Cognichip wants AI to design the chips that power AI, and just raised M to try
In the rapidly evolving world of artificial intelligence (AI), the design and manufacturing of silicon chips play a crucial role. Cognichip, a startup focused on leveraging AI for chip design, has recently raised $60 million in funding to further its mission of revolutionizing the semiconductor industry.
The Challenge of Chip Design
Chip design is a complex and time-consuming process. Advanced chips can take anywhere from three to five years to transition from conception to mass production. The design phase alone can extend up to two years before physical layout begins. For instance, Nvidia’s latest GPU line, Blackwell, contains an astonishing 104 billion transistors, highlighting the intricacy involved in chip design.
Cognichip aims to address these challenges by utilizing AI tools that can streamline the design process. According to Cognichip CEO and founder Faraj Aalaei, the goal is to integrate AI into the semiconductor design space, similar to how it has been adopted in software engineering. “These systems have now become intelligent enough that by just guiding them and telling them what the result is that you want, it can actually produce beautiful code,” Aalaei stated.
Funding and Growth
Emerging from stealth mode last year, Cognichip has successfully raised a total of $93 million since its founding in 2024. The latest funding round was led by Seligman Ventures, with notable participation from Intel CEO Lip-Bu Tan, who will join Cognichip’s board. Umesh Padval, a managing partner at Seligman, will also take a seat on the board.
Despite the significant funding, Cognichip has yet to showcase a new chip designed using its AI system. The company has not disclosed the names of its customers, although it has been collaborating with various partners since September.
Unique Approach to AI in Chip Design
What sets Cognichip apart from its competitors is its use of a proprietary model trained specifically on chip design data, rather than relying on a general-purpose large language model (LLM). This approach required Cognichip to access domain-specific training data, a challenging task given that chip designers typically guard their intellectual property closely.
To overcome this hurdle, Cognichip has developed its own datasets, including synthetic data, and has licensed data from partners. Additionally, the firm has implemented procedures that allow chipmakers to securely train Cognichip’s models on their proprietary data without exposing sensitive information. In cases where proprietary data is unavailable, Cognichip has utilized open-source alternatives.
Demonstrating Capabilities
Last year, Cognichip organized a hackathon at San Jose State University, inviting electrical engineering students to test its AI model. The participants successfully used the model to design CPUs based on the RISC-V open-source chip architecture, demonstrating the potential of Cognichip’s technology in a real-world setting.
Competitive Landscape
Cognichip is entering a competitive market, facing established players like Synopsys and Cadence Design Systems, as well as well-funded startups such as ChipAgents and Ricursive. ChipAgents recently closed a $74 million extended Series A funding round, while Ricursive raised a substantial $300 million in its Series A round.
Umesh Padval remarked on the current investment climate, stating, “The current flood of capital into AI infrastructure is the largest I’ve seen in 40 years of investing.” He further emphasized the potential for companies like Cognichip to thrive in what he described as a “super cycle for semiconductors and hardware.”
Conclusion
Cognichip is positioning itself at the intersection of AI and semiconductor design, aiming to transform how chips are developed. With significant funding and a unique approach to AI integration, the company has the potential to make a substantial impact in the industry. As the demand for advanced chips continues to grow, Cognichip’s innovative solutions may pave the way for faster, more efficient chip design processes.
Note: This article is based on information available as of October 2023 and may not reflect the latest developments in the industry.
