AMPulse

China and Germany, full-census demography

The Converted and the Born

Germany's AM census is anchored by companies that predate the industry itself, converts by construction. Three-quarters of China's was founded after 2010. Catching up is a race on one road; this data shows two.

AMPULSE DATA REPORT · 2026Data snapshot: Sep 7, 2026

Older than the industry

The trade-press shorthand for these two countries in AM is a race, and it has been running for years. In 2021, VoxelMatters introduced BLT to Western readers as the first Chinese maker able to "compete with Western (especially German) industry leaders" and wrote that on metal systems "Chinese companies are catching up." By 2024, Forbes had China "catching up to and, in many ways, surpassing the U.S. and Europe." Germany holds the equipment heritage and the qualification depth; China closes the distance machine by machine, price cut by price cut. The frame assumes the two sectors are the same kind of thing at different points on one road.

Counted whole, they are not. The median company in AMPulse's German AM census was founded in 2007. The median Chinese one was founded in 2015. Of Germany's 661 dated companies, 273, 41.3%, existed before 2000, and 182 of those, 27.5% of dated rows, existed before 1990; commercial AM systems date from 1987. The Chinese census holds 67 pre-2000 companies, 6.6%, against 755 founded since 2010, 74.8%.

The German end of that chart is not a quirk of a few conglomerates. Kurtz Ersa, whose Flying Ray machines put it in the census as an AM hardware maker, was founded in 1779. Siemens dates to 1847, TRUMPF to 1923. Even the country's AM-native anchor is old: EOS, founded in 1989, is the same age as the industry it supplies. The census reads as an existing industrial base, machine builders, materials houses, finishing shops, that added a process.

The Chinese census reads as the opposite event. UnionTech, founded in 2000, is the old guard. Farsoon followed in 2009, BLT in 2011, Creality and Eplus3D in 2014, Bambu Lab in 2020. These named companies were formed to do AM. The census cannot say that of every one of the 755 companies founded since 2010, and does not need to: a company founded before 1990 can only be a convert, and a company founded in the 2010s cannot bring a century of prior industrial base with it.

The decade between the two poles is where the censuses come closest. The 2000s hold 187 Chinese companies, 18.5% of the dated census, and 74 German ones, 11.2%: UnionTech's and SHINING 3D's generation on one side, Concept Laser's and SLM Solutions' on the other. It is the only era in which the two shares are within eight points of each other, and in Germany it is small. Each country's AM census was mostly formed either before that decade or after it.

"Catching up" measures China's position on Germany's road. The demography says the roads are different: Germany converted an industrial base it already had, and China formed one.

What converted, and what was born

Composition shows the same split from a second angle. 65.8% of categorized Chinese companies sit upstream, making hardware or materials, against 44.5% of German ones. Services are the biggest German block at 33.3%, roughly double China's 16.6%.

Inside Germany's pre-2000 cohort, the converted layer is concrete: 81 service providers, 72 hardware makers, 59 materials companies, and 32 post-processing specialists, 11.7% of the cohort, mostly finishing and heat-treatment businesses that existed long before there was anything additive to finish. China's whole pre-2000 cohort is 67 companies; it contains 2 post-processing firms.

What composition does not show is an era trend. China's census is upstream-heavy at both age poles, 61.2% of its pre-2000 companies and 67.2% of its post-2010 ones, and Germany's service share exceeds China's at both, 29.7 against 19.4% pre-2000 and 35.7 against 15.1% post-2010. The compositional gap between the countries is stable across founding eras; the demographic gap is the one that moves.

When the converts converted

Founding year says how old a company is, not when it took up additive. The patent record can date it. The additive-only class B33Y was created in 2015, so a company's first B33Y filing is floored there. Patent titles are not: a company that filed on "generative Fertigung" or "三维打印" in 2005 filed on additive in 2005 whatever class the office assigned. The table dates each patent holder's first additive-titled filing.

