Cadence Launches AI Chip Packaging Design Agent AuraStack
Claire Weston
Cadence Design Systems on July 15 launched AuraStack, an AI agent that drives PCB and chip-packaging design via natural language, claiming it can cut time-to-market by up to 50%; Nvidia, TSMC, and Schneider Electric are early adopters.
What does AuraStack actually do?
AuraStack is an AI "super-agent" built for printed circuit board (PCB — the green board inside phones and computers that carries all the chips) and chip-packaging design.
Engineers describe design goals in natural language. The system then calls Cadence's existing software tools to run layout planning and virtual testing automatically.
This means → the multi-tool, trial-and-error workflow engineers once ran by hand can now be driven by conversation, with Nvidia chips powering the compute underneath.
Are the efficiency numbers credible?
Cadence claims two headline figures: up to 50% shorter time-to-market and up to 15× productivity gains on individual tasks.
In a live demo, an engineer used AuraStack to redesign a 5G smartphone PCB — the goal was a lower-cost variant for a new market.
The system recommended consolidating components, projecting 28% cost savings, and suggested a compatible, cheaper power-management chip.
In plain terms = this is not design from scratch — it is "subtraction" on an existing board. The AI finds which parts can merge and which chip can swap for a cheaper one.
Who is already using it?
Confirmed early adopters include Nvidia, TSMC, and Schneider Electric.
Cadence VP Michael Jackson said: "The bottleneck isn't automation itself — the real bottleneck is engineering intelligence" — meaning the system must reason about cost-vs-performance trade-offs.
This reflects Cadence's pitch beyond "faster tools" toward "tools that make judgment calls for engineers" — a pivotal shift in the EDA industry (the software sector that designs chips) from automation to intelligence.
How does the pricing work?
Users can pair AuraStack with OpenAI's ChatGPT, Google's Gemini, Anthropic's Claude, or open-source models.
Pricing follows a consumption model — charges scale with actual AI compute used — on top of the existing Cadence tool subscription.
This means → Cadence collects two revenue streams: the base subscription stays, and AI usage adds incremental fees. For users, total cost hinges on how much AI compute they consume.
What does the market watch next?
AuraStack goes live this year. Jackson said the full rollout will complete by September.
The key validation point: whether consumption-based pricing can generate predictable revenue at scale in engineering-design workflows.
In plain terms = whether the tool works is one question; whether engineers will keep paying for AI compute — and how much — is what determines if this commercialization path holds up.
Content is for reference only, not financial advice.