Never miss a story from Corporate and Culture. Subscribe for in-depth analysis and articles.
The Design Automation Conference (DAC) recently celebrated its 60th anniversary in San Francisco with an outstanding lineup of speakers and presentations. Academic, research, government, and industry attendees participated in robust research and engineering tracks, hands-on training sessions, and a bustling exhibition. Moreover, the conference offered educational hands-on training sessions by Cadence, enhancing attendees' practical skills in the ever-evolving field of design automation.
This year’s DAC was a crucible for groundbreaking discussions with hot topics that ignited fervent debates, including generative AI (GenAI), chiplet integration, heterogeneous integration in an SoC package, and other underlying semiconductor technologies. Beyond these technical intricacies, DAC served as a forum to explore vital subjects such as data security, reliability, automotive electronics, and the future of autonomous vehicles, with the overarching theme being the urgent need for AI and machine learning to process data at an unprecedented pace, reducing costs and time to market—a challenge the industry is poised to confront head-on.
The conference featured a lineup of four insightful keynotes, along with an additional quartet of visionary talks. Attendees also had the opportunity to engage with technical SKYTalks and TechTalk presentations. In total, the event hosted an impressive array of over 100 sessions. In this blog post, I will provide summaries of Lip-Bu Tan's Visionary talk, Paul Cunningham's SKYTalk, the panel discussion in which KT Moore participated as a panelist, and the Keynote address delivered by Alberto Sangiovanni-Vincentelli.
Lip-Bu Tan, former CEO and current executive chair of the Board of Directors at Cadence, gave a visionary talk on “Advancing Precision Medicine through Generative AI-Driven Drug Development.”
According to Lip-Bu, GenAI could transform drug discovery and precision medicine, providing a multitude of opportunities for researchers and clinicians. Using deep learning algorithms, GenAI can potentially pinpoint potential candidates for new drugs and predict a patient's reaction to various treatments, leading to more tailored and accurate medicine.
Lip-Bu Tan explained that the biosystem market, specifically pharma, is experiencing a slowdown in productivity, with pharmaceutical companies set to lose $50 billion due to expired drugs. AI and machine learning simulation can drive growth in this sector, making it more predictable and increasing FDA approval rates. GenAI is also predicted to impact the industry significantly. AI-driven drug discovery is vital to this growth, with companies investing heavily in 3D molecular structure and new drug development. His talk explored how GenAI transforms drug discovery and precision medicine and discussed the potential for future innovation in these fields. He highlighted the use of AI in precision drug discovery, including the development of new antibiotics and the first FDA-approved organ drug.Lip-Bu Tan discussed the application of computational software in drug discovery and how Cadence drives innovation there. In addition to Cadence’s CFD solutions for system simulation, there are also opportunities for drug discovery simulation. Cadence recently acquired OpenEye Scientific, a leader in molecular modeling and simulation that 19 of the top 20 pharmaceutical companies already use, to create a new business unit called Cadence Molecular Sciences.Lip-Bu concluded by focusing on the productive uