Nvidia’s $2 billion Synopsys bet aims to make AI chip design faster

Nvidia builds the chips that train AI. Now it wants AI to help build the chips.

The company announced a strategic multiyear partnership with Synopsys on December 1, 2025, backed by a $2 billion investment in Synopsys common stock at $414.79 per share. The goal is to rewire electronic design automation (EDA), the specialized software engineers use to plan, simulate and verify chips before anything gets manufactured.

What Nvidia is actually buying into

Think of EDA as the CAD software and crash-test lab for semiconductors, rolled into one. Before a chip with billions of transistors goes to a fab, engineers have to model how every circuit behaves and confirm the design works.

The partnership plans to fold three Nvidia strengths into Synopsys’ software: high-performance computing, agentic AI, and digital twin technology. Agentic AI refers to systems that can carry out multistep tasks with some autonomy, instead of just answering prompts. A digital twin is a virtual replica of a physical system that can be tested without touching the real thing.

On the technical side, the collaboration leans on Nvidia’s CUDA accelerated computing stack, along with AI tools such as NVIDIA NIM and the NeMo Toolkit.

The speed claims

The headline numbers are aggressive. Circuit simulation is projected to run up to 30x faster on Nvidia’s Grace Blackwell platforms.

Verification may see an even bigger jump. Synopsys AgentEngineer, integrated with Nvidia technology, can deliver up to 50x faster verification signoff, along with 20-30% productivity gains, according to the companies.

The work also extends beyond pure chip layout. The partnership is expected to speed up applications like molecular simulation and electromagnetic simulation, which matter for everything from materials research to antenna and packaging design.

At the GTC 2026 conference, the two companies showed off early customer results, including shorter verification closure times and stronger simulation capabilities.

A three-decade relationship gets a cap table

This is not a first date. Nvidia has relied on Synopsys tools to design its own chips for more than three decades.

The $2 billion equity check changes the nature of the relationship. Nvidia moves from longtime customer to shareholder, with a direct interest in how quickly Synopsys can turn its software into an AI-native platform.

Synopsys, for its part, has signaled the arrangement is not exclusive. The company remains open to working with other semiconductor firms, which matters a great deal for a vendor whose customers include companies that compete directly with Nvidia.

What this means for the chip industry

For Nvidia, the logic is circular in a useful way. Faster EDA tools running on Nvidia GPUs create another source of demand for Nvidia GPUs. Every chip designer who moves simulation onto Grace Blackwell becomes, in effect, an Nvidia customer twice over.

Nvidia paid $414.79 per share, which gives the market a clear reference point for how one of the most informed buyers in the industry priced Synopsys at the time of the deal.