Systems Engineering Enters New Era as SysML v2 Standard Gains Final Approval, AI Integration

Exploring The Ne

xt Frontier of Systems Engineering: SysMLv2

OMG formally adopts next-generation modeling language after eight-year development; Dassault Systèmes leads commercial implementation as AI capabilities promise to ease steep learning curve

The Object Management Group (OMG) formally adopted the Systems Modeling Language (SysML) version 2.0 specification on June 30, 2025, marking a watershed moment for model-based systems engineering after eight years of development. The formal specification was published in September 2025, enabling commercial tool vendors to begin full-scale implementation of a standard that industry leaders believe will fundamentally transform how engineers design complex systems.

The adoption package includes three interconnected specifications: SysML v2.0, the Kernel Modeling Language (KerML) v1.0 that provides its semantic foundation, and the Systems Modeling Application Programming Interface (API) and Services v1.0 that enables interoperability across tools.

First Mover Advantage

Dassault Systèmes is positioned to dominate early adoption, having announced during an October 2024 technical webinar that it will release comprehensive SysML v2 support in its Cameo Systems Modeler portfolio with its 2026x release, scheduled for late November or early December 2024. The company has been running early access programs since the beta specification phase, giving it significant lead time over competitors.

"We definitely have the most powerful solution there in terms of features," said Andrew Lytvynov, SysML v2 Implementation Lead at Dassault Systèmes. "You don't need five tools to do MBSE with SysML v2. You just use Cameo for this."

Other major vendors including IBM, PTC, Siemens, and multiple open-source initiatives have also committed to SysML v2 support. IBM unveiled its Rhapsody Systems Engineering cloud platform in September 2024, leveraging SysML v2 as a foundational technology.

Breaking with the Past

SysML v2 represents a complete architectural departure from its 15-year-old predecessor. Unlike SysML v1, which was built on the Unified Modeling Language (UML) designed for software engineers, v2 is constructed on the Kernel Modeling Language (KerML), which provides formal semantics and was designed specifically for systems engineering.

The most visible innovation is dual syntax support: engineers can work with models using traditional graphical notation (boxes and lines) or a new textual syntax, with full bidirectional synchronization between the two representations. Dassault claims to be the only vendor currently offering true synchronization between syntaxes.

More fundamentally, SysML v2 introduces a standardized expression language that enables automated analysis and verification. Unlike v1, where expressions were tool-specific and non-standard, v2 expressions are universally understood across implementations, enabling engineers to define mathematical relationships, constraints, and dependencies that tools can automatically evaluate.

Requirements can now include formal constraints that evaluate to true or false, moving beyond purely textual descriptions to enable automated verification—a game-changer for complex systems with thousands of requirements.

The Cost of Progress

However, the transition presents substantial challenges. Thomas Baniszewski, Teamwork Cloud Product Manager at Dassault, revealed a striking statistic: SysML v2 models are approximately ten times larger than equivalent v1 models, requiring significant infrastructure upgrades for organizations using server-based collaboration.

"All of these fancy language features like formal grounding, ontological semantics, high degree of expressiveness—it actually comes with a cost that resides at persistency of such models," Baniszewski explained.

The learning curve is equally steep. Lytvynov acknowledged that while the language is more consistent internally, "it's more powerful than v1 but at the same time it means you will need to learn a little more as an expert." Engineers must master not only new syntax but entirely different modeling paradigms, including usage-focused modeling concepts absent from v1.

A March 2024 Department of Defense technical report on SysML v1-to-v2 migration emphasizes that transformation is "lossy"—information will be lost—and requires substantial post-processing to leverage v2 capabilities. Organizations cannot simply run conversion tools; they must undertake expert-led remodeling efforts that could span months or years for large repositories.

AI as the Great Equalizer

The textual syntax that contributes to SysML v2's complexity may paradoxically become its greatest adoption accelerator. Multiple research initiatives are demonstrating that large language models (LLMs) can generate SysML v2 models from natural language specifications, potentially reducing the time to create initial model structures from weeks to minutes.

A September 2024 paper in the INCOSE International Symposium proposed that LLMs can serve as an interpretive layer for SysML v2, reducing dependency on technical expertise traditionally needed for API navigation and model management.

Researchers at multiple institutions introduced SysTemp in June 2025, a multi-agent system that uses LLMs with template generation and parsing agents to automatically create syntactically correct SysML v2 models. An August 2025 study published in a ScienceDirect journal demonstrated that combining LLMs with retrieval-augmented generation and validation engines produces substantial improvements in model correctness.

Practical applications are emerging rapidly. AI assistants could provide real-time syntax help, suggest entire model structures based on natural language descriptions, and help engineers translate v1 thinking patterns into v2's paradigm. ThunderGraph AI has developed systems that build SysML v2 graphs incrementally while validating output, checking for inconsistencies, and ensuring traceability back to source documentation.

"AI LLM specifically interact through text," Lytvynov confirmed when asked about AI integration, "so definitely we understand that and we'll use that as much as possible."

The convergence is particularly timely. Industry thought leaders including Doug Rosenberg, Tim Weilkiens, and Brian Moberley have published guides on AI-assisted MBSE with SysML, providing roadmaps for hardware-software co-design accelerated by AI at every step.

Strategic Implications

"This is an essential contribution to our strategic ambition to evolve systems engineering to a fully model-based discipline," said Ralf Hartmann, president of the International Council on Systems Engineering (INCOSE).

