Building a Dataspace for Manufacturing as a Service in Factory-X

Abstract: One way to solve the challenge of small and medium-sized enterprise (SME) manufacturers of acquiring sufficient orders is by joining digital Manufacturing-as-a-Service (MaaS) platforms for on-demand manufacturing. However, joining such platforms brings about new challenges such as efficient quoting handling in the face of potentially low success rates and the need for high production quality for low lot sizes. Automating the complete interaction between manufacturers and MaaS platforms, from registering the manufacturer and its capabilities to handling incoming requests and managing offers, orders, and production quality reporting, helps to overcome these challenges. Thus, the increased number of requests can be handled efficiently, and the production quality can be maintained at a high level even for low lot sizes. This paper presents an architecture for automating the interaction and functional building blocks between manufacturers and MaaS platforms, along with a prototype implementation and evaluation of its effectiveness in addressing the challenges SME manufacturers are faced with.

• The paper introduces a robust dataspace framework that integrates semantic AAS models and modular MX-Port components to automate the end-to-end MaaS process chain.
• The paper demonstrates that automated supplier onboarding combined with ML-driven cost modeling effectively addresses challenges posed by heterogeneous legacy manufacturing data.
• The paper highlights practical implications for SMEs by streamlining RFQ handling, order execution, and digital quality reporting to enhance competitiveness in manufacturing.

Building a Dataspace for Manufacturing as a Service in Factory-X: An Expert Analysis

Context and Motivation

This paper presents a robust architecture and prototype for an industrial dataspace enabling Manufacturing-as-a-Service (MaaS) within the Factory-X ecosystem. Factory-X, as the technical flagship of the larger Manufacturing-X initiative, focuses on distributed, secure, and semantically interoperable data sharing to support resilient, sustainable, and competitive manufacturing. The authors address critical challenges for small- and medium-sized manufacturers (SMEs): low success rates and high effort in quoting, and the quality demands of low-lot-size production. Their approach is predicated on industry-standard semantic models and secure, automated data exchange.

Technical Architecture

Central to the solution is the integration of the Asset Administration Shell (AAS) for semantic interoperability and the MX-Port for secure, modular communication. The proposed system automates the MaaS process chain from supplier onboarding, standardized capability notification, search and matchmaking, automated quoting, order execution, through to digital quality reporting.

The architecture orchestrates three central stakeholder roles:

• Buyers initiate and specify requirements.
• Suppliers (often SMEs) provide manufacturing capabilities.
• Platform Applications mediate, matching supply/demand, and facilitating the automation of transactions.

The data exchange leverages standardized AAS submodels for all entities (factories, products, machines), using modular MX-Port implementations (e.g., Leo, Hercules) for discovery, access control, and secure inter-company exchange.

Scenario-Driven Digitalization

The paper defines and implements three primary interaction scenarios:

• Supplier Capability Notification: Automated, structured publishing of supplier capabilities using AAS submodels, enabling precise matchmaking.
• Search, Request, Offer, and Order: Automated, feature-based matchmaking of manufacturing capabilities, streamlined RFQ handling with ML-based feature recognition and cost estimation, and standardized communication using AAS-compliant artifacts.
• Order Execution and Quality Control: End-to-end digital workflow including CAM automation, inline quality data capture, and AAS-documented quality reports leveraging the concept of a "Quality Fingerprint" for statistical process monitoring.

The demonstrator covers all these process steps, providing evidence for seamless technical integration and automation capability.

Semantic Standardization and Automation

A salient innovation is the deployment of a capability modeler which transforms heterogeneous, partially unstructured legacy descriptions into standardized, machine-interpretable AAS submodels. The modeler utilizes both structured and textual data sources, and exposes its functionality via API and GUI, supporting both manual and automated supplier onboarding. This is tightly integrated with feature recognition and ML-driven cost modeling to provide automated, component-level quoting based on both historical and inferred manufacturing data.

For secure, interoperable connectivity, MX-Port implementations encapsulate adapters for legacy systems (MES/ERP), converters to AAS, standardized data gateways, access/usage control (including token-based and policy-driven authorization), and discovery mechanisms.

Numerical and Empirical Outcomes

While the paper does not report large-scale benchmarking with systematic quantitative figures, the prototype demonstrates that:

• The automation of quoting, order processing, and quality control is feasible for a typical SME with heterogeneous legacy data.
• Structured and automated supplier data onboarding accelerates the time to visibility on digital marketplaces.
• RFQ handling is streamlined, reducing bottlenecks in request/offer/order cycles.
• Machine-level quality monitoring and automated reporting support digital traceability and enable higher lot-specific quality assurance.

The adoption effort is weighted towards upfront semantic mapping, but ongoing operations are strongly automated and scalable.

Theoretical and Practical Implications

Technically, the paper underscores the viability and necessity of strict semantic alignment (AAS) and controlled data sharing (MX-Ports) in open, federated manufacturing ecosystems. It extends current paradigms by supporting fine-grained, feature-based search and quoting, and integrating real-time production and quality information into interoperable dataspace assets.

Practically, these capabilities enable SMEs to compete in digital marketplaces, optimizing asset utilization and reducing administrative overhead. The platform-centric approach paves the way for decentralized, resilient supply chains, which are particularly pertinent in volatile economic and ecological contexts.

A key observation is the current dependency on high data quality and consistent semantic mapping; limitations in onboarding effort and granular capability specification remain open challenges. Furthermore, the implications for IP-sensitive processes highlight the need for next-generation access and usage controls.

Future Directions

The approach is extensible across the manufacturing value chain, including assembly, surface treatment, and distributed supply networks. The authors propose further research into:

• Data-driven learning systems, leveraging aggregated, anonymized process data for operational optimization.
• The application of LLMs for automated data transformation and semantic mapping, reducing onboarding effort for new participants.
• Large-scale industrial evaluation to assess scalability, real-world acceptance, and economic impact in SMEs.
• Enhanced mechanisms for IP and compliance assurance in contract manufacturing.

Conclusion

This work delivers a comprehensive, technically mature reference architecture and prototype for federated MaaS in the context of Factory-X, leveraging AAS semantic models and modular, secure data exchange. The system enables end-to-end automation from capability publication through quoting, order execution, and quality control—addressing long-standing pain points in SME manufacturing integration with digital supply chains. Practical deployment depends on further validation, data mapping automation, and systemic adoption of semantic standards. These findings have direct implications for future industrial data spaces, collaborative AI systems, and resilient manufacturing networks.