Siemens and NVIDIA Agentic EDA Update: A WorldIndustrialWeekly Guide to 3D IC Electrothermal Design

From a WorldIndustrialWeekly standpoint, the release signals a shift in design governance: an agentic layer can use seven EDA tools and evaluate hundreds of design points per cycle while Siemens and NVIDIA extend the EDA agentic AI path tied to DAC 2026 and NVIDIA accelerated computing. The practical target is electrical, power, thermal and co-design analysis for 3D ICs.

Reported Efficiency and Exploration Data

Reported effect	Why it matters to industrial teams
up to a 95% reduction in engineering effort	Teams can spend less time on repetitive coordination and more time on architecture decisions.
up to a 99% reduction in workflow setup time	Faster setup lowers the barrier to running coupled analyses across multiple domains.
up to 50 times greater design exploration coverage	Broader coverage improves the chance of detecting power and thermal risks before they become schedule issues.

Case View: High-Density AI Compute Raises Electrothermal Risk

The NVIDIA Rubin GPU illustrates the pressure created by advanced AI systems. The company lists support for up to 50 petaflops of NVFP4 inference and up to 288 GB of HBM4 per GPU. These GPUs can be combined into systems such as NVIDIA Vera Rubin NVL72. When compute capacity, memory bandwidth and package resources are integrated more closely, power density and heat concentration rise. For readers responsible for product delivery, that makes electrothermal co-design a practical early-stage activity rather than a late verification chore.

How the Agentic Layer Coordinates Design Work

An engineer sets the objectives. The Fuse EDA AI Agent, integrated with Intelligence Center X, then coordinates multiple Siemens tools. It runs coupled electrothermal analyses, identifies unmet targets and assesses whether the root causes involve power delivery, placement, materials or a combination of factors. The Siemens Design Space Exploration and Optimization technology, referenced as DSE/DSO, lets teams test alternatives with physics-based signoff solvers.

Results from those analyses guide design-space exploration, trade-off assessment and reruns of selected configurations. The workflow coordinates nine steps across seven Siemens EDA tools. It performs 50 to 100 reasoning steps during an eight- to 10-hour runtime and can evaluate heat transfer coefficient (HTC) boundaries, power budgets and material stack-ups across 200 to more than 500 design points per cycle.

Siemens Tool Chain Covered by the Coordination Layer

• Innovator3D IC Integrator software: supports floor planning, 3D IC assembly and package design.
• Solido Characterizer software: characterizes libraries to generate Liberty files.
• mPower Digital software: performs power analysis and calculates temperature-dependent instance power.
• Calibre 3DThermal software: supports 3D IC thermal signoff and power mapping.
• Simcenter Flotherm software: provides system-level thermal simulation and visualization.
• Calibre xACT software: performs temperature-dependent parasitic extraction.
• Solido SPICE software: supports power- and temperature-aware electrical SPICE simulation.

NVIDIA Components Supporting the Agentic EDA Path

Siemens combines the NVIDIA Agent Toolkit with NVIDIA accelerated computing and NVIDIA CUDA-X libraries to support the agent-driven EDA workflow. Accelerated simulation tools provide analysis results that agents evaluate during design exploration.

• Nemotron open models: support orchestration and execution of agentic EDA subflows.
• NeMo Gym: tracks agent trajectories and identifies areas for refining agent skills and Model Context Protocol (MCP) tool descriptions.
• NeMo RL: supports post-training fine-tuning of language models with EDA knowledge and tool-calling capabilities.
• NeMo Relay: provides observability and audit records across agent harnesses.
• NeMo Switchyard: routes tasks based on compute requirements, accuracy and token costs.
• NVIDIA OpenShell: sandboxes agent execution to help protect intellectual property and foundry process design kits, or PDKs.
• CUDA-X libraries: accelerate supported engineering solvers used in the workflow to reduce simulation time during iterative design.

Common Mistakes to Avoid

• Treating electrothermal co-design as a final verification activity only after electrical, power, thermal and extraction decisions have hardened.
• Reducing the number of evaluated scenarios when time and computing constraints tighten, which can postpone recognition of power and thermal problems.
• Assuming one specialist tool can solve a coupled problem when causes may involve power delivery, placement, materials or a combination of factors.
• Overlooking governance, observability and sandboxing while expanding agent-driven workflows across sensitive design data and process IP.

Action Suggestions for Design and Operations Leaders

1. Define the objectives, constraints and acceptance criteria before using the Fuse EDA AI Agent, so the agent optimizes against an explicit target rather than an implicit one.
2. Use Design Space Exploration and Optimization, or DSE/DSO, to connect exploratory analysis with physics-based signoff solvers instead of leaving signoff as a disconnected endpoint.
3. Document the tool handoffs across the seven EDA tools so engineers know which result feeds power, thermal, extraction, SPICE, package or system-level decisions.
4. Adopt tracking and audit capabilities, including NeMo Relay, as part of production workflows so agent decisions can be reviewed and improved.
5. Protect process data and IP when agent execution touches foundry process design kits and internal design records.

FAQ for Industrial Readers

What is the main change?

The main change is the addition of an agentic coordination layer for 3D IC electrothermal workflows. Siemens and NVIDIA link the agentic AI capabilities introduced at DAC 2026 with NVIDIA accelerated computing to support electrical, power, thermal and co-design analysis.

Why does it matter beyond semiconductors?

The same logic applies to industrial systems where compute, memory and thermal management are becoming more compact. When multiple engineering domains must be evaluated together, coordination quality affects risk, schedule and product performance.

What should managers watch first?

Managers should watch setup time, exploration breadth, signoff discipline and governance. The reported gains, including up to a 95% reduction in engineering effort, up to a 99% reduction in workflow setup time and up to 50 times greater design exploration coverage, are meaningful only if teams can audit how the workflow chooses and reruns design points.