For defense contractors, the timing is critical. The Department of Defense has sponsored transition guidance through its Digital Engineering, Modeling and Simulation office, emphasizing that SysML v2 adoption is essential for following current best practices and supporting strategic missions.

The standardized API specification—the first time an OMG metamodel has included a companion API standard—creates new opportunities for tool integration across aerospace, automotive, defense, and other industries. Model "patching" capabilities specifically address OEM-supplier collaboration by allowing black-box interfaces to be exchanged while protecting intellectual property.

However, the incomplete tool ecosystem poses near-term challenges. Dassault's 2026x release lacks dynamic behavior simulation (coming in 2026 refresh versions) and some third-party integrations. The Unified Architecture Framework (UAF) v2, critical for defense and aerospace, won't have a solid specification until late 2025.

Transition Realities

Dassault is taking a pragmatic migration approach, supporting both v1 and v2 within the same installation "for as long as needed" with no forced timeline. "Some programs will never move to SysML v2 and that's fine," Lytvynov said.

OMG notes that SysML v1.7, adopted in June 2024, will continue being used for several years as industry transitions.

Organizations face strategic choices: early adopters on new programs can gain competitive advantages through advanced capabilities, while teams maintaining legacy systems or in critical program phases may find disruption unacceptable. Those with AI development resources or access to emerging AI-assisted tools may find the transition significantly easier than expected.

"To take advantage of the rich features of the language and API, organizations are encouraged to develop a transition strategy and plan to adopt these technologies and leverage these practices," said Chris Scheiber, Chief Engineer at Lockheed Martin.

As the systems engineering community adapts to this generational shift, the unusual convergence of a more complex modeling language with increasingly capable AI assistance may paradoxically make SysML v2 easier to adopt than its simpler predecessor—transforming what appeared to be the technology's greatest weakness into its most compelling advantage.

---

Sources

1. Dassault Systèmes and OMG. (2024). "Exploring the Next Frontier of Systems Engineering: SysML v2" [Webinar transcript]. October 2024. Presented by Andrew Lytvynov, Nerijus Šatkauskas, and Thomas Baniszewski.

2. U.S. Department of Defense. (2024). "SysML v1 to SysML v2 Model Conversion Approach." Technical Report, March 2024. Office of the Under Secretary of Defense for Research and Engineering. Available at: https://www.cto.mil/wp-content/uploads/2025/02/SysML-v2-TransitionApproach-1.3.pdf

3. Object Management Group. (2025). "Object Management Group Approves Final Adoption of the SysML V2 Specification." Press Release, July 21, 2025. Available at: https://www.omg.org/news/releases/pr2025/07-21-25.htm

4. Object Management Group. (2025). "About the OMG System Modeling Language Specification Version 2.0." Formal specification, Publication Date: September 2025. Available at: https://www.omg.org/spec/SysML/2.0/About-SysML

5. Systems Modeling Community. (2025). "SysML-v2-Release." GitHub Repository. Available at: https://github.com/Systems-Modeling/SysML-v2-Release

6. OMG MBSE Wiki. (2025). "SysML v2 Transition." Last modified February 20, 2025. Available at: https://www.omgwiki.org/MBSE/doku.php?id=mbse:sysml_v2_transition

7. DeHart, John K. (2024). "Leveraging Large Language Models for Direct Interaction with SysML v2." INCOSE International Symposium, September 7, 2024. Available at: https://incose.onlinelibrary.wiley.com/doi/abs/10.1002/iis2.13262

8. Researchers from multiple institutions. (2025). "SysTemp: A Multi-Agent System for Template-Based Generation of SysML v2." arXiv:2506.21608v1, June 20, 2025. Available at: https://arxiv.org/html/2506.21608v1

9. Simon, Célina. (2025). "Introduction to SysML v2." Sodius Willert Blog, February 7, 2025. Available at: https://www.sodiuswillert.com/en/blog/introduction-to-sysml-v2

10. Obeo. (2025). "SysON 2025.2: Adding New Features to SysON is Our Ongoing Commitment." Company Blog. Available at: https://blog.obeosoft.com/syson-2025-2-adding-new-features-to-syson-is-our-ongoing-commitment

11. ThunderGraph AI. "Automating Model Based Systems Engineering with AI." Company Blog. Available at: https://www.thundergraph.ai/blog/automating-mbse

12. Rosenberg, Doug, Weilkiens, Tim, and Moberley, Brian. "AI Assisted MBSE with SysML." Leanpub. Available at: https://leanpub.com/aim

13. Smith, Jamie. (2024). "Automated Reasoning for SysML v2." Imandra Inc. Blog, September 30, 2024. Available at: https://medium.com/imandra/automated-reasoning-for-sysml-v2-ad7e87addba8

14. Friedenthal, Sanford. (2025). "The Next Generation Systems Modeling Language (SysML v2)." INCOSE presentation, May 2, 2025. Available at: https://www.incose.org/docs/default-source/content-library/the_next_generation_systems_modeling_language-sysml_v2-sfriedenthal-incose_sectoriii.pdf

15. Siemens. (2025). "SysML v2 for modern systems engineering: A practical guide." Teamcenter Blog, June 25, 2025. Available at: https://blogs.sw.siemens.com/teamcenter/sysml-v2-guide/

Note: This article synthesizes information from the Dassault Systèmes technical webinar, OMG official announcements, Department of Defense guidance, academic research papers, and industry analysis. Readers should consult OMG documentation at omg.org for official specification details.