Cadence Q3 Earnings: Chip Design Demand Expands Beyond AI, EDA Revenue Up 20%
nashnova research
Cadence's EDA revenue grew 20% year-on-year in Q3, with chip-design tool demand spreading from AI hardware into analog, memory, aerospace and finance — the clearest customer-mix shift the company has signaled in years.
Who is buying chip-design tools now? Not just AI companies?
EDA (electronic design automation — the software engineers use to design chip circuits) growth was driven mainly by AI hardware clients. This quarter, analog chipmakers, memory manufacturers, systems companies, and even aerospace and finance teams building custom chips all expanded purchases.
This means → EDA's growth engine has shifted from "one AI leg" to multiple legs, making the revenue base broader and more resilient.
Today 45% of Cadence's business comes from systems companies, not traditional chipmakers. In plain terms = roughly half the money now comes from tech firms designing their own chips, not legacy fabs.
All three segments are growing — and hardware hit a record. How?
Core EDA revenue rose 20% YoY, semiconductor IP (reusable chip-design blocks) grew 24%, and systems design & analysis grew 30%.
On the hardware side, the Palladium emulation and Protium prototyping systems — machines that verify chip designs before fabrication — posted another record quarter. The CFO said backlog is deep, the company is building inventory to ship, and "there are no capacity or supply-chain bottlenecks."
The CEO cited a TSMC forecast: chip sizes will expand 48× over the next five years as process shrinks combine with 3D packaging. This means → bigger, more complex chips drive exponentially more verification work, and the ceiling for emulation hardware demand is still far off.
How exactly does "agentic AI" drive EDA demand? Is it more than a buzzword?
A Citi analyst pressed the point: is hardware or agentic AI — AI systems that autonomously call tools to complete multi-step tasks — the real driver of EDA growth?
The CEO responded: stripping out hardware, recurring revenue grew 18% this quarter, up from 11% a year ago. The CFO called agentic AI "a demand multiplier for EDA" — an AI agent calls Cadence tools far more often than a single engineer does.
In plain terms = one AI agent generates the tool-usage volume of several engineers; Cadence charges by usage, so more calls mean more revenue. This reflects a double dynamic: AI is not just a source of customer demand — it is directly amplifying Cadence's own tool consumption.
Why hasn't the profit margin kept up with revenue growth?
A Wells Fargo analyst asked directly: why isn't high revenue growth translating proportionally into profit? The CFO's answer: revenue will grow roughly $1.1 billion this year, with non-GAAP operating profit up about $500 million.
The drag comes from M&A — the acquired Hexagon design and engineering business is expected to generate about $160 million in full-year revenue, slightly above plan, but with minimal profit contribution.
This means → strip out Hexagon, and Cadence's incremental operating margin stays at 54–55%. In plain terms = the legacy business's profitability hasn't changed; the new acquisition just hasn't started earning its keep yet.
Rival Synopsys set long-term targets — how did Cadence respond?
A Bank of America analyst raised Synopsys's September investor-day targets: mid-teens revenue growth through FY2030, a non-GAAP operating margin of roughly 50%, and a royalty-linked IP deal with Amazon worth over $1 billion.
The CEO declined to offer a multi-year outlook, saying only that this year's revenue growth rate plus non-GAAP operating margin will exceed 65%. Long-term guidance will follow the usual schedule in February. On the royalty model, he called it "just one of many paths."
This signals that Cadence is unwilling to let a competitor set its cadence — but Synopsys's targets have already become the market's reference frame. The February earnings call will be the key window to watch.
市场有风险，内容仅供研究参考，不构成投资建议。