The Integrated Shipyard: Leveraging AI and Digital Twins to Mitigate Labor Shortages and Data Silos
The MOC
By
Cailyn Yong
June 8, 2026
Introduction
The need to modernize U.S. shipbuilding has grown increasingly evident. Both naval and commercial sectors require complex, customized vessels, but the supporting workforce and data infrastructure remain outdated. Consequently, the United States has lost its competitive edge in the global shipbuilding industry. Modernization of domestic shipyards is therefore essential to improve workflows and achieve greater speed, precision, and consistency.
However, modernization efforts are frequently constrained by two primary factors: a shortage of skilled naval architects and engineers, and the absence of an integrated operating system to manage overall yard productivity. Technical knowledge required for ship construction is often dispersed across spreadsheets, outdated design files, and among veteran engineers nearing retirement. The lack of a unified operating system prevents automatic updates across teams, resulting in manual processes and inefficient communication. To overcome these critical bottlenecks, this paper argues that combining artificial intelligence (AI) with digital twin technologies presents a viable, integrated solution for modern shipyards.
While digital twin technology is already utilized in the maritime industry, its current applications focus heavily on real-time operational modeling and manufacturing logistics. As highlighted in the Center for Maritime Strategy’s Pier Review report, shipyards such as Fincantieri typically deploy digital twins as virtual replications of a vessel to integrate design, simulation, and operational tracking to manage day-to-day yard productivity. However, this paper explores an expanded, less-utilized capability of combining digital twin infrastructure with generative and predictive AI. By training machine learning models on deep historical data—such as past contract histories, design archives, and legacy engineering data—shipyards can unlock predictive insights before a ship ever enters production.
This paper examines how adopting these technologies can address critical bottlenecks across three phases of modern shipbuilding: design and bidding, parallel procurement and compliance, and physical assembly with cross-yard coordination. The paper concludes by outlining the requirements for successful technological adoption and the risks associated with inaction.
Section 1: The Design and Bidding Phase
Manual Extrapolation in Custom Vessel Design
As domestic shipyards rarely build repetitive vessel classes, each new contract starts close to scratch. Field interviews with shipyard executives consistently identify a primary bottleneck: adapting legacy designs for new customers. Especially during the sales engineering phase, a shipyard’s competitiveness comes from its ability to develop new products, indicating that strong in-house design capability is crucial. Yet this becomes harder due to the severe shortage of U.S. naval architects. While international competitors heavily subsidize maritime education, the United States suffers from a limited number of specialized maritime degree programs, historical wage gaps, and an accelerating retirements. Consequently, domestic shipyards are forced to rely on third-party engineering firms, stretching timelines and driving up design costs.
For example, a yard may hold a proven blueprint for a 1,000-foot general carrier, while a new customer requests a 300-foot vessel carrying entirely different cargo. Engineering teams cannot scale the template linearly. Even with consistent structural class rules, modifying the vessel’s length and functional purpose requires manual, case-by-case extrapolation from older drawings often kept in formats that resist automated reuse. Without enough in-house designers to create basic designs, this directly affects the shipyard’s ability to secure sustainable growth and profitability.
AI-Assisted Design Extrapolation
With AI, shipyards can speed up this basic design process. A shipyard holds decades of historical data: contract histories, 3D design files, regulatory submittals, and material schedules. The opportunity here is to train an AI model on this archive—including the shipyard’s previous trial-and-error—so the yard can query this data context when a new contract arrives.
When a naval architect receives a new contract for a specialized vessel, such as a 300-foot ship, the requirements—vessel dimensions, cargo type, regulatory class, and intended route—are provided to the AI. The model searches the shipyard’s historical records for analogous prior builds and generates a modern design based on the closest precedent. It accounts for how dimensional changes affect frame layouts, structural rules, and stability calculations, removing the need for manual extrapolation from older drawings.
