Chip Design

Chip layout and verification workflows accelerated using reinforcement learning and generative AI from design data and specifications

Business impact

• design cycle time reduced from weeks or months to hours — AI accelerates layout generation and verification, drastically shortening design timelines
• up to 40% engineer time saved — Automation of repetitive tasks frees engineers to focus on innovation
• up to 40x faster RTL validation cycles — Agentic AI enables massively parallel simulations and faster verification
• up to 20x faster multiphysics and lithography performance — Accelerated computing and AI models speed complex simulations and analyses
• more than 10x faster library characterization — AI orchestration optimizes multi-tool workflows for rapid component characterization

Data requirements

• chip layout and floorplan data (Structured) — Used to train reinforcement learning models for placement optimization
• simulation and verification logs (Text) — Provide feedback for AI agents to improve design accuracy and coverage
• multiphysics and lithography simulation outputs (Numeric) — Enable AI to accelerate complex physical modeling tasks
• historical design parameters and performance metrics (Structured) — Inform AI-driven design space exploration and optimization
• natural language design specifications (Text) — Allow generative AI to produce RTL code and testbenches from text inputs

AI methods and techniques

• Reinforcement learning — Learns optimal chip floorplanning by sequentially placing components and receiving layout quality rewards
• Generative AI — Generates RTL code, verification scripts, and design collateral from natural language prompts
• Agentic AI — Autonomously orchestrates multi-step design and verification workflows with minimal human input

AI models and model families

AlphaChip, NVIDIA Nemotron, GPT-4o, Synopsys.ai Copilot

Industries

Real-world evidence

12 documented case studies on record.

Companies using this: Cadence, Chip Agents, Google, Keysight, Media Tek, NVIDIA, OpenAI, Samsung, Siemens AG, Silvaco, Synopsys, Tesla.