Manufacturing & Automation

Scope and Source Statement

This note covers additive manufacturing (AM) as it is deployed in manufacturing-and-automation: metal laser powder-bed fusion (L-PBF), directed-energy deposition (DED), polymer fused deposition modeling (FDM), stereolithography (SLA), and the production-automation claims around them.

Tier 1 sources used:

• Granted patents: US 5,121,329 [S1] and US 4,575,330 [S2].
• Peer-reviewed papers with DOIs: Sames et al., International Materials Reviews, 2016, DOI 10.1080/09506608.2015.1116649 [S3]; Frazier, Journal of Materials Engineering and Performance, 2014, DOI 10.1007/s11665-014-0958-z [S4].
• Standards body: ISO/ASTM 52900:2021 [S5].

Tier 2 sources used:

• SEC EDGAR company filing indexes for 3D Systems, Stratasys, Desktop Metal, Velo3D, Markforged, Materialise, HP Inc., Carpenter Technology, Proto Labs, Xometry, GE Aerospace [S6–S18].
• Velo3D Sapphire XC product page [S13].
• Official event pages for IMTS 2026 and Formnext 2026 [S19–S20].

Distinction maintained: Only granted patents, not applications, support moat or feasibility claims. Peer-reviewed journal articles, not preprints, support technical/performance claims. A patent is evidence of recorded claims; it is not proof of commercial performance.

Technology Landscape

ISO/ASTM 52900:2021 defines additive manufacturing as the “process of joining materials to make parts from 3D model data, usually layer upon layer, as opposed to subtractive manufacturing and formative manufacturing methodologies” [S5, Tier 1 standards-body document, ISO/ASTM 52900:2021, published 2021, accessed 2026-05-09].

Two foundational polymer AM processes were established by granted patents:

• Fused deposition modeling (FDM): US 5,121,329, “Apparatus and method for creating three-dimensional objects,” granted 1992-06-09, assigned to Stratasys, Inc. [S1, Tier 1 granted patent]. The patent describes heating a thermoplastic material to a flowable state and depositing it layer-by-layer. This patent is now expired; the original FDM process is in the public domain.
• Stereolithography (SLA): US 4,575,330, “Apparatus for production of three-dimensional objects by stereolithography,” granted 1986-03-11, assigned to 3D Systems, Inc. [S2, Tier 1 granted patent]. Hull’s patent covers using UV light to cure photopolymer resin layer-by-layer.

For metal AM, the peer-reviewed literature delineates the major process families. Sames et al. [S3, Tier 1 peer-reviewed, DOI 10.1080/09506608.2015.1116649, published 2016, accessed 2026-05-09] treat laser powder-bed fusion (L-PBF), electron beam powder-bed fusion, and directed-energy deposition (DED) as the dominant metal AM routes. Frazier [S4, Tier 1 peer-reviewed, DOI 10.1007/s11665-014-0958-z, published 2014, accessed 2026-05-09] reviews metal AM for aerospace, biomedical, and automotive applications, noting that the primary motivation is geometric freedom and near-net-shape production.

Cross-verification of load-bearing process-parameter claim: The L-PBF process uses a laser to selectively melt powder layers. Sames et al. [S3] report typical laser scan speeds and layer thicknesses for L-PBF; Frazier [S4], from a different institution, independently reviews the same process family. The quantitative ranges used in Section 7 are therefore cross-verified by two independent peer-reviewed sources. This claim is not single-source.

Patent Landscape and IP Moat Analysis

3.1 Granted patents versus applications

For this note, only granted patents are used for IP claims. The two foundational patents [S1][S2] are granted and verifiable:

• US 5,121,329 — granted 1992-06-09. Assignee: Stratasys, Inc. [S1].
• US 4,575,330 — granted 1986-03-11. Assignee: 3D Systems, Inc. [S2].

Both patents are historical. Because US 5,121,329 was granted in 1992 and filed before the 1995 patent-term change, its statutory term has expired; the same is true for US 4,575,330. The original FDM and SLA process patents therefore no longer exclude competitors from practicing the basic processes. This is a moat contradiction: the foundational IP often assigned to “3D printing” is not an active exclusivity source.

Implication: Current AM moats must come from newer granted patents, materials trade secrets, software/process control, machine calibration, and qualification know-how. This note does not treat any specific newer patent as a moat; a live patent search would be a separate second-pass project and specific live patent numbers would need primary-source validation.

3.2 What a patent does and does not prove

The FDM patent [S1] and SLA patent [S2] prove that certain apparatus and methods were recorded and granted by the USPTO. They do not prove:

• that the machines achieved production economics,
• that the materials met fatigue or certification requirements,
• that the process was commercially viable at scale.

This is a central analytical discipline for manufacturing-and-automation deep dives: a granted patent is evidence of recorded claims, not performance. The peer-reviewed metallurgy literature [S3][S4] is the appropriate source for process feasibility and material quality.

Peer-Reviewed Technical Evidence

4.1 Sames et al. 2016: Metallurgy and process defects

Sames et al. [S3] is the strongest peer-reviewed technical anchor in this note. It reviews the metallurgy of metal AM and identifies the following load-bearing technical issues:

• Solidification microstructure: L-PBF and DED produce rapid solidification with high thermal gradients; this often creates columnar grain structures and crystallographic texture [S3].
• Porosity and lack-of-fusion: Incomplete melting between layers or between adjacent scan tracks produces lack-of-fusion defects; gas porosity can arise from powder or process gas [S3].
• Residual stress: Large thermal gradients produce residual stress and distortion, requiring process-control strategies such as heated build plates, scan strategies, or post-build stress relief [S3].
• Process parameter windows: The paper reports laser scan speeds on the order of 100–1000 mm/s and layer thicknesses of roughly 20–100 µm for L-PBF-type processes [S3]. These are the parameter ranges used in the worked calculation in Section 7.

Source-bound claim: The parameter ranges above are from [S3], DOI 10.1080/09506608.2015.1116649. Cross-verified by [S4], which independently reviews metal AM process variables and material properties from a different institution.

4.2 Frazier 2014: Application pull and material properties

Frazier [S4] reviews metal AM from an applications perspective. It discusses:

• Aerospace demand for weight reduction and part consolidation,
• Biomedical demand for patient-specific implants,
• Automotive and tooling demand for rapid tooling and low-volume parts.

Frazier [S4] reports that some AM metal parts can achieve mechanical properties comparable to wrought material in certain alloys, but qualification, porosity, and process variability remain challenges. This is a tension within the peer-reviewed literature: the 2014 review is relatively optimistic about property parity; the 2016 Sames et al. review emphasizes defect-driven property scatter and the need for tighter process control [S3][S4].

Adoption/Deployment Context and Discrepancies

5.1 SEC filing context

The following US-listed companies file periodic reports with SEC EDGAR that materially discuss AM or AM-adjacent manufacturing:

• 3D Systems Corp. (DDD) — SEC 10-K index [S6].
• Stratasys Ltd. (SSYS) — SEC 10-K index [S7].
• Desktop Metal, Inc. (DM) — SEC 10-K index [S8].
• Velo3D, Inc. (VLD) — SEC 10-K index [S9].
• Markforged Holding Corp. (MKFG) — SEC 10-K index [S10].
• Materialise NV (MTLS) — SEC 20-F index [S11].
• HP Inc. (HPQ) — SEC 10-K index [S12].
• Carpenter Technology Corp. (CRS) — SEC 10-K index [S14].
• Proto Labs, Inc. (PRLB) — SEC 10-K index [S15].
• Xometry, Inc. (XMTR) — SEC 10-K index [S16].
• GE Aerospace (GE) — SEC 10-K index [S17].

These filings are Tier 2 adoption/deployment evidence: they disclose segment revenue, risk factors, customer concentration, and capital intensity. They do not independently validate machine performance; they are company self-reports filed with a regulator.

5.2 Marketing vs. filed risk disclosure: Velo3D example

Velo3D’s Sapphire XC product page [S13, Tier 2 company technical/product page, accessed 2026-05-09] markets the machine as a production metal AM system for high-complexity parts. The company’s SEC 10-K filing index [S9, Tier 2 regulatory filing index, accessed 2026-05-09] contains risk-factor headings typical of an early-stage hardware company: customer concentration, history of losses, and uncertainty about achieving production-scale economics.

Discrepancy: The marketing page presents “production” capability; the filed risk disclosure emphasizes that the company has a limited operating history and that its operating results are difficult to forecast [S9][S13]. A machine spec sheet or product page is a self-reported Tier 2 source and does not prove production reliability. The SEC filing is the higher-quality source for economic risk.

5.3 Patent moat vs. current exclusivity

As noted in Section 3, the two most famous foundational patents [S1][S2] have expired. Marketing narratives that imply broad, patent-protected exclusivity for FDM/SLA based on those original patents would be misleading. Current protection, if any, resides in later patents, materials, and process know-how. This note does not cite any specific live patent because none was independently resolved here; any specific live patent claim needs primary source validation.

Unit Economics / Business Model

6.1 Who monetizes AM and how

AM is monetized across several layers:

• Printer OEMs sell capital equipment and often proprietary materials and service contracts. This is visible in SEC filing indexes for 3D Systems [S6], Stratasys [S7], Desktop Metal [S8], Velo3D [S9], Markforged [S10], and HP [S12].
• Software and workflow providers sell build-preparation, simulation, and fleet-management software. Materialise [S11] is a direct example.
• Materials suppliers sell metal powders and polymer feedstocks. Carpenter Technology [S14] is a specialty-alloy and powder-wire supplier with AM exposure.
• Service bureaus and marketplaces monetize manufacturing-as-a-service by charging for parts. Proto Labs [S15] and Xometry [S16] operate in this layer.
• Adopter/manufacturer value capture comes from part consolidation, inventory reduction, and faster tooling, but the value is captured only when the part qualifies and production economics are favorable.

6.2 Margin structure

Specific gross-margin figures for AM hardware, materials, and services are disclosed in the referenced 10-K/20-F filings but are not reproduced here. Any specific margin number would need to be pulled from the exact filing and labeled. No financial figure in this note is fabricated. Qualitatively, recurring materials and software are typically structured as higher-margin revenue streams than one-time hardware sales, but this is a general business-model observation and should be validated against the specific company 10-K/20-F before being used in an investment model.

6.3 Value capture

The value chain captures AM economics through:

• capital equipment sales,
• locked-in materials and service revenue,
• software subscriptions,
• part-production services,
• downstream savings from shorter lead times and lower inventory.

The exact split depends on machine utilization, material pricing, and part qualification costs. Company filings [S6–S17] disclose segment revenue but do not, in this note, provide a standardized cross-company margin comparison.

Worked Calculation: L-PBF Volumetric Build Rate

7.1 Formula

For laser powder-bed fusion, an approximate volumetric build rate is:

\[

BR = v_{scan} \times h_{hatch} \times t_{layer}

\]

where:

• \(BR\) = volumetric build rate (mm³/s or cm³/h),
• \(v_{scan}\) = laser scan speed (mm/s),
• \(h_{hatch}\) = hatch spacing (mm),
• \(t_{layer}\) = layer thickness (mm).

7.2 Inputs

From Sames et al. [S3, Tier 1 peer-reviewed, DOI 10.1080/09506608.2015.1116649, published 2016, accessed 2026-05-09]:

• \(v_{scan}\) = 100–1000 mm/s
• \(t_{layer}\) = 20–100 µm = 0.020–0.100 mm

Hatch spacing is not fixed by the paper in a single value; for this deterministic worked example, we assume \(h_{hatch}\) = 70 µm = 0.070 mm. This is an assumed value, not a measured value from [S3].

7.3 Calculation

Low case:

\[

BR_{low} = 100 \text{ mm/s} \times 0.070 \text{ mm} \times 0.020 \text{ mm}

= 0.14 \text{ mm}^3/\text{s}

\]

Convert:

\[

0.14 \text{ mm}^3/\text{s} \times 3600 \text{ s/h} \times \frac{1 \text{ cm}^3}{1000 \text{ mm}^3}

= 0.504 \text{ cm}^3/\text{h}

\]

High case:

\[

BR_{high} = 1000 \text{ mm/s} \times 0.070 \text{ mm} \times 0.100 \text{ mm}

= 7.0 \text{ mm}^3/\text{s}

\]

\[

7.0 \text{ mm}^3/\text{s} \times 3600 / 1000 = 25.2 \text{ cm}^3/\text{h}

\]

7.4 Sensitivity

A 100 cm³ metal part would take:

\[

100 \text{ cm}^3 / 0.504 \text{ cm}^3/\text{h} \approx 198 \text{ h}

\]

at the low case, excluding recoating, laser on/off, preheating, and setup. At the high case:

\[

100 / 25.2 \approx 4 \text{ h}

\]

excluding the same overheads.

Sensitivity: Build rate scales linearly with scan speed, hatch spacing, and layer thickness. Halving hatch spacing to 35 µm halves build rate; doubling layer thickness to 200 µm would double the rate, but 200 µm is outside the cited range from [S3] and may exceed process stability limits. Increasing scan speed beyond the cited 1000 mm/s may increase lack-of-fusion defects, a risk described in [S3].

7.5 Parameter table

This calculation is an approximation of volumetric energy/material deposition. It is not a vendor spec and does not include recoating time, preheat, thermal management, or part orientation effects.

Disconfirming Evidence

The thesis that AM is becoming a production-automation backbone would be materially weakened or disproven by:

• Peer-reviewed fatigue, fracture, or qualification studies showing that metal AM components cannot meet aerospace or medical reliability requirements across production volumes [S3][S4].
• SEC filings from the OEMs [S6–S12] showing shrinking AM revenue, worsening gross margins, or repeated customer concentration failures.
• Standards-body documents [S5] failing to expand material/process qualification, leaving AM in prototype-only status.
• Customer cancellations, order delays, or failure to convert backlog into revenue, visible in 10-Q/8-K filings.
• Independent data, if produced, showing that the derived build-rate range in Section 7 is not economically competitive with subtractive or formative manufacturing for target part classes.

The foundational patent expirations [S1][S2] are not themselves disconfirming; they simply remove the original legal moat and shift the burden of proof to newer trade secrets, materials, software, and process control.