The AI generates outputs in standard CAD file formats, allowing the resulting arrangements, drawings, and component schedules to be imported directly into widely used design software such as SSI’s ShipConstructor, AVEVA Marine, Siemens NX, or Dassault CATIA. This provides architects with a functional starting point: a 3D model informed by relevant historical designs, customized to the new contract, and ready for refinement into a final proposal. By producing compliant, contract-specific designs in days rather than weeks, shipyards can commit to bids earlier and initiate procurement while design finalization is ongoing.
Section 2: The Parallel Procurement and Compliance Phase
Regulatory Gap Analysis Under Parallel Procurement
After a contract is secured, design and supply chain processes become decoupled. Typical shipbuilding projects span twelve to eighteen months, and to reduce material price inflation, shipyards often conduct design and procurement activities in parallel. However, this approach is frequently disrupted by regulatory gap analysis. An interview with a Director of Engineering from Fincantieri Marine Group revealed that when U.S. shipyards adopt European vessel designs, engineers must manually modify blueprints to comply with U.S. Coast Guard (USCG) and American Bureau of Shipping (ABS) regulations. The USCG imposes domestic requirements that are significantly stricter than international class rules, and even minor differences can lead to substantial layout changes.
For instance, international regulations may permit 700-millimeter structural passageways on conventional cargo vessels, whereas the USCG requires 900 millimeters for certain passenger and domestic vessel layouts. The increased width requires moving bulkheads outward, which can cause conflicts with pre-planned electrical and piping systems. Material standards present additional challenges. European designs use metric measurements that do not align precisely with U.S. imperial manufacturing tolerances. A senior shipbuilding professional noted that when vessel designs developed under one regulatory and industrial framework are adapted to another, material equivalencies, tolerances, and procurement availability may not translate perfectly. In some cases, these discrepancies can lead to heavier material selections or cumulative design changes, increasing vessel weight, affecting the center of gravity, and reducing stability margins
Automated Compliance and Clash Detection
An AI-driven compliance layer can identify regulatory and material conflicts during the design process, rather than after procurement decisions have been made. AI models trained on USCG and ABS rule libraries can operate alongside parametric design platforms, serving as a compliance translation layer.
When a foreign design becomes introduced, the AI compliance layer automatically conducts gap analysis. It identifies discrepancies, such as the 700-millimeter versus 900-millimeter passageway requirement, and determines which structural, electrical, and piping components are affected. While CAD platforms detect geometric clashes once they are present in the model, the AI layer anticipates cascade effects before they occur. Similarly, the AI system links metric-to-imperial material substitutions to existing stability calculations, enabling shipyards to assess the impact of procurement changes on top-side weight prior to finalizing purchase orders. Early resolution of these conflicts secures both the build schedule and capital invested in long-lead materials, keeping alignment between design and procurement processes.
Section 3: The Physical Assembly and Cross-Yard Alliance Phase
Cross-Yard Coordination and Module Mismatch
Once design compliance is achieved and materials are procured, the primary bottleneck shifts to the physical shipyard. Yard capacity is limited by fixed physical layouts; facilities cannot immediately expand their footprint or add new drydocks. Bottlenecks between fabrication and assembly can delay schedules, and shipyards in the Upper Midwest of the United States such as Marinette Marine may lose up to six months of outdoor work due to winter conditions.
To address demand within fixed layouts, shipyards often form cross-yard manufacturing alliances. Geographically dispersed facilities fabricate distinct submodules, which are then transported by truck or barge to a central drydock for final assembly. Regional partnerships, such as the Fourth Coast Shipbuilding Alliance (Fincantieri Marine Group, Fraser Shipyards, and Donjon Marine), exemplify this approach. However, this process requires precise alignment of submodules upon assembly.
When submodules do not align correctly, the primary cause is often due to the absence of a cohesive, real-time project information system and comprehensive communication records. Revisions, corrected measurements, or quick fixes are communicated only through manual exchanges such as emails, phone calls, and re-sent files. As these compound, shipyards may receive completed submodules that are significantly out of spec or slightly misaligned. These discrepancies often go undetected until final assembly at the drydock, resulting in costly rework and reconstruction.
