Cadence Predicts AI Will Boost Chip Design Efficiency by 1,000 Times, Again Raises Annual Performance Expectations

·by Henderson·AI
Cadence Predicts AI Will Boost Chip Design Efficiency by 1,000 Times, Again Raises Annual Performance Expectations

Cadence stated during its third-quarter earnings call that with the integration of artificial intelligence into electronic design automation tools, chip design efficiency is expected to increase by approximately 1,000 times compared to 20 to 25 years ago. The company also raised its annual performance expectations again.

Cadence CEO Anirudh Devgan pointed out that 20 years ago, IBM and Intel required 500 people working for five years to design a CPU, whereas today, about 50 people can complete the task in six months, achieving an efficiency improvement of around 100 times; artificial intelligence is expected to bring an additional 5 to 10 times efficiency improvement on top of this.

Cadence Expects Significant Improvement in Chip Design Efficiency

On the same day, Cadence launched the Maestro AI model series aimed at semiconductor and system design. This series is based on open-weight models and utilizes Cadence design tools and data for post-training, allowing selection between proprietary models and external cutting-edge models based on design tasks. Cadence stated that preliminary results show that compared to general foundational models, Maestro can improve design quality by up to 30% and increase token usage efficiency by five times.

Cadence plans to integrate Maestro with AI design agents such as ChipStack, InnoStack, and AuraStack. Devgan noted that Samsung foundry adopted InnoStack in the third quarter to enhance process migration and design technology co-optimization efficiency. After using InnoStack in several projects, the Renesas MCU development team achieved a production efficiency increase of 34 times; MediaTek used ChipStack in mass production digital verification processes, resulting in a verification coverage increase of 50 to 70 percentage points, while the workload for engineers was reduced to about 50% to 70% of the original.

Cadence's third-quarter revenue was $1.63331 billion, a year-on-year increase of 22%; the non-GAAP operating profit margin, excluding one-time expenses, was 44.1%, with earnings per share at US$2.11 (approximately HK$16), marking a 9% year-on-year increase. The backlog at the end of the third quarter reached $8.2 billion, a record high. The company raised its full-year revenue expectations to between $6.387 billion and $6.417 billion, and further increased the expected year-on-year growth rate from 19% at the time of the second-quarter earnings release to 21%.

H
About the author
Henderson