Skip to content
Work samples

What our analyses look like

Four extracts from ongoing work — ordered by the question they answer, not by date. Every chart appears here together with the methodology behind it and with its source.

These charts are extracts. Every report covers competitor profiles, benchmarking, market sizing, policy and regional analysis, plus the patent layer across every company on record.

Chart labels are in English throughout, as they come from the bilingual analysis series. The reading guide below each chart explains every axis.

This is what a process comparison looks like

Six joining processes side by side — which one is pulling ahead, which one has matured?

What the chart shows
The number of patent families per joining process and year, worldwide, from 2015 to 2023. The comparison covers laser welding, arc and plasma welding, polymer joining, soldering and brazing, friction stir and pressure welding, and resistance welding.
How to read it
The horizontal axis shows the earliest filing year of a family, the vertical axis the number of families (0 to roughly 5,750). Each line is one process and is labelled directly. The dip at 2019 is not a market movement but a coverage gap in the source; it affects all processes equally.
Line chart of six joining processes from 2015 to 2023. The laser curve rises most steeply and sits clearly above all other processes by 2023.
Source: Google Patents Public Data (Google & IFI CLAIMS Patent Services, 2026). Analysis by Vertumen, primary CPC classification, families 2015–2023. · Open chart at full size
Finding
Laser grows fastest and keeps widening the gap to every other process. That does not make laser the largest market — mature processes such as resistance welding see few new patents even though they run in millions of production lines.
Methodology in detail
The counting unit is the patent family, deduplicated by earliest filing — the same invention is filed in several countries and would otherwise be counted repeatedly. Exactly one primary CPC class is counted per process: laser B23K26; arc and plasma B23K9 and B23K10; polymer joining B29C65; soldering and brazing B23K1 and B23K3; friction stir and pressure welding B23K20; resistance B23K11. Deliberately not the broad class B23K, which mixes soldering and laser cutting into one bucket. Filing years from 2024 onward are excluded because filings from roughly the last eighteen months are not yet fully published — otherwise every curve would drop off artificially at the end.

This is what a competitor comparison looks like

Who actually holds the intellectual property in a process — and does that match brand perception?

What the chart shows
The seven most active applicants in arc welding, measured in patent families filed between 2015 and 2023, sorted in descending order.
How to read it
Each bar is one corporate group, its length the number of patent families, and the axis starts at zero. Subsidiaries are consolidated under their parent — which is why Miller and Hobart do not appear separately but inside the ITW row.
Bar chart of the seven most active applicants in arc welding. ITW leads with 561 patent families ahead of Lincoln Electric with 263 and Panasonic with 217.
Source: Google Patents Public Data (Google & IFI CLAIMS Patent Services, 2026). Analysis by Vertumen, primary CPC B23K9, corporate applicants aggregated. · Open chart at full size
Finding
Illinois Tool Works, a diversified group the industry rarely sees as a welding brand, leads with 561 families. Lincoln Electric follows with 263, Panasonic with 217 and Kobelco with 187. Fronius, the best-known European name on the list, ranks sixth with 105 families.
Methodology in detail
The counting unit is the deduplicated patent family. Only the primary CPC class B23K9 is counted, meaning true arc welding without soldering and without laser. Corporate applicants are consolidated into one group across all name variants — for ITW that means Illinois Tool Works, Miller, Hobart, Bernard and Tregaskiss — and then checked for double counting. For context, patent families measure invention and IP activity, not revenue, market share or installed base. The metric also favours companies with a strong patenting culture; those relying more on trade secrets appear relatively under-counted.

This is what stock versus momentum looks like

The same data, two ways of cutting it — and a result that reverses.