Synchronized Digital Twin Across Yards
A shared digital twin provides shipyards with a live, continuously updated 3D model of the vessel and its modules. An AI layer can integrate meeting transcripts, calls, and email threads, associating them with specific design revisions or material orders. This ensures that every change within the digital twin is linked to the corresponding discussion or decision, allowing any yard to trace modifications back to their source rather than relying on memory or fragmented records.
When this shared state becomes integrated with real-time data—such as panel line availability, paint bay capacity, and local weather forecasts—the AI system can further support build-schedule coordination. This information enables the development of algorithms that optimize the movement of sub-assemblies across the alliance, interfacing with existing design and production planning systems.
These technological layers transform fragmented and delayed coordination into a unified source of truth, enabling an alliance of separate shipyards to operate with the precision of a single facility.
Implementation Risks and Adoption Framework
The AI is only as good as the data it learns from, which means shipyards must feed the model the full archive of past designs and contracts. The problem is that this data is also a shipyard’s most valuable and sensitive asset. That is what makes adoption difficult, and it comes down to several main issues.
Data Security and Sovereignty
The main limitation is where both the data and the AI models are stored. Ship designs, particularly those related to naval or dual-use vessels, are sensitive intellectual property. Transmitting such data to a shared cloud-based model poses unacceptable risks for most shipyards, as proprietary hull forms and regulatory details could be exposed or incorporated into models accessible to competitors. Thus, the system should operate on a shipyard’s local infrastructure, utilizing locally managed, open-weight models fine-tuned in-house. This approach ensures that a shipyard’s data is used exclusively for its own system and never leaves its premises.
Data Integrity
To ensure high-quality outputs from any AI model, input data must be thoroughly curated and standardized. Decades of contract histories, 3D files, and material schedules often exist in inconsistent formats and incomplete records. A model trained on contradictory or flawed data will replicate these issues. Therefore, validating and standardizing the data archive is a critical initial step.
Return on Investment and Phased Adoption
Shipyards typically operate with narrow profit margins and are unlikely to invest capital without a clear return on investment. Adoption of modern technologies is most effective when demonstrated through targeted pilot projects focused on specific, measurable outcomes, such as reduced basic-design cycle times or decreased late-stage rework. Since shipyards cannot suspend production to implement system revisions, technology rollout must be incremental. Implementation should begin where leverage is greatest and risk is lowest, typically in the design and bidding phase, before progressing to procurement and compliance, and ultimately cross-yard assembly.
Conclusion
The integration of AI and digital twin technology within the maritime industrial base has become imperative, shifting the focus from whether to adopt these innovations to how rapidly and responsibly they can be implemented. Continued dependence on manual gap analysis and tracking only widens the competitive gap with global shipbuilders, while spatial clashes, weight errors, and late-stage change orders continue to consume capital. The most significant impact is the loss of workforce expertise, as the retirement of veteran engineers leaves behind decades of troubleshooting knowledge.
While technology cannot replace skilled laborers such as welders and pipefitters, a unified, secure digital framework serves as a force multiplier for the existing workforce. It extends the capabilities of labor-short shipyards and preserves the expertise of experienced engineers. The recommended path forward involves integrating systems that retain proprietary data within the shipyard while enhancing speed, precision, and coordination. To maintain maritime readiness and strengthen U.S.’s maritime capability on the global stage, domestic shipbuilding must standardize its data architecture, secure it in accordance with its own regulatory requirements, and implement the integrated shipyard model.
Cailyn Yong is the founder of Forge, which develops AI and digital twin systems for shipyards, and a graduate of NYU. Her work on the maritime industrial base draws on ongoing field interviews with domestic shipyard operators and international naval architects.
The views expressed in this piece are the sole opinions of the author and do not necessarily reflect those of the Center for Maritime Strategy or other institutions listed.
