Summary

• Cadence OnCloud is now available on sovereign infrastructure in Munich, providing European data residency for semiconductor and system design workloads.
• The company plans to add Nvidia Vera and Grace CPU systems, Millennium M2000 supercomputing resources and additional Reality Digital Twin Platform instances as the facility expands.
• Vertiv, HPE and Nation Data Center are using or adopting Cadence's wider digital twin technology. Their involvement does not establish that every capability is already operating in Munich.

Cadence has made sovereign cloud services available for semiconductor and system design in Munich, giving European engineering organisations a new option for processing sensitive technical information within European jurisdiction.

The facility provides access to Cadence OnCloud, the company’s cloud environment for electronic design automation and associated engineering workloads. Additional computing resources, including Nvidia Vera and Grace CPU systems and the Cadence Millennium M2000 AI Supercomputer, are planned as part of its continuing expansion.

Cadence announced the Munich development at GTC Berlin on 21 October, positioning the service as a way for customers to access demanding computing resources while maintaining European data residency.

The distinction between current and planned availability is important. OnCloud is available through the sovereign infrastructure now, but the announcement’s broader description of an integrated AI engineering platform includes hardware and digital twin instances that have not all been confirmed as operational at the Munich site.

Chip development brings demanding data requirements

Semiconductor design requires engineers to evaluate electrical circuits, physical layouts and interactions between components before designs are submitted for manufacturing.

The work involves software capable of analysing complex mathematical problems, including electrical behaviour, heat generation, signal interference and the effects of arranging components within increasingly dense packages.

These calculations can require substantial computing capacity, particularly when teams are working on advanced chips, multiple connected semiconductor devices or large electronic systems.

Cloud infrastructure allows organisations to obtain computing resources without purchasing and maintaining every system themselves. It can also provide temporary capacity when simulation requirements exceed the resources available through an internal engineering department.

However, chip designs and engineering models can contain highly sensitive commercial information. Detailed layouts, simulation results and product specifications may represent intellectual property developed over several years.

Moving such information into an externally operated environment introduces questions about where it is processed, who can administer the underlying systems and which contractual or legal arrangements govern access.

Cadence’s Munich offering addresses the location of workloads by providing infrastructure within European jurisdiction, expanding the company’s existing European cloud presence, which includes resources supported through public cloud facilities in Frankfurt.

The company says Cadence OnCloud meets ISO 27001, ISO 27017 and SOC 2 Type II assurance requirements, supporting the security proposition associated with the service.

These standards concern information security and related controls. They do not independently establish that every customer’s deployment satisfies all applicable regulatory, export control or contractual requirements.

Each organisation must still assess the access arrangements, security configuration and responsibilities associated with its particular use of the service.

Additional computing capacity remains planned

Cadence is working with Nvidia to expand the processing resources available for engineering calculations and AI-assisted development.

The planned Munich additions include Nvidia Vera and Grace CPU systems and the Millennium M2000 Supercomputer, which uses Nvidia Blackwell technology.

Cadence is also developing access to additional Reality Digital Twin Platform instances through the facility.

These technologies can support different stages of engineering work. Electronic design automation tools help teams create and verify chip designs, while accelerated computing can reduce the time required for suitable numerical calculations.

Supercomputing systems may also allow multiple complex simulations to operate concurrently, enabling engineers to compare alternative designs without waiting for each calculation to finish before beginning another.

Those benefits depend on the software and hardware configuration involved. Applications that cannot use accelerated processing effectively may not achieve the same improvements as workloads designed for GPUs or specialist computing systems.

Cadence has not provided a completed inventory showing all of the planned equipment operating in Munich as of 21 October.

Its announcement explicitly identifies OnCloud as available today while describing the additional hardware and software instances as part of ongoing infrastructure expansion.

Customers evaluating the service will therefore need to establish which resources can be reserved now, which remain on the roadmap and what commercial terms apply to accessing them.

Digital twins extend the engineering proposition

Cadence is also applying its simulation technology to the physical infrastructure used by large AI computing installations.

The Reality Digital Twin Platform creates computational representations of data centre equipment and operating conditions, allowing engineers to evaluate designs before hardware is installed.

Such models can be used to examine power distribution, cooling, airflow, temperatures and equipment placement, helping identify where one part of a design may impose additional requirements on another.

Large AI systems can place considerable demands on electricity and cooling infrastructure, particularly when computing racks are configured for high power density.

Engineers must therefore coordinate several systems whose performance is interconnected. A change in equipment placement may affect cooling requirements, while differences in electrical distribution can influence the amount of computing capacity a facility can support.

Cadence is working with Nvidia’s Omniverse technologies and DSX Blueprint to support these evaluations.

It has also developed a validated digital representation of the Millennium M2000 system that customers can incorporate into infrastructure simulations.

The model allows engineers to examine how the equipment could interact with the surrounding facility before it is installed, although the results still depend on the accuracy of the assumptions and supporting equipment data.

Simulation can reduce some design uncertainty without eliminating the need to test actual infrastructure after construction.

Industrial partners use the wider platform

Vertiv is using the Cadence Reality Digital Twin Platform to design and validate its modular MegaMod HDX AI data centre system.

HPE is adopting the platform for its Data Center AI Mod POD solution, while France-based Nation Data Center is using the technology to evaluate sovereign, high-density computing infrastructure.

These applications demonstrate customer activity involving the wider Cadence simulation portfolio, but they are not evidence that each organisation is using the new Munich OnCloud facility.

Cadence also reports that its digital twin technology increased data centre capacity utilisation from 71% to 89% in a recent deployment, while operational availability improved from 90% to 98%.

The company has not identified that result as a measurement from the new Munich installation. It should therefore be understood as a reported outcome from a separate application of the technology.

The figures also require context. Higher capacity utilisation may improve the use of existing infrastructure, but the commercial benefit depends on operating costs, reliability and the ability to support customer workloads.

Likewise, improved availability may reflect several engineering or operational changes rather than being attributable entirely to a digital twin.

European sovereignty requires more than geography

Cadence presents the Munich facility as an option for organisations seeking stronger European control over engineering workloads.

Locating information within the EU can help customers meet data residency requirements and reduce some concerns associated with processing sensitive information in another jurisdiction.

However, data residency is not identical to complete technological independence. European infrastructure may still depend on processors, software, maintenance arrangements and external suppliers based in other regions.

The distinction matters for semiconductor development because intellectual property protection involves contractual rights, technical access controls and security procedures alongside the physical location of computing equipment.

Cadence also says the facility can support startups participating in its EuroCDP programme, potentially extending access to specialist design resources beyond companies able to operate their own substantial computing installations.

Smaller engineering businesses may benefit from flexible access to software and computing capacity, although they must still consider usage charges, integration requirements and the treatment of proprietary design information.

For larger organisations, the new service provides another deployment option alongside internally operated infrastructure and conventional public cloud arrangements.

The Munich development establishes commercial availability of sovereign Cadence OnCloud services. The planned expansion will determine when customers can also access the additional computing systems and digital twin instances described in the company’s wider platform strategy.

Its longer-term commercial significance will depend on those capabilities becoming operational, the reliability and cost of the service, and whether European engineering businesses regard the resulting control arrangements as sufficient for their most sensitive workloads.