AI Design Automation: From Sequential Steps to Continuous Exploration
AI design automation is the use of machine learning, generative design tools and domain‑specific models to automate simulation, layout and verification tasks that previously required long, manual engineering iterations, shrinking timelines from months to minutes while enabling engineers to test far more design options at acceptable accuracy. In computational fluid dynamics (CFD), reduced‑order models are emerging that can approximate detailed solvers and support CFD simulation acceleration, so thousands of geometry variants can be screened in minutes instead of weeks of batch runs. Similar trends appear in electronics, where AI-powered EDA is moving beyond rule checking into architecture generation, assisted schematic capture and automatic placement and routing. Across these fields, the shared goal is engineering workflow optimization: shift simulation and trade‑off studies to earlier concept phases, cut down on handoffs between tools, and let specialists focus on corner cases and innovation instead of repetitive configuration work.
CFD Simulation Acceleration with AI-Based Reduced-Order Models
In aerodynamics and fluid systems, generative design tools linked to AI-based reduced‑order models are changing how teams approach design exploration. Instead of running full CFD for each geometry, engineers can build a trained surrogate that predicts flow and performance for new shapes at a fraction of the cost. When coupled with automated geometry variation, this supports evaluation of thousands of design candidates in the time it once took to simulate a handful. A new add‑on for AI-powered CFD design exploration in Siemens’ Simcenter ecosystem illustrates the trend: physics-informed models approximate high‑fidelity solvers closely enough for early design ranking, freeing expensive CFD jobs for final refinement and certification. For automotive aerodynamics, cooling systems, or turbomachinery, that means faster convergence on efficient concepts, better use of computational resources, and a shorter loop from idea to validated shortlist.
Generative Design and AI-Powered EDA in Automotive Electronics
Automotive electronics design is becoming a key proving ground for AI-powered EDA. In the Zuken Valeo InnoLab program, generative design drives architecture creation: Zuken’s System Planner combined with Valeo’s AI agents can instantly propose and assess multi‑criteria electronic system architectures against internal standards. Digital continuity keeps those choices traceable for ASPICE 4.0 hardware engineering requirements, while AI assistants support engineers during detailed design with solution searches, rule checks and constraint application. Auto‑placement and routing rely on AI Place and Route algorithms in Zuken’s Design Force engine to push layout density and signal quality without the same level of manual tuning. This approach cuts manual edits, improves repeatability and reduces late‑stage layout surprises. More broadly, it shows how generative design tools are moving upstream in the flow, from point accelerators to active collaborators that shape design intent from the first block diagrams.
Faster Early-Stage Power Electronics Design with PLECS-Based Checks
In power electronics, AI-linked and model-based simulators are compressing the time needed for early feasibility checks. ROHM’s browser‑based PLECS Simulator focuses on fast loss and thermal calculations so engineers can screen device options before detailed SPICE work. Designers pick a topology, select ROHM power devices and obtain loss and temperature‑rise estimates in seconds to minutes, which helps narrow choices for converters and inverters. According to ROHM, the simulator already supports 20 circuit topologies and will expand across SiC devices, IGBTs and power modules. Used alongside high‑precision SPICE tools such as the ROHM Solution Simulator, this workflow lets teams separate early design exploration from final verification. The result is fewer unsuitable device selections carried into lab prototypes and a shorter loop between concept, simulation and hardware, especially in applications where efficiency and thermal margins dominate the design envelope.
Multi-Domain Signal Path Simulation for AI and High-Speed Links
High‑speed data links for AI infrastructure and high‑performance computing are pushing design tools beyond single‑domain analysis. Keysight’s new Electrical‑Optical‑Electrical (EOE) workflow in ADS 2026 models full signal paths across electrical and optical domains inside one environment, joining its High Speed Digital flow with Keysight Photonic Designer. Engineers can now simulate transmitters, photonic integrated circuits and electrical receivers in a single chain and catch signal‑integrity problems that only appear when both domains interact. The workflow supports bidirectional optical links and wavelength‑division multiplexing, so teams can study nonlinear effects and crosstalk across multi‑wavelength channels such as 800G and 1.6T interconnects. According to Keysight, 87% of hyperscale optical transceivers are expected to run at 800Gbps or higher by 2029, making this kind of multi‑domain co‑simulation critical for reliable link architectures and for keeping design iterations ahead of rapid bandwidth growth.