AMPulse

January 2026 retrospective edition

The 2025 AM Capital Map

A count of 604 capital events shows an industry shaped by several different capital systems. The mix matters more than one aggregate total.

Capital composition

604 visible events

47%venture + growth

47%

14%

15%

Venture + growth

284 · 47%

Grants

84 · 13.9%

Strategic

42 · 7%

M&A

32 · 5.3%

Public

42 · 7%

Debt

23 · 3.8%

Other

90 · 14.9%

Unknown

7 · 1.2%

Unresolved classification97 events · 16.1%

AMPULSE DATA REPORT · JANUARY 2026 EDITION · UPDATED SEPTEMBER 2026Data snapshot: Sep 8, 2026

The additive manufacturing capital market did not have one funding story in 2025. It had several, running in parallel.

AMPulse records 604 visible capital events across 514 AM companies during the year, one event per company, instrument and quarter. Conventional venture and growth stages account for 284 of them. The other side of the ledger is not one coherent alternative: it holds 84 grants, 42 strategic investments, 32 acquisitions, 42 public-market events, and 23 debt financings. Another 90 events remain in the ledger's unresolved other category, and 7 lack a stage altogether.

That composition changes the question. A single total-dollar figure would treat a seed round, a public offering, a government grant, an acquisition, and corporate strategic capital as interchangeable units. A count has the same problem in miniature: a £2,500 Kickstarter campaign and a $25M Department of Defense grant are each one row. The count is still the better instrument for this edition, because 359 of the 604 events carry a normalized amount and the missing amounts are not missing at random, so a dollar total would rank instruments by their disclosure habits.

The more useful map begins with the instruments, then changes the denominator.

This was a multi-capital market

Venture still supplied the widest channel. At 47.0% of visible events, the combined pre-seed through later-stage venture and growth group is the plurality of the ledger, not the whole market. Explicit non-VC instruments account for 223 events, or 36.9%, and that share is a floor: the unresolved group holds private placements, SBIR contracts, secondary sales and crowdfunding alongside rounds labelled "Pre-A+" or "additional funding". An AM capital analysis restricted to venture stages would omit more than a third of the visible activity before any unresolved record is adjudicated.

What the non-VC rows contain, as the ledger records them:

• Grants and public programs (84 events) are mostly named agencies writing cheques to named companies: a €60M German federal grant to H.C. Starck Tungsten Powders, $25M from the US Department of Defense to IperionX, $14.2M from the UK's Aerospace Technology Institute programme to GKN Aerospace, and a long tail of Innovate UK awards (Photocentric, Domin, Rapid Fusion), NASA and SBIR contracts, and state economic-development funds. This is among the best-disclosed instruments in the ledger: 79.8% of grant rows carry a parsed amount.

• Strategic capital (42 events) is a stage label, not one instrument. The two biggest rows are financial sponsors: Fortissimo Capital's $120M investment in Stratasys and Partners Group's $104M in restor3d. The corporate rows are smaller and more varied: Han's Laser into Han's Matrix3D, AVIC Heavy Machinery's capital increase in China Aerospace Laser, Lockheed Martin Ventures in Venus Aerospace, Stratasys in Tritone Technologies. Winner Technology's row is a control transfer. Only 31% of the bucket carries an amount.

• Acquisitions (32 events) transfer ownership rather than finance operations. The 2025 rows include Nano Dimension closing its $183M acquisition of Desktop Metal in April; ExOne bought by Anzu Partners out of the Desktop Metal bankruptcy in July, then unified with voxeljet under ExOne Global Holdings in October; AMETEK's $920M purchase of FARO; Patinex taking United Grinding private; Sodick buying AltForm and KEYENCE buying CADENAS. The bucket also carries stake purchases and non-specialist targets, so it is read here as a count of ownership events, never as a dollar figure.

• Public markets (42 events) split between listings and post-IPO raises. The listings include Firefly Aerospace ($868M), Carlsmed ($100.5M) and Ottobock (€100M), Scantech's STAR Market IPO, Lincsolution in Korea, and Yuding Additive's ¥847M pre-IPO round. The post-IPO rows are mostly small placements by already-listed companies.

• Debt (23 events) ranges from the European Investment Bank's €15M venture debt to Artec 3D and the Stifel AM-Forward Fund's $10M to Sintavia, to Commerzbank's €50M facility for Destinus and Sinochem's €478M debt-to-equity conversion at KraussMaffei.

Volume is not penetration

Raw event volume points immediately to Hardware. Its 268 events are 44.4% of the full ledger, 2.3 times Application and 3.5 times Materials. If the analysis stopped there, Hardware would look like the uncontested center of capital activity.

The company denominator changes the picture.

Among 1,754 eligible tracked Hardware companies, 222 recorded at least one 2025 capital event, a company-participation rate of 12.7%. Application reached 14.4% on 116 events. Platform recorded only 33 events, but 29 of its 208 eligible tracked companies had one, a rate of 13.9%.

