Nvidia Corp. has bought a $2 billion stake in Synopsys Inc. as part of a new multi-year partnership the companies announced today.

The chipmaker purchased the stock at a price of $414.79 per share, a slight discount to Synopsys’ Friday close. The deal has a similar structure as the $5 billion investment in Intel Corp. that Nvidia a few weeks ago. The latter transaction includes not only a stock purchase, but also a a technical collaboration that will see the companies integrate some of their products. 

Nvidia and Synopsys are likewise launching a product integration collaboration. The partnership’s first goal will be to speed up the latter company’s engineering applications, which chipmakers use to create semiconductor designs and test that they work as expected. Nvidia is a long-time user of the software. 

Synopsys also competes in other parts of the engineering software market thanks to its recently closed acquisition of Ansys. The deal bought it a suite of applications that can be used to design batteries, car engines, wind turbines and a wide range of other systems. Synopsys paid $35 billion to obtain Ansys’ technology.

The company plans to speed up its engineering applications using an Nvidia software bundle called CUDA-X. It’s based on CUDA, the programming language that developers use to write software for the chipmaker’s graphics processing units. CUDA-X includes compiles, libraries and other software modules that use GPUs to speed up applications.

One CUDA-X module, nvCOMP, is optimized to speed up data compression and decompression workloads. Applications compress data before sending it over the network to save bandwidth. Nvidia says that nvCOMP speeds up decompression using a hardware module called the Decompression Engine that ships with its latest Blackwell chips. CUDA-X can also accelerate other workloads such as Apache Spark analytics environments.

The second focus of the Nvidia-Synopsys partnership is enhancing the latter company’s AgentEngineer toolkit. The offering, which debuted in March, provides AI agents that partly automate tasks such as testing chip designs. Synopsys plans to enhance AgentEngineer using Nvidia’s NeMo Agent Toolkit and Nemotron collection of agent-optimized AI models.

After engineers put together a system design, they test it for errors using digital twins. A digital twin is a type of simulation that incorporates data from real-world deployments of a system. Synopsys will collaborate with Nvidia to help customers create such virtual representations using the chipmaker’s Omniverse and Cosmos tools. Omniverse eases the task of generating simulations, while Cosmos includes AI models that make those simulations more realistic.

The partnership will also see the companies collaborate on software distribution initiatives. Synopsys and Nvidia plan to enable “cloud access for GPU-accelerated engineering” software. Additionally, they will launch a joint go-to-market effort with the help of Synopsys’ direct sellers and channel partners.

Photo: Nvidia