earlier. Yet among manufacturers generating less than $50 million in annual revenue, 44% said they operated below 70% of capacity.

Among companies with more than $100 million in revenue, the equivalent share was only 16%. Of the companies operating below 70% capacity, 48% cited finding enough work as a major obstacle. Increased backlogs for one part of the market can coexist with inadequate demand for another.

These percentages do not establish that the same factories had both swelling backlogs and idle equipment. The survey covered 150 executives, and the figures refer to different respondent groups. It is not a representative census of all US manufacturing capacity. But it prompts a decision question: How can equipment that already exists become a trustworthy option for new customer orders?

Only 59% planned to hire, despite 77% reporting larger backlogs

Manufacturing Dive's October 7 report found that while 77% of the surveyed executives reported larger production backlogs, 59% said they planned to increase hiring in 2027. The figures do not reveal the overlap between those groups; a simple 18-percentage-point subtraction does not identify the share of order-rich firms holding back recruitment.

Backlogs may reflect order growth, timing, production constraints or customer schedules; they are not a direct count of unavailable production hours. Firms can adjust scheduling, yield, automation, external sourcing or staffing in different ways depending on what the orders require.

Before purchasing another machine or opening another building, buyers can ask whether qualified outside capacity is accessible. Before treating that as a ready-made solution, they must test whether a different shop can actually run the specific materials, tolerances, inspection and delivery terms.

The small-versus-large utilization gap is 28 percentage points

In the manufacturing outlook, 44% of manufacturers below $50 million in annual revenue said they were running at less than 70% capacity. For manufacturers above $100 million, it was 16%. The 28-percentage-point gap means the smaller-group share was 2.75 times the larger group's share.

It does not mean the typical small factory operates at 44% utilization, or that manufacturers between $50 million and $100 million have a particular rate. No reliable industrywide number of uncommitted CNC hours can be calculated from these two group percentages without more information on the firms and equipment involved.

Underutilized small suppliers face a customer-acquisition challenge, while larger buyers trying to increase output face a supplier-discovery and validation challenge. These are related commercial problems, but having available machinery does not establish that the machine can serve a particular buyer.

An idle CNC machine is not necessarily qualified production capacity

Consider an illustrative case: a manufacturer has a backlog of aerospace components, while another shop runs compatible-looking CNC equipment at 60% utilization. Transferring the order would still require material and tolerance capability, appropriate fixtures and programming, inspection evidence, traceability, certifications and reliable delivery. Similar machine labels cannot establish all of these.

Among aerospace and defense respondents to Xometry's survey, 42% named compliance as a top barrier to winning new work. Requirements such as AS9100 quality-system certification, ITAR controls or CMMC cybersecurity protections may matter depending on the contract and product. None is automatically required for every aerospace job.

Xometry itself clarifies that its company certifications do not extend by default to every independent partner. Shops must maintain their own credentials, and work requiring specific certification is directed to eligible verified partners. The installed asset and the asset qualified for a particular order are different units of supply.

Manufacturing AI is expected to act before the machine cuts metal

The areas executives expected AI to affect most in 2027 were manufacturability and costing (51%), quality control and inspection (47%), production execution on the shop floor (32%), and supplier selection and qualification (25%). These are expectations reported by respondents, not measured improvements in productivity.

When a buyer submits a part drawing, a prospective supplier must examine material, tolerance, finishing, quantities, lead time and inspection conditions before quoting or accepting it. Design for Manufacturability (DFM) and cost assessment depend on real process constraints and equipment experience, not just a text model's ability to summarize a document.

Fifty-seven percent of respondents reported meaningful AI returns, up from 44% a year earlier, but these are self-reported investment outcomes rather than a 57% increase in industrial productivity. Xometry's own argument stresses the difference between general-purpose AI and decision support grounded in live production, capacity and pricing data.

Workcenter's redesign targets the decision to accept a job

On September 8, before releasing its manufacturing survey, Xometry launched a redesigned Workcenter for manufacturing partners. The interface centralizes new opportunities, jobs needing attention, messages, performance records and anticipated payouts. Its Job Board puts 3D previews, engineering drawings and requirements alongside filters for manufacturing processes and certifications.

Xometry reported that in its early testing the rebuilt experience reduced job-review time by more than 60% and raised partners' job-acceptance rates by about 15%. These are vendor-reported early results; the announcement does not independently establish the same effect across the industry or a proportional increase in completed output.

Still, the product effort reveals a specific operating problem: suppliers must find, understand, verify fit and accept work before any cutting, assembly or inspection begins. Streamlining that judgment can expand access to available shops without increasing the number of installed machines. It does not remove the need for qualification and reliable execution.

Hard-to-hire workers own different parts of the capacity conversion

Survey respondents identified AI and automation operators (34%) as difficult to hire, followed by quality engineers (31%), maintenance technicians (29%), machinists (23%) and CNC programmers (20%). These figures indicate which occupations respondents named, not vacancy rates across the US workforce.

Those skills address distinct constraints. CNC programming and machining convert designs into acceptable geometry, quality engineering verifies customer specifications, maintenance supports uptime, and automation operations connect equipment with workflow and control systems. Grouping them all as a generic manufacturing labor shortage conceals where an order-to-delivery process can fail.

Pay increases were reported as helpful for hiring or retention by 81% of employers with over 500 employees versus 46% of employers with under 200. This does not prove smaller employers cannot recruit with higher wages. It does suggest firms should also identify which scarce technical capabilities belong in-house and when training or external expertise makes sense.

BANSEOG VIEW | Installed Capacity ≠ Qualified Capacity ≠ Deliverable Capacity

A factory can own a machine and still fail to qualify for a customer's work. A qualified process can still fail a delivery schedule or repeat-volume requirement. We distinguish Installed Capacity (the equipment physically available), Qualified Capacity (the processes, material and certifications suitable for the order), and Deliverable Capacity (the quantity, quality and lead time a supplier can repeatedly achieve).

For buyers, sourcing strategy becomes a question of whether independently verified suppliers can provide meaningful alternative output. For smaller manufacturers, business development means demonstrating actual process capability, credentials and delivery performance, not merely advertising idle equipment. Xometry's quality intake and order-review UI attempt to bridge parts of that gap.

This three-layer framework is Banseog's interpretation of public evidence, not a proven causal finding from the survey. It also changes the talent discussion: the critical role might be a DFM engineer, CNC programmer, quality leader, maintenance specialist or supplier-integration operator. The strategic advantage may lie not in owning more capacity, but in turning qualified existing capacity into dependable deliveries.

BANSEOG VIEW

Banseog View — Installed, qualified and deliverable capacity are separate

The 77% backlog increase and 44% underutilization describe different sets of survey respondents. They cannot be combined into a claim that individual factories were both fully booked and idle.

The >60% early review-time reduction and ~15% job-acceptance increase are Xometry's internal test claims, not independent evidence of sectorwide output or profit gains.

Separate physical equipment (installed), compliance/process fit (qualified) and proven volume/lead-time performance (deliverable), and evaluate whether each relevant operator and engineer has solved the specific problem.

SOURCES

Primary sources and references

Published 13 October 2026. Xometry's 2027 outlook and Manufacturing Dive's October 7 analysis rely on a proprietary survey of 150 US manufacturing executives, not a US manufacturing census. Increased backlog (77%) and underutilization among sub-$50m shops (44%) refer to distinct respondent groups. 16% pertains to firms above $100m with utilization below 70%; 48% describes insufficient work among respondents below 70% utilization. Percentages for AI impact (51/47/32/25%) are expectations, not measured effectiveness. Hard-to-fill occupation shares (34/31/29/23/20%) are survey responses, not labor-market vacancy rates. Xometry's September 8 Workcenter >60% job-review-time reduction and ~15% acceptance gain are early vendor-run tests; neither proves a corresponding increase in delivered parts. Xometry certifications are not inherited automatically by each independent supplier; order-specific standards vary. Installed/Qualified/Deliverable Capacity is Banseog's strategic distinction, not a statistically established causal link between idle capacity and any one hiring gap.