AMPulse

Cohort analysis

The New Shape of AM Formation

Service, polymer and education gave up share of new AM company formation between the 2010s and the 2020s; application, metal and aerospace/defense took it. The aerospace/defense move survives a strict supplier-only count.

Value-chain share of new AM companies · by formation cohort

Hardware

Service

Application

AMPULSE DATA REPORT · 2026Data snapshot: Sep 8, 2026

Correction, 2026-08-05. The closing section gave the largest single-cohort Aerospace/Defense step as +4.9pp; the correct figure at the 2026-05-14 snapshot was +4.8pp, which § 3 already reported. The frozen JSON was unchanged.

Revision, 2026-09-08. Re-extracted against a fresh snapshot with the shared census filter, following the series review. Cohort sizes grew from 2,811 / 929 to 3,034 / 1,130; Service moved from 34 to 25% to 33 to 26%, Polymer from 56 to 40% to 53 to 38%, A/D from 6.6 to 13.0% to 8.5 to 13.9%, and the 2024-26 Hardware and Polymer rebounds reported in § 4 did not survive the larger cohort. The counter-thesis section and the country split are new; one spotlight card was dropped for failing AM relevance. Full list in data/manifest.json.

Setup: three shifts, one direction

Newly founded additive-manufacturing companies look different in the 2020s than they did in the 2010s, and the difference sits in three categories. Service value-chain share of new AM companies fell from 33% in the 2010s to 26% in the 2020s. Polymer-AM share fell from 53% to 38% while Metal-AM share rose from 19% to 23%. Aerospace/Defense target-market share rose from 8.5% to 13.9%, while Education fell from 4.8% to 1.1%.

The shifts mostly resolved by 2020. The 2020-23 cohort (n=858) already sits within a few percentage points of the 2024-26 cohort (n=272) on every headline dimension; the current cohort reflects a change that has already happened rather than one still accelerating.

Two categories did not shift. Hardware as a value-chain position was 31% of 2010s formations and 31% of 2020s formations. Industrial as a target market was 65% across the 2010s and 62% across the 2020s, still the dominant target market in both decades. The shifts are in the categories around hardware and industrial, not in those two. That is the least obvious finding in the report: the largest category on each axis held its place while everything around it moved.

§ 1: Service down, application up

Service share fell from 33% of new AM company formations in the 2010s to 26% in the 2020s. Over the same window, Application share rose from 8% to 13%. Hardware stayed at 31% and 31%. The movement within the value chain is Service-to-Application, not Service-to-Hardware; the residual is spread thinly across Materials, Software, Platform and Post-Processing, each of which moved by less than 2pp at the decade level.

The Service decline is monotonic across the four cohorts: 36.4% in 2010-14, 30.0% in 2015-19, 25.8% in 2020-23, 25.4% in 2024-26. Most of the decade fall happened between 2010-14 and 2015-19 (6.5pp) and between 2015-19 and 2020-23 (4.2pp); the 2020-23 to 2024-26 step is under half a point. Service is still the second-largest value-chain category in the 2020s, below Hardware and above Application; it did not collapse.

The Application rise is monotonic through 2020-23: 6.4% / 9.8% / 13.9% / 11.8%. The Application category groups companies whose primary product is something made using AM (an aerospace component, a medical device, a defense platform) rather than the printers, materials, software or services the rest of the value chain provides. Where that rise came from is not where a reader would guess. Of the 151 2020s Application rows, 46 target Consumer, 39 Industrial, 35 Aerospace/Defense and 27 Medical; Aerospace/Defense was 28% of Application rows in the 2010s and 23% in the 2020s. The Application rise is a consumer-product and industrial-part story at least as much as a defense one.

Hardware was the largest category in both decades, and the spotlights show how much variety that flat number hides: GRU Space (lunar-infrastructure printers) and Creator Precision (Electric-Field-Driven jet deposition for sub-micron precision) are both Hardware. Deep Manufacturing, whose multi-robot WAAM cells sell ultra-large metal parts rather than machines, is Service, and is the kind of 2020s service formation the aggregate decline conceals.

Where the displaced service-bureau demand went is a separate question, and formation data cannot reach it. Fewer new bureaus is also what consolidation into fewer, larger production-oriented providers looks like from the formation side; Protolabs, Xometry and Materialise publish service revenue in their filings across the same window, and that is where the answer would come from. Formation counts the entrants. It does not count the work.