Three observations follow.

First, Hardware's volume lead is partly structural. It is a large tracked category, its companies are capital-intensive, and a company can record more than one event. Across the whole report cohort there were 1.18 events per company with an event. Event counts measure transaction activity, not the breadth of companies reached.

Second, the volume lead does not become a participation lead. Application and Platform sit above Hardware on the point estimate, and all three intervals overlap, so the ranking is unsettled. What the denominator does establish is narrower and holds: a category with 44.4% of the ledger's events reaches its tracked base at about the same rate as Application, with well under half its event count, and Platform, with an eighth.

Third, Service shows the reverse pattern. It has the largest eligible tracked base in this cut, 2,062 companies, but only 47 had a visible capital event, a 2.3% rate. Part of that is financing behaviour: service bureaus grow through revenue, equipment ownership, or project cash flow rather than external rounds. Part of it is the denominator: the Service base is the least-curated tail of the tracked universe, with lifecycle status unfiltered, so its rate is the one most exposed to dormant and directory-scraped rows.

The disagreement between the left and right sides of the figure is not a measurement problem to smooth away. It is the finding.

Geography changes after the same test

Country counts are even easier to overread. The United States led the 2025 event ledger with 187 visible events, followed by China with 149. Together they accounted for 55.6% of all visible records.

The order of those two depends on the collapse rule. Before source records of one financing were merged, China had 190 rows against 195 for the United States; the merge removed 41 Chinese rows and 8 American ones. Chinese rounds enter the ledger through more sources (a lead announcement, a syndicate announcement, a database sweep) and were being counted once per source. The raw count was measuring ingestion, not financing.

When each country is divided by its eligible tracked-company base, smaller ecosystems rise. Israel recorded a 21.3% company-participation rate, India 12.4%, Canada 12.3%, and China 11.6%. The United Kingdom recorded 9.3% and the United States 8.9%. Germany, fourth by raw event count, fell to 4.7% of its much larger tracked base.

Of the three drivers in the caption, discovery is the one this ledger can weigh, and it cuts the other way from the count: the rate is built on distinct companies, so the multi-source ingestion that inflated China's row count leaves its rate almost untouched. What remains is a ranking of small bases. Israel's interval spans 13.6% to 31.9%, and India, Canada and Australia sit within a point of China.

Who received more than once

The diligence question a count raises is whether capital reached many companies or circulated through a smaller set of repeat recipients. On the collapsed ledger, 75 of the 514 companies (14.6%) recorded more than one 2025 event, and between them they hold 165 events, 27.3% of the ledger.

The concentration is mild, and the repeat rows are not the usual suspects. No company in the ledger recorded more than four events, and the 13 companies with three or more are mostly small: public-program stackers such as X-wave Innovations (three grants) and Verustruct (two grants, a pre-seed round and one unresolved row), companies combining a grant with a round (XO Armor, CPS Technologies, SmartKem), and Chinese hardware makers with Series A rows in more than one quarter (FastForm, Rongsu Technology, Synergy Hi-Tech). ELEGOO's two Series B rows are the one repeat pattern at consumer-hardware scale.

Repeat events in 2025 were a grants-and-extensions phenomenon at the bottom of the ledger, not a re-up phenomenon at the top.

Why there is no total-dollar headline

Only 359 of the 604 visible events carry an amount normalized to USD, and 160 carry a source URL.

A dollar total built on that base would inherit disclosure bias on top of the coverage gap, and the bias runs by instrument. Grants, debt and public-market events disclose: 79.8%, 82.6% and 81% of those rows carry a parsed amount, because the agencies, banks and exchanges publish them. Private rounds disclose about half the time (58.8%), strategic investments a third (31%), and acquisitions rarely (18.8%). A sum of the disclosed subset would weight the ledger toward public money and public markets and away from the private and strategic rounds that carry most of the industry's growth capital.

This edition uses counts for what counts can support, and withholds aggregate dollars until the largest parsed events are source-verified and transaction status separates closed, announced, planned, and acquisition consideration.

What the map says, and what it does not

The strongest conclusion is structural. The AM capital market visible in 2025 was split across venture, public programs, corporate strategy, ownership transfers, public markets, and debt. Investors looking only at venture rounds see the widest slice but miss a material part of the system. Policymakers looking only at grants miss how private and strategic capital select different parts of the value chain. Operators looking only at transaction totals miss that 27.3% of events went to companies that recorded more than one.

The category and country cuts add the same warning twice: a raw count ranks Hardware and the United States, and a denominator puts Application and Platform level with Hardware and Israel, India and Canada above China. A capital map should show both.

The map describes how capital appeared in the ledger, not whether it was well allocated. A grant is not private-market validation, an acquisition is not growth capital, and a participation rate is not a forecast.

Implications

Methodology

Lee, S. (2026). The 2025 AM Capital Map. AMPulse Data Reports. https://www.ampulse.online/reports/2025-am-capital-map