Cadence Maestro brings NVIDIA AI models to chip design

A new tool for chip engineers

Cadence Design Systems has announced a new portfolio of AI models called Cadence Maestro. This suite is designed specifically to help engineers develop computer chips more efficiently. Designing modern semiconductors is a massive undertaking that requires arranging billions of transistors while balancing power, cost, and performance.

The goal of Maestro is to introduce AI agents into the engineering workflow. These agents are intended to perform specific tasks such as generating initial designs, evaluating the results of those designs, and providing support for difficult technical decisions. By automating these parts of the process, Cadence hopes to help engineers manage the increasing complexity of hardware development.

The first models in the Maestro portfolio are scheduled to be available for a small group of early-access clients starting in the final quarter of 2026. This rollout follows a growing trend where software companies are racing to integrate AI into specialized industrial fields. Cadence aims to provide a tool that understands the specific nuances of semiconductor and system design rather than acting as a general-purpose assistant.

The NVIDIA partnership

Cadence is building Maestro on a foundation of NVIDIA technology. The portfolio includes parts of NVIDIA’s open-weight Nemotron model, which is a type of AI model where the underlying weights are available for public use. To ensure security, the suite also incorporates NVIDIA’s OpenShell technology to protect data during the design process.

By combining these models with its own electronic design automation tools, Cadence is creating a specialized system. The company claims that these specialized models perform significantly better than general-purpose AI in engineering contexts. According to Cadence, early tests showed a quality improvement of as much as 30% and a token efficiency boost of up to five times compared to standard models.

Token efficiency refers to how effectively an AI model uses its output units to complete a task. By optimizing this, Cadence aims to make the AI more practical for heavy engineering workloads. The company also noted that while these results were seen in early testing, actual performance might differ depending on the specific task a customer performs.

Deployment and data privacy

One of the standout features of Maestro is the flexibility it offers for where the AI actually runs. Customers have the option to use the models through the cloud platform provided by Cadence, via public cloud services, or on their own internal hardware. This is a critical distinction for companies that handle highly sensitive chip designs and proprietary engineering data.

Keeping AI systems in-house allows these companies to maintain much tighter control over how their information is processed and who can access it. By keeping the data behind corporate firewalls, firms can mitigate the risks associated with external cloud services. This setup is particularly appealing to semiconductor giants who view their designs as some of their most valuable intellectual property.

Furthermore, customers have the ability to fine-tune the models using their own specific chip designs and internal workflows. This customization helps the AI agents produce results that align more closely with unique engineering requirements. However, Cadence notes that users will still need to perform their own evaluations of security and performance before moving these tools into a full production environment.

The competitive landscape

The launch of Maestro occurs while Cadence faces competition from its main rival, Synopsys. Synopsys is currently developing a tool called GPT-Synopsys through a partnership with OpenAI. Unlike the Cadence approach, that model is intended to run on infrastructure hosted by OpenAI, highlighting a different philosophy regarding AI deployment in the chip design industry.

Cadence is positioning itself as a versatile provider by stating that Maestro is built to work with various types of AI models. This includes both open-weight systems and proprietary ones. The company has also signaled that it intends to continue collaborating with various commercial AI developers as part of its broader long-term strategy.

The move reflects a broader shift in the technology sector to use AI for more than just writing code or answering simple questions. Instead, the industry is moving toward using AI to handle complex engineering work that requires deep technical knowledge. If these systems can reliably improve design quality and shorten development cycles, they could become essential tools for companies facing increasingly difficult technical demands.