Synopsys, a developer of software for semiconductor design, recently introduced a technology that aims to facilitate the automation of tasks involved in creating new computer chips.
The company provides tools that assist engineers in arranging tens of billions of transistors, the fundamental electrical switches that power computers. This already intricate process has become more challenging as companies like Nvidia, a Synopsys client, transition from designing single chips to AI server systems that incorporate hundreds or thousands of chips, with the goal of launching new servers annually.
This shift necessitates the simultaneous design of numerous chips and components, placing significant strain on engineering teams, according to Synopsys CEO Sassine Ghazi, who addressed attendees at the company’s annual user conference.
"These are very complex and difficult to design," Ghazi remarked about the new AI computers. He emphasized that the pressures engineers face stem not only from the complexity of their work but also from the rapid pace of product delivery and associated costs.
On that note, Synopsys launched a new technology called AgentEngineer, which will initially focus on AI "agents" that can follow instructions from human engineers. These agents are designed to handle specific chip design tasks, such as verifying whether a circuit design functions as intended.
Looking ahead, Synopsys envisions these agents aiding in the coordination of intricate systems that involve various chips and components, ensuring timely product delivery.
"AI plays a huge role, because your R&D capacity is not growing," said Shankar Krishnamoorthy, who leads the technology and development group at Synopsys. He explained that companies cannot simply increase their teams exponentially, making it essential to enhance R&D capacity through innovative solutions.