The German conversion has a first decade and China's does not. 6 pre-2000 German holders filed on additive before 2010 and 16 before 2015; the Chinese pre-2000 cohort has 0 and 2. The earliest German filings are the industry's own history: EOS in 1996, seven years after its founding; TRUMPF in 2002, the same year as Concept Laser, which had been founded two years earlier; Bego, the Bremen dental house founded in 1890, in 2005, when it moved laser-melted crowns into production. Machine builders, a dental supplier, a materials house: the converts were already filing while the process was a prototyping tool.

The earliest Chinese additive filing in the census is from 2013, and four companies share the year: Print-Rite, the Zhuhai printer-consumables maker founded in 1981, Tiertime, Regenovo and Anyprint Unique Printing. Prismlab and Farsoon follow in 2014. Farsoon is the instructive case: founded in 2009 by Xu Xiaoshu, formerly technical director of DTM, the laser-sintering pioneer now inside 3D Systems, and China's flag-carrier in industrial polymer systems, its first additive-titled filing in this record is five years after founding. China's additive patent record begins in 2013 for every age of company, and 73 of the 134 post-2010 Chinese holders with a dated first filing made it in 2020 or later. The born were born recently, and they started filing more recently still.

What the born make

The Chinese post-2010 cohort is 755 companies, and the reading a Western reader reaches for is that it is Creality, Bambu Lab and their imitators: a consumer-printer boom counted as an industry. The count says otherwise. Mark every post-2010 company whose target market is consumer or whose description names desktop, hobbyist or FDM systems: 187 of the 755, 24.8%. Germany's post-2010 cohort, 314 companies, carries 22.3% under the same rule. The desktop share of the young cohort is the same in both countries, and in both it is a quarter. Three-quarters of the Chinese born are industrial: 73.8% of the labeled cohort names an industrial target market, 7.6% aerospace and defense, 5.1% medical. 504 of the 755 make hardware or materials.

Filing is where the born cohort shows its intent. 34.7% of China's post-2010 companies hold at least one active patent link, and 262 of the 375 Chinese patent holders in the census were founded in 2010 or later. The born are the country's filers, and they file young.

Where the two censuses live

The Chinese census is concentrated and the German one is not. The five largest Chinese cities hold 39.8% of the census and Shenzhen alone 15.2%, 157 companies of which 75.7% were founded since 2010. The five largest German cities hold 17.2%. Germany's AM companies sit where its Mittelstand sits, in Ditzingen, Krailling, Lübeck, Herzogenaurach, one or two to a town; the one German city with a young cohort is Berlin, 43 companies and 85% of them founded since 2010, and it is a software-and-services cluster rather than a machine-building one. Shenzhen is what a born industry looks like on a map, one city, one decade, 157 companies, and it is the desktop city: Creality, Anycubic, ELEGOO, Snapmaker and Bambu Lab all sit there, while the industrial born sit in Changsha (Farsoon), Xi'an (BLT) and Hangzhou (Eplus3D, SHINING 3D). The German pattern is what a converted one looks like: the process went to where the factories already were.

The same shape under three instruments

The census objection writes itself: AMPulse finds German companies partly through directories and trade-fair rosters that reach old firms, and Chinese companies heavily through news flow that reaches young ones. The cuts below cannot answer for companies the census never found at all; that residual stays in the limitations. What they test is the composition version of the objection, whether the shape comes from which channel found which company. If it did, the shape should weaken under a different membership rule. It does not.

Among patent-holding companies, where the cut criterion is a patent office's records rather than anything about how AMPulse found the company, the gap widens: 58.6% of German patent holders predate 2000 against 8% of Chinese ones. And read through news flow alone, the channel that most favors young companies, Germany still carries 32.4% pre-2000 to China's 7.5%. No membership rule available makes the two countries look like one population.

Consolidation has an age requirement

The two demographies are already producing different corporate histories. 64 of Germany's 681 census companies, 9.4%, are recorded as acquired, another 65, 9.5%, as subsidiaries of a larger group, and 32 more, 4.7%, as defunct. The Chinese census records 0.9% acquired, 1.8% subsidiaries and 1.1% defunct.

