VDE Updates GIS Switchgear Certification: AI PD Modules Require New Certificate

Frankfurt, 18 May 2026 — The German Association for Electrical, Electronic & Information Technologies (VDE) has updated its mandatory certification list for gas-insulated switchgear (GIS) on 18 May 2026. The revision introduces a new requirement: AI-powered partial discharge (PD) online diagnostic modules must now hold the standalone VDE-AR-E 2800-10:2026 certificate and undergo joint testing with the host GIS system. This change directly impacts manufacturers, exporters, and service providers supplying GIS equipment to grid operators across Germany, Austria, and Switzerland.

Event Overview

On 18 May 2026, the VDE published an update to its GIS switchgear certification framework. Under the revised rules, AI-based PD diagnostic modules — whether embedded or retrofitted — are no longer covered under the base GIS type approval. Instead, they must obtain the newly issued VDE-AR-E 2800-10:2026 certificate. Crucially, this certification requires co-validation with the specific GIS model during full-system testing. The requirement applies retroactively to all pending and future grid projects in the D-A-CH region, including those involving Chinese GIS suppliers.

Industries Affected

Direct Trade Enterprises: Exporters of GIS systems from China to the D-A-CH market face immediate compliance risk. Since the AI PD module is now a certified subsystem, failure to secure VDE-AR-E 2800-10:2026 — even if the GIS itself holds valid VDE-AR-E 2800-10:2023 — invalidates the entire unit’s certification. This may trigger contract renegotiations, delivery delays, or rejection at grid connection stages.

Raw Material & Component Procurement Enterprises: Suppliers of AI inference hardware (e.g., edge AI accelerators), PD sensor arrays, or firmware-integrated diagnostic boards must now align technical documentation and test protocols with VDE-AR-E 2800-10:2026 requirements. Procurement teams will need to verify supplier readiness for joint-system validation — not just component-level conformity — adding complexity to sourcing decisions.

Manufacturing Enterprises: GIS OEMs and system integrators must redesign internal certification workflows. Module integration can no longer be treated as a post-certification software upgrade; it now triggers retesting of the complete GIS assembly. This increases time-to-market by an estimated 8–12 weeks per configuration and raises non-recurring engineering (NRE) costs for test setup and third-party lab coordination.

Supply Chain Service Enterprises: Certification consultants, test laboratories, and conformity assessment bodies active in the high-voltage equipment sector must expand their scope to cover AI module–GIS co-validation. Their service offerings now require expertise in both IEC 60270-compliant PD measurement and AI model traceability (e.g., data provenance, inference reproducibility), which few currently possess.

Key Focus Areas and Recommended Actions

Verify module certification status before tender submission

For any GIS bid targeting German transmission system operators (TSOs) or Austrian grid companies, confirm that the AI PD module carries VDE-AR-E 2800-10:2026 — not just CE marking or legacy VDE certificates. Joint testing evidence must accompany the application package.

Engage VDE-accredited labs early in design phase

Delaying co-validation planning until prototype completion risks cascading delays. Manufacturers should initiate pre-assessment discussions with VDE-certified labs no later than the functional specification review stage — especially where proprietary AI models or custom sensor fusion logic are deployed.

Review contractual liability clauses

Contracts signed prior to 18 May 2026 may lack provisions for post-delivery recertification triggered by AI module updates. Legal and commercial teams should audit existing agreements for ‘certification continuity’ language and negotiate addenda where necessary.

Editorial Perspective / Industry Observation

Observably, this regulatory move signals a structural shift: safety-critical AI functions are no longer treated as ‘software features’, but as integral, certifiable subsystems — akin to circuit breakers or SF₆ monitoring units. Analysis shows that VDE-AR-E 2800-10:2026 emphasizes not only detection accuracy, but also AI transparency (e.g., uncertainty quantification per measurement, audit-ready inference logs). From an industry perspective, this is less about restricting AI adoption and more about embedding accountability into AI-enabled grid assets. Current more relevant interpretation is that the standard sets a precedent likely to influence CENELEC and IEC working groups — particularly for AI applications in protection relays and digital substations.

Conclusion

This update marks a maturation point in how regulators treat AI within high-reliability electrical infrastructure. It does not halt innovation, but redirects it toward verifiable, interoperable, and auditable implementation. For global GIS suppliers, compliance is no longer optional — it is a prerequisite for market access in one of the world’s most technically demanding grid environments.

Source Attribution

• VDE-AR-E 2800-10:2026, Requirements for Artificial Intelligence-Based Partial Discharge Diagnostic Modules in GIS, published 18 May 2026, VDE e.V.
• VDE Certification Bulletin No. 2026/05, Annex A: Transitional Arrangements for GIS Projects Submitted Before 18 May 2026 (status: effective immediately)
• Note: Implementation timelines for legacy installations undergoing AI retrofitting remain under consultation; stakeholders should monitor VDE’s official portal for updates.