What the chart shows
The annual count of new patent families in laser material processing for five suppliers, 2015 to 2023. Not the accumulated stock, but the inflow per year.
How to read it
The horizontal axis shows the earliest filing year, the vertical axis the number of new families in that year (0 to 150). Each line is one corporate group, labelled directly. A falling curve means declining filing activity, not the loss of existing rights. The dip at 2019 is again the coverage gap in the source.
Line chart of annual patent filings by five major laser suppliers. The Han's Laser curve drops sharply after 2020 while TRUMPF rises and leads from 2021 onward.
Source: Google Patents Public Data (Google & IFI CLAIMS Patent Services, 2026). Analysis by Vertumen, primary CPC B23K26, families by earliest filing year. · Open chart at full size
Finding
Summed across the whole period, Han's Laser leads with 660 families ahead of TRUMPF with 621 — so China leads on stock. Plot the same data over time and the picture flips. Han's Laser falls from its 2020 peak of 139 families to 28 in 2023, a drop of roughly 80 per cent, while TRUMPF climbs from 32 to 81 and has led every year since 2021. Both statements are correct; they answer different questions.
Methodology in detail
The counting unit is the distinctly counted patent family — not a sum across name variants, which double-counts families held by several corporate entities. Primary CPC class B23K26, laser material processing. Corporate applicants are consolidated across all name variants including non-Latin spellings. The aggregate sums 2015 to 2025, while the trend plots by earliest filing year and excludes 2024 and 2025 due to publication lag.

This is what a value-chain analysis looks like

Two technology layers as axes — and the question of which one a supplier actually occupies.

What the chart shows
Thirteen laser suppliers plotted by their patent families across two separate layers of the value chain: the laser source and laser processing.
How to read it
The horizontal axis counts patent families in laser processing, CPC B23K26, meaning the machine and process layer, from 0 to 800. The vertical axis counts families in the laser source, CPC H01S3, the light engine itself, from 0 to 120. Each dot is one company, its colour marking the country of headquarters (Germany, China, Japan, USA, Switzerland). Far right and low means machine builder; high and to the left means source maker; anyone away from both axes is active on both layers.
Scatter chart of thirteen laser suppliers. Most sit on either the laser-source axis or the laser-processing axis, with only TRUMPF present on both layers.
Source: Google Patents Public Data (Google & IFI CLAIMS Patent Services, 2026). Analysis by Vertumen, primary CPC H01S3 and B23K26, families 2015–2025. · Open chart at full size
Finding
Almost every supplier occupies exactly one layer. Raycus, IPG Photonics, Coherent and nLIGHT sit on the source axis; Han's Laser, DISCO, Amada and Bystronic on the processing axis. TRUMPF is the only company in the picture with meaningful presence on both. This is why a source maker like Raycus never appears in a machine ranking despite playing a substantial role in the market.
Methodology in detail
The counting unit is the distinctly counted patent family, covering 2015 to 2025, with both classes counted as primary. Corporate variants are consolidated per company. One clarification concerns the horizontal axis: B23K26 covers working by laser beam, not machine building in general — presses, punching and bending machines sit in B21D and B30B and are not included here, not even for TRUMPF. A disclosed limitation applies to the source axis: makers of lithography sources such as Gigaphoton and Cymer are absent because they serve chip lithography, a different market. That exclusion works company-wide rather than proportionally — anyone running an EUV driver-laser business themselves carries that share undivided inside their H01S3 figure. For TRUMPF, some of the 117 families likely relate to EUV lithography; separating them out would require a subclass breakdown that we have not performed.
What these figures do not say

Patent families measure invention and IP activity. They are not a proxy for revenue, market share or installed base, and they favour companies with a strong patenting culture over those relying on trade secrets. In a full report this layer therefore never stands alone but sits alongside financial metrics, product and specification data, policy and regional analysis.

Data basis and licence

All four analyses draw on Google Patents Public Data, accessed via BigQuery. The dataset is published by Google and based on data from IFI CLAIMS Patent Services; it is licensed under the Creative Commons licence CC BY 4.0. The analysis, class selection, corporate consolidation and presentation are the work of Vertumen.

The same methodology applied to your question

These extracts come from the laser and joining technology clusters. If you want to know what the picture looks like in your market, describe your question to us.