§ 2: the polymer-metal reshuffle

Polymer-AM share of new AM company formations fell from 53% in the 2010s to 38% in the 2020s. Metal-AM share rose from 19% to 23%. Polymer remains the plurality material category in the 2020s; it ceased to be a majority. The ratio of polymer to metal formations moved from about 2.7:1 to about 1.6:1.

The Polymer decline did most of its work early: 61.8% (2010-14), 45.6% (2015-19), 37.8% (2020-23), 39.0% (2024-26). The first step alone is 16pp. The Metal rise did the same: 14.1% / 24.0% / 24.0% / 21.0%, with the +10pp move between 2010-14 and 2015-19 as the dominant step, then a plateau. Both headline material moves were finished by 2015-19; the 2020s cohorts hold the level.

The other material categories remained a long tail across both decades. Multi-Material moved from 5.0% to 4.4%, Bioprinting from 3.2% to 3.0%, Ceramics/Composites from 4.1% to 4.8%. Of the 7 2020s spotlights, Deep Manufacturing and Astrobase are Metal AM, GRU Space is Construction AM, and KeratOPrinter (bioprinted corneas) and Magnet Defense (Multi-Material via hybrid AM) round out the spread.

The metal rise has a geography. China is the one large formation base where metal was already a third of new AM companies in the 2010s and stayed there, and its weight in the world's new metal-AM formations rose from 27% to 32%. The United States went the other way on both readings. The United Kingdom and India doubled their within-country metal shares from a low base; Germany held its level.

§ 3: Education collapse, aerospace-and-defense rise

Aerospace/Defense target-market share of new AM company formations rose from 8.5% in the 2010s to 13.9% in the 2020s. Education fell from 4.8% to 1.1%. Industrial remained the dominant target market in both decades (65% and 62%), at a magnitude no other target market approaches.

The A/D rise is monotonic across the first three cohorts and plateaus: 6.1% (2010-14), 10.6% (2015-19), 14.3% (2020-23), 12.5% (2024-26). The +4.5pp step from 2010-14 to 2015-19 sits inside the 2010s; the +3.8pp step from 2015-19 to 2020-23 crosses the decade boundary. By 2020-23 the share had reached its decade level and the 2024-26 cohort sits at it. Report #6 in this series takes up the why (sub-segmentation, geography, policy); this section establishes the what and, in the next section, whether the what is real.

The Education collapse is concentrated even earlier: 8.6% (2010-14), 1.7% (2015-19), 0.8% (2020-23), 1.8% (2024-26). The MakerBot-era school and maker position (desktop polymer hardware sold into K-12 and universities) was a 2010-14 phenomenon; new AM companies targeting Education have been at or below 2% in the three cohorts since.

The other target markets held shape. Consumer moved from 11.7% to 12.0%, Medical from 9.1% to 8.5%, Automotive from 0.8% to 1.2%. Industrial's 64.7% to 62.5% is the only non-headline decade movement, and it is smaller than the A/D gain, so A/D's share came from Education and the residual as much as from Industrial.

The A/D rise is an American rise first. The United States accounts for 52% of the 2020s A/D formations, and A/D went from 20% to 28% of all new American AM companies. Its largest step, 2015-19 to 2020-23, predates Section 849 of the FY2026 NDAA (effective 2026-12-18, barring DoD procurement or operation of covered AM machines from covered AM companies), so the statute is a tailwind for a formation pattern already in place, not its cause. China's A/D share within its own formations rose from 6.2% to 10.2%, on a different demand base; the United Kingdom and India moved from single digits to 12% and 17% on small counts.

Counter-thesis: is the classification doing the work?

The strongest objection to § 3 is that the A/D rise is a taxonomy artifact. AMPulse counts a launch-vehicle startup, a lunar-base startup and an unmanned-vessel builder as AM companies because AM is decisive to how they build, not because they sell anything additive. If the 2020s A/D rows are mostly that kind of company, "A/D share rose" means "AMPulse started counting rocket companies", and a buyer should discount the finding to zero.

The value-chain field lets the objection be tested. Of the 157 2020s A/D rows, 115 are AM-producing (Hardware 65, Service 30, Materials 14, Software 4, Post-Processing 2), 35 are AM-decisive Application rows, and 7 are Platform, adjacent-equipment or unclassified. In the 2010s the same split was 179 producing, 69 Application and 10 other out of 258. The AM-decisive share of A/D formations went down, from 27% to 22%.