One reading, and the one the ages support: the acquired 9.4% is what an old population looks like after the ownership cycle has passed through it. SLM Solutions sells metal systems as Nikon SLM Solutions, the name it took after Nikon's takeover, launched in 2022 and completed the following year. TRUMPF exited its laser metal fusion business to the DUBAG group. EOS moved the other way along the same cycle, pulling titanium powder capability in-house by acquiring Metalpine in April 2026. The subsidiary 9.5% is a different thing: Heraeus AMLOY, LOCTITE 3D Printing and Siemens Energy Additive Manufacturing are divisions of groups that added additive, corporate structure rather than a cycle, counted as not independent without ever having been independent. If churn follows age, most of China's census, median founding year 2015, has not yet been where Germany's has. The competing reading is coverage, taken up below.

The Chinese defunct figure is a floor: closures in China surface through channels this pipeline reads less densely than English- and German-language trade press.

The counter-thesis

The strongest objection is that the demography here is AMPulse's discovery method, measured and renamed. German companies enter the census through directories, trade-fair rosters and a decade of trade press that names old firms; Chinese companies enter through news flow and the funding ledger, which name young ones. The instrument cuts answer the composition version: inside the patent-holding population, where membership is set by a patent office, the gap is wider, not narrower, and inside the news population, the channel most tilted toward young companies, Germany still carries 4.3 times China's pre-2000 share. What the cuts leave open is the residual: an old Chinese machine-tool group with an additive line and no English-language footprint, or a young German job shop with a metal system and no press release. Both exist. They would have to exist in the hundreds to close a 35-point gap, and the patent cut, which does not depend on either one issuing a press release, points the other way.

The second objection is that "converted" flatters Germany. A company founded in 1923 that bought a printer in 2016 is not an additive company in any sense that matters to a buyer; it is a machine-tool company with a printer. Agreed, and the entry-date table is the answer: 6 pre-2000 German holders were filing on additive before 2010, when the process was a prototyping tool and the filing was a bet, not a purchase. A convert that filed in 2002 is a different thing from one that filed in 2019, and the German cohort contains both. China's pre-2000 cohort contains only the second kind.

The third is that "born" flatters China, because a quarter of the Chinese born make desktop printers and the industry that matters is the other three-quarters. The quarter is there, and the same quarter is in Germany's young cohort. The desktop share is a property of the decade, not the country.

What each reader should take

What to watch

• Consolidation in China. The census records 2.7% of Chinese companies as no longer independent. The first year that figure moves by more than a point is the year the born population meets the cycle the German one has been through.
• Section 849. From 2026-12-18, US defense procurement and operation of AM machines from covered AM companies is barred, with testing, intelligence and electronic-warfare exceptions. Chinese equipment makers' export mix in the following year, and any resulting closures or mergers in the census, are where the consequence will show.
• Berlin. Germany's one young cluster, 43 companies and 85% founded since 2010. Whether it grows, or its companies get acquired by the old base the way Bigrep's operating business was, says whether Germany is forming a born layer of its own.
• German early converts. EOS, TRUMPF, Bego and the other pre-2010 filers are the companies whose additive lines have survived the longest. A divestiture by one of them, on the pattern of TRUMPF's exit from one of its additive lines, laser metal fusion, is the signal that conversion is reversible.
• Acquirer identity. The census holds ownership status without the owner. Collecting the parent for the 129 German rows is the one data addition that changes what this report can say.

Appendix: the company panel

The companies behind the figures. Each country's panel is its twelve deepest patent holders plus its twelve most-covered companies in the trailing 24 months, deduplicated. Local subsidiaries of foreign groups and rows whose first additive filing predates their founding are left out (methodology).

Links are active patent links; B33Y is the subset carrying the additive-only class; first additive filing is the application year of the company's earliest additive-titled patent, blank where no title matched. Stories are pipeline-collected news in the trailing 24 months.

Lee, S. (2026). The Converted and the Born. AMPulse Data Reports. https://www.ampulse.online/reports/china-germany-converted-born-2026