Center for Maritime Strategy
By Cailyn Yong
Introduction
The need to modernize U.S. shipbuilding has grown increasingly evident. Both naval and commercial sectors require complex, customized vessels, but the supporting workforce and data infrastructure remain outdated. Consequently, the United States has lost its competitive edge in the global shipbuilding industry. Modernization of domestic shipyards is therefore essential to improve workflows and achieve greater speed, precision, and consistency.
However, modernization efforts are frequently constrained by two primary factors: a shortage of skilled naval architects and engineers, and the absence of an integrated operating system to manage overall yard productivity. Technical knowledge required for ship construction is often dispersed across spreadsheets, outdated design files, and among veteran engineers nearing retirement. The lack of a unified operating system prevents automatic updates across teams, resulting in manual processes and inefficient communication. To overcome these critical bottlenecks, this paper argues that combining artificial intelligence (AI) with digital twin technologies presents a viable, integrated solution for modern shipyards.
While digital twin technology is already utilized in the maritime industry, its current applications focus heavily on real-time operational modeling and manufacturing logistics. As highlighted in the Center for Maritime Strategy’s Pier Review report, shipyards such as Fincantieri typically deploy digital twins as virtual replications of a vessel to integrate design, simulation, and operational tracking to manage day-to-day yard productivity. However, this paper explores an expanded, less-utilized capability of combining digital twin infrastructure with generative and predictive AI. By training machine learning models on deep historical data—such as past contract histories, design archives, and legacy engineering data—shipyards can unlock predictive insights before a ship ever enters production.
This paper examines how adopting these technologies can address critical bottlenecks across three phases of modern shipbuilding: design and bidding, parallel procurement and compliance, and physical assembly with cross-yard coordination. The paper concludes by outlining the requirements for successful technological adoption and the risks associated with inaction.
Section 1: The Design and Bidding Phase
Manual Extrapolation in Custom Vessel Design
As domestic shipyards rarely build repetitive vessel classes, each new contract starts close to scratch. Field interviews with shipyard executives consistently identify a primary bottleneck: adapting legacy designs for new customers. Especially during the sales engineering phase, a shipyard’s competitiveness comes from its ability to develop new products, indicating that strong in-house design capability is crucial. Yet this becomes harder due to the severe shortage of U.S. naval architects. While international competitors heavily subsidize maritime education, the United States suffers from a limited number of specialized maritime degree programs, historical wage gaps, and an accelerating retirements. Consequently, domestic shipyards are forced to rely on third-party engineering firms, stretching timelines and driving up design costs.
For example, a yard may hold a proven blueprint for a 1,000-foot general carrier, while a new customer requests a 300-foot vessel carrying entirely different cargo. Engineering teams cannot scale the template linearly. Even with consistent structural class rules, modifying the vessel’s length and functional purpose requires manual, case-by-case extrapolation from older drawings often kept in formats that resist automated reuse. Without enough in-house designers to create basic designs, this directly affects the shipyard’s ability to secure sustainable growth and profitability.
AI-Assisted Design Extrapolation
With AI, shipyards can speed up this basic design process. A shipyard holds decades of historical data: contract histories, 3D design files, regulatory submittals, and material schedules. The opportunity here is to train an AI model on this archive—including the shipyard’s previous trial-and-error—so the yard can query this data context when a new contract arrives.
When a naval architect receives a new contract for a specialized vessel, such as a 300-foot ship, the requirements—vessel dimensions, cargo type, regulatory class, and intended route—are provided to the AI. The model searches the shipyard’s historical records for analogous prior builds and generates a modern design based on the closest precedent. It accounts for how dimensional changes affect frame layouts, structural rules, and stability calculations, removing the need for manual extrapolation from older drawings.
The AI generates outputs in standard CAD file formats, allowing the resulting arrangements, drawings, and component schedules to be imported directly into widely used design software such as SSI’s ShipConstructor, AVEVA Marine, Siemens NX, or Dassault CATIA. This provides architects with a functional starting point: a 3D model informed by relevant historical designs, customized to the new contract, and ready for refinement into a final proposal. By producing compliant, contract-specific designs in days rather than weeks, shipyards can commit to bids earlier and initiate procurement while design finalization is ongoing.
Section 2: The Parallel Procurement and Compliance Phase
Regulatory Gap Analysis Under Parallel Procurement
After a contract is secured, design and supply chain processes become decoupled. Typical shipbuilding projects span twelve to eighteen months, and to reduce material price inflation, shipyards often conduct design and procurement activities in parallel. However, this approach is frequently disrupted by regulatory gap analysis. An interview with a Director of Engineering from Fincantieri Marine Group revealed that when U.S. shipyards adopt European vessel designs, engineers must manually modify blueprints to comply with U.S. Coast Guard (USCG) and American Bureau of Shipping (ABS) regulations. The USCG imposes domestic requirements that are significantly stricter than international class rules, and even minor differences can lead to substantial layout changes.
For instance, international regulations may permit 700-millimeter structural passageways on conventional cargo vessels, whereas the USCG requires 900 millimeters for certain passenger and domestic vessel layouts. The increased width requires moving bulkheads outward, which can cause conflicts with pre-planned electrical and piping systems. Material standards present additional challenges. European designs use metric measurements that do not align precisely with U.S. imperial manufacturing tolerances. A senior shipbuilding professional noted that when vessel designs developed under one regulatory and industrial framework are adapted to another, material equivalencies, tolerances, and procurement availability may not translate perfectly. In some cases, these discrepancies can lead to heavier material selections or cumulative design changes, increasing vessel weight, affecting the center of gravity, and reducing stability margins
Automated Compliance and Clash Detection
An AI-driven compliance layer can identify regulatory and material conflicts during the design process, rather than after procurement decisions have been made. AI models trained on USCG and ABS rule libraries can operate alongside parametric design platforms, serving as a compliance translation layer.
When a foreign design becomes introduced, the AI compliance layer automatically conducts gap analysis. It identifies discrepancies, such as the 700-millimeter versus 900-millimeter passageway requirement, and determines which structural, electrical, and piping components are affected. While CAD platforms detect geometric clashes once they are present in the model, the AI layer anticipates cascade effects before they occur. Similarly, the AI system links metric-to-imperial material substitutions to existing stability calculations, enabling shipyards to assess the impact of procurement changes on top-side weight prior to finalizing purchase orders. Early resolution of these conflicts secures both the build schedule and capital invested in long-lead materials, keeping alignment between design and procurement processes.
Section 3: The Physical Assembly and Cross-Yard Alliance Phase
Cross-Yard Coordination and Module Mismatch
Once design compliance is achieved and materials are procured, the primary bottleneck shifts to the physical shipyard. Yard capacity is limited by fixed physical layouts; facilities cannot immediately expand their footprint or add new drydocks. Bottlenecks between fabrication and assembly can delay schedules, and shipyards in the Upper Midwest of the United States such as Marinette Marine may lose up to six months of outdoor work due to winter conditions.
To address demand within fixed layouts, shipyards often form cross-yard manufacturing alliances. Geographically dispersed facilities fabricate distinct submodules, which are then transported by truck or barge to a central drydock for final assembly. Regional partnerships, such as the Fourth Coast Shipbuilding Alliance (Fincantieri Marine Group, Fraser Shipyards, and Donjon Marine), exemplify this approach. However, this process requires precise alignment of submodules upon assembly.
When submodules do not align correctly, the primary cause is often due to the absence of a cohesive, real-time project information system and comprehensive communication records. Revisions, corrected measurements, or quick fixes are communicated only through manual exchanges such as emails, phone calls, and re-sent files. As these compound, shipyards may receive completed submodules that are significantly out of spec or slightly misaligned. These discrepancies often go undetected until final assembly at the drydock, resulting in costly rework and reconstruction.
Synchronized Digital Twin Across Yards
A shared digital twin provides shipyards with a live, continuously updated 3D model of the vessel and its modules. An AI layer can integrate meeting transcripts, calls, and email threads, associating them with specific design revisions or material orders. This ensures that every change within the digital twin is linked to the corresponding discussion or decision, allowing any yard to trace modifications back to their source rather than relying on memory or fragmented records.
When this shared state becomes integrated with real-time data—such as panel line availability, paint bay capacity, and local weather forecasts—the AI system can further support build-schedule coordination. This information enables the development of algorithms that optimize the movement of sub-assemblies across the alliance, interfacing with existing design and production planning systems.
These technological layers transform fragmented and delayed coordination into a unified source of truth, enabling an alliance of separate shipyards to operate with the precision of a single facility.
Implementation Risks and Adoption Framework
The AI is only as good as the data it learns from, which means shipyards must feed the model the full archive of past designs and contracts. The problem is that this data is also a shipyard’s most valuable and sensitive asset. That is what makes adoption difficult, and it comes down to several main issues.
Data Security and Sovereignty
The main limitation is where both the data and the AI models are stored. Ship designs, particularly those related to naval or dual-use vessels, are sensitive intellectual property. Transmitting such data to a shared cloud-based model poses unacceptable risks for most shipyards, as proprietary hull forms and regulatory details could be exposed or incorporated into models accessible to competitors. Thus, the system should operate on a shipyard’s local infrastructure, utilizing locally managed, open-weight models fine-tuned in-house. This approach ensures that a shipyard’s data is used exclusively for its own system and never leaves its premises.
Data Integrity
To ensure high-quality outputs from any AI model, input data must be thoroughly curated and standardized. Decades of contract histories, 3D files, and material schedules often exist in inconsistent formats and incomplete records. A model trained on contradictory or flawed data will replicate these issues. Therefore, validating and standardizing the data archive is a critical initial step.
Return on Investment and Phased Adoption
Shipyards typically operate with narrow profit margins and are unlikely to invest capital without a clear return on investment. Adoption of modern technologies is most effective when demonstrated through targeted pilot projects focused on specific, measurable outcomes, such as reduced basic-design cycle times or decreased late-stage rework. Since shipyards cannot suspend production to implement system revisions, technology rollout must be incremental. Implementation should begin where leverage is greatest and risk is lowest, typically in the design and bidding phase, before progressing to procurement and compliance, and ultimately cross-yard assembly.
Conclusion
The integration of AI and digital twin technology within the maritime industrial base has become imperative, shifting the focus from whether to adopt these innovations to how rapidly and responsibly they can be implemented. Continued dependence on manual gap analysis and tracking only widens the competitive gap with global shipbuilders, while spatial clashes, weight errors, and late-stage change orders continue to consume capital. The most significant impact is the loss of workforce expertise, as the retirement of veteran engineers leaves behind decades of troubleshooting knowledge.
While technology cannot replace skilled laborers such as welders and pipefitters, a unified, secure digital framework serves as a force multiplier for the existing workforce. It extends the capabilities of labor-short shipyards and preserves the expertise of experienced engineers. The recommended path forward involves integrating systems that retain proprietary data within the shipyard while enhancing speed, precision, and coordination. To maintain maritime readiness and strengthen U.S.’s maritime capability on the global stage, domestic shipbuilding must standardize its data architecture, secure it in accordance with its own regulatory requirements, and implement the integrated shipyard model.
Cailyn Yong is the founder of Forge, which develops AI and digital twin systems for shipyards, and a graduate of NYU. Her work on the maritime industrial base draws on ongoing field interviews with domestic shipyard operators and international naval architects.
The views expressed in this piece are the sole opinions of the author and do not necessarily reflect those of the Center for Maritime Strategy or other institutions listed.