Under the strict supplier-only rubric the A/D share moves from 6.9% to 12.9%, a larger move than the published one, because AM-decisive users were a bigger part of the 2010s A/D base than of the 2020s one. The rise is carried by companies that sell printers, feedstock, software and printed parts into aerospace and defense, in that order of count. The objection, taken seriously, strengthens the finding.

Two qualifications survive. The value-chain field is itself an AMPulse classification, and a handful of the 2020s Application rows read as metal service bureaus that name aerospace as their market; moving them to Service would shift a few rows between the numerator buckets without changing the direction of any rubric. And a supplier that targets A/D is a positioning count, not a demand count: a defense-targeting powder startup is evidence of where founders and their investors think the customers are, not of a contract. Report #6 pairs this formation series with the contract ledger.

The reader who came to discount the A/D rise as a taxonomy artifact leaves with a bigger number.

Spotlights: 10 companies across the shift

10 companies in two groups. Three era-illustrative companies whose formation predates the 2010 window but whose visibility defined the ecosystem the 2010s cohort formed within, and 7 2020s-archetype companies from the 2020-2026 cohort that exemplify the categories toward which composition has moved. Each 2020s card states whether its AM relevance is verified (the company markets an AM product or service) or inferred from the company's stated build method.

Group A: Era-illustrative companies

Founded before the cohort window opens in 2010, so not cohort members; the most visible exemplars of the 2010s service-and-consumer AM ecosystem that the cohort numbers describe in aggregate.

Group B: 2020s archetype companies

§ 4: 2024-26, a rebound that did not survive

The first edition of this report, at a 2026-05-14 snapshot with 163 companies in the 2024-26 cohort, found two moves inside the 2020s pointing back toward the categories the decade had shifted away from: Hardware from 34% to 40% and Polymer from 39% to 46%. At this snapshot the cohort holds 272 companies, and both moves are gone. Hardware reads 31.1% in 2020-23 and 30.9% in 2024-26; Polymer reads 37.8% and 39.0%.

What happened between the two snapshots is discovery, not founding: the 109 companies added to the 2024-26 cohort in four months were founded before the first snapshot and surfaced after it, and they were less often hardware and polymer than the 163 that had surfaced first. The first-found rows of an open cohort are the ones with early press, which in polymer AM means the desktop and prosumer entrants that ride the fastest-growing segment of the equipment market (desktop polymer under EUR 10,000 growing at 30%+ a year, per the industry context in the methodology). A polymer rebound in new formations may still exist inside that segment; the whole-cohort share no longer shows it.

What the cohort does show is the three decade shifts holding. Service sits at 25.4% against 25.8% in 2020-23, both far below 36.4% in 2010-14. Aerospace/Defense sits at 12.5% against 14.3%. The window stays open through 2026, so the 2024-26 readings will move again on the next extraction, and by less each time.

§ 5: what this signals

Closing: the asterisks

The composition movement is real in the AMPulse database and consistent across three classification axes that describe different aspects of each company. Four things bound it, each owned by the methodology box: the numbers are shares of a cohort, not formation rates; the material taxonomy changed in April 2026 and two categories are excluded from the headline for it; older-cohort discovery bias can inflate the magnitudes, not the directions; and every axis is an AMPulse classification that a reader can re-run from the frozen JSON. The counter-thesis section adds the one that matters most for the headline this report is quoted for: under a supplier-only count the A/D rise is larger, not smaller.

Which categories moved is settled in this data. Why they moved is not, and formation counts cannot carry it; the single-axis deep-dives in Reports #5 and #6 take it on.

This is the second report in the AMPulse Data Reports series. Report #1, India's Sovereignty AM, covered the same database from a country-pair frame. Forthcoming reports in the series:

• R3, Series A Squeeze: capital bifurcation in AM funding rounds (in progress).
• R4, Active Investors 2026: who writes AM checks.
• R5, Where Service Went: single-axis deep-dive on the Service-VCP formation decline and the consolidation behind it.
• R6, Defense-Industrial AM: single-axis deep-dive on the Aerospace/Defense rise, paired with the contract ledger.

Send corrections to [email protected] (or [email protected] until the alias exists).

Lee, S. (2026). The New Shape of AM Formation. AMPulse Data Reports. https://www.ampulse.online/reports/am-industrialization-2026