The capacity you already own

Before you read

Check existing capacity before approving a new line

A request for a new line usually reaches the capital committee with one sentence doing most of the work: the plant is full. This report tests that sentence against public data, then gives the committee a way to test it inside its own plant before the money is committed.

The argument runs in three steps. National statistics show that American plants, on average, run below their own long-run utilization and operate well under half the hours in a week. Plants themselves report why: weak orders explain most of the gap, but some plants name limits inside the plant. And the hours lost inside scheduled time, to stops, changeovers, slow running and scrap, can be priced and set against the other ways of buying an hour.

Two levers run through every chapter. The first is unscheduled time: hours in the week the constraint is not planned to run. The second is losses in scheduled time: hours it is planned to run but does not produce good output. A new line competes with both, and which is cheapest depends on how many hours the plant actually needs.

Summary of findings

Numbers carry a mark: a CAP or TC-V identifier for a published statistic, TC-D for our own arithmetic on one, and Illustrative for an output of the report’s model line.

Manufacturing runs below its own long-run average

The Federal Reserve estimates that US manufacturing ran at 75.7% of capacity in August 2026, 2.5 percentage points below its 1972–2025 average of 78.2% [CAP-001] [CAP-003].

The average plant works 66 of the week’s 168 hours

In the second quarter of 2026, US manufacturing plants operated an average of 66 hours a week [CAP-006], or 39% of the calendar week [TC-D-002].

This is the bluntest capacity number in public statistics, and it describes the first lever: unscheduled time. It does not mean every plant could run around the clock. Continuous-process industries such as refineries and paperboard mills already report 168 hours [CAP-006], and many discrete plants are bound by demand, staffing or regulation.

Utilization gaps differ across industries

In August 2026, 14 of 19 manufacturing industry groups ran below their own 1972–2025 averages and one ran at it [TC-D-004]. Furniture, printing and primary metals ran about 10 points below; machinery, electrical equipment and nonmetallic minerals ran above [CAP-026].

Weak orders are the most cited reason

When the Census Bureau asks plants why they run below full capability, 67.1% cite insufficient orders. Next come insufficient supply of labor, at 22.2%, and equipment limitations, at 11.5% [CAP-005].

This is the honest starting point: most idle capacity is a market problem, and no improvement program fixes a market problem.

Manufacturers raised capital spending by 10.6% in 2022

US manufacturing companies with employees spent $314.3 billion on structures and equipment in 2022, up 10.6% from $284.2 billion in 2021 [CAP-017].

Output per hour has grown 0.5% a year this cycle

Manufacturing labor productivity has grown at an annualized 0.5% in the current business cycle, above the previous cycle’s 0.1% but well below the long-term rate of 2.1% since 1987 [CAP-024].

Downtime took 7.8% of planned time in the NIST sample

In a NIST study of US discrete manufacturers, downtime averaged 7.8% of planned production time, ranging from 4.0% to 12.8% across groups of plants by size and industry [CAP-012].

Lost sales may be the largest maintenance loss

NIST estimates $100.2 billion of sales lost to delays and defects from preventable maintenance issues [CAP-007], with a 90% confidence interval of $33.5 billion to $166.8 billion, and $88.3 billion when estimated by size and industry [TC-V-001]. That is revenue, not profit.

The most reactive plants report the most downtime

In the NIST sample, plants in the quartile most reliant on reactive maintenance reported 3.3 times the downtime and 16.0 times the defects of those in the least reactive quartile [CAP-010].

These are associations from a small, self-selected sample, not proof that a different maintenance mix will produce the same gains [CAP-010]. Changeovers, slow running and minor stops are harder still. No government series measures them, and no public dataset supports an industry-average OEE. For these losses the reliable number is the one a plant measures itself.

In the illustrative line, losses exceed a weekend day every week

The four losses take 1,312 of the line’s 6,000 planned hours, about 26 hours a week: more than the line would gain by running round the clock every Saturday (Illustrative).

On the illustrative line, one OEE point is worth $72,000

Priced at $1,200 an hour of contribution, the four losses cost the illustrative line $1.65 million a year, including $77,000 of variable cost forfeited on scrapped units (Illustrative).

One point of OEE on this line is 60 hours of planned time, 60,000 units, or $72,000 of contribution a year (Illustrative). The price applies only on a constraint with demand. On a line that is not the bottleneck, a recovered hour adds inventory, not sales.

Price your own hour before you price a new one

In 2021, value added per production-worker hour ranged from $62 in apparel to $1,022 in petroleum and coal products, against $180 for manufacturing as a whole [CAP-029] [TC-D-006-01] [TC-D-006-06] [TC-D-006-22].

Three ways to buy an hour

A plant short of hours on its constraint can recover losses, schedule more time, or build. On the illustrative line, recovery is cheapest for a few hundred hours, recovery plus weekend scheduling for up to about 2,200, and only a new line covers more (Illustrative).

Seven figures a capital request should show first

The Hidden Line Test is a one-page gate for any request to add capacity. It asks the sponsor to prove demand, then to show the capacity already standing, scheduled or not, and to price it, before the committee prices the capacity to be built.

The decision rule

Buy hours in order of cost: recover the losses on the constraint first, schedule more of its week second, and build only for the hours the first two cannot supply.