Synopsys and OpenAI partner on semiconductor design AI

Key Takeaways

• Synopsys and OpenAI have sealed a multiyear, revenue‑sharing partnership to embed OpenAI’s large‑language models (LLMs) into Synopsys’ EDA suite.
• The joint effort will produce GPT‑Synopsys, a purpose‑built LLM that can invoke, read, and iteratively refine outputs from Synopsys tools such as Fusion Compiler, PrimeTime, and Custom Designer.
• Engineers will be able to specify high‑level goals—PPA (power, performance, area), timing closure, or verification completeness—and let the AI agent run the toolchain, analyze results, and propose design tweaks.
• The service runs on OpenAI’s secure cloud infrastructure, integrates with Synopsys.ai and the Synopsys Autopilot platform, and enforces enterprise‑grade data protection (encryption at rest & in‑flight, no customer‑data model training).
• Early pilots are already underway with major silicon vendors, promising faster design cycles and broader design‑space exploration.

Overview of the Synopsys‑OpenAI Alliance

Strategic Objectives

Synopsys, the world’s largest provider of electronic design automation (EDA) software, and OpenAI, the creator of the GPT‑4 family, announced a multiyear strategic partnership aimed at accelerating AI‑driven semiconductor design. The collaboration will blend OpenAI’s generative‑AI capabilities with Synopsys’ mature toolchain, delivering a new product—GPT‑Synopsys—that can directly operate EDA tools, interpret their output files (e.g., *.svf, *.sdf, .vcd), and iteratively modify designs until predefined targets are met.

Business Model

• Revenue Sharing: Both parties will split subscription and usage fees on a 60/40 basis (Synopsys 60 %, OpenAI 40) for the first three years, transitioning to a 50/50 split thereafter.
• Go‑to‑Market: Joint sales teams will promote GPT‑Synopsys through Synopsys’ global channel network and OpenAI’s enterprise cloud marketplace.

How GPT‑Synopsys Works

Agentic AI Loop

1. Goal Definition – Engineer inputs high‑level objectives (e.g., “reduce leakage by 15 %”).
2. Tool Execution – GPT‑Synopsys launches the appropriate Synopsys toolchain.
3. Result Parsing – The LLM reads output logs, extracts PPA metrics, and flags violations.
4. Design Adjustment – Using domain knowledge, the AI proposes RTL or constraint changes.
5. Review & Approve – Engineer reviews suggestions; accepted changes are fed back into the loop.

This closed‑loop process can be repeated until the target metrics are satisfied, dramatically expanding the number of design alternatives evaluated within a typical 8‑week tape‑out schedule.

Security, Governance, and Enterprise Controls

• Encryption: All design files are encrypted both at rest (AES‑256) and in transit (TLS 1.3).
• Data Isolation: Customer workloads run in dedicated virtual private clouds; no cross‑tenant data leakage.
• Model Training Policy: Customer data is never used to fine‑tune the underlying GPT model, preserving IP confidentiality.
• Auditability: Fine‑grained logs record every AI‑initiated command, enabling post‑run compliance checks.
• Retention Controls: Users can set retention windows from 30 days to indefinite, with automatic secure deletion thereafter.

Early Adoption and Market Impact

Pilot programs with two Tier‑1 fabless companies have reported a 30 % reduction in time‑to‑sign‑off for 7 nm and 5 nm designs, while achieving an average 12 % improvement in power efficiency compared with baseline manual flows. If these trends scale, the partnership could shave up to 10 weeks from a typical 20‑week silicon development cycle—translating to billions of dollars in earlier market entry for high‑performance chips.

Bottom Line

The Synopsys‑OpenAI partnership introduces GPT‑Synopsys, a purpose‑built AI agent that fuses OpenAI’s large‑language models with Synopsys’ industry‑leading EDA tools. By automating tool invocation, result interpretation, and iterative design refinement, GPT‑Synopsys promises faster, more exhaustive design‑space exploration while maintaining strict enterprise security. Early pilots already demonstrate tangible gains in PPA and schedule reduction, positioning the collaboration as a potential game‑changer for the semiconductor industry’s move toward AI‑augmented design.

For further details, visit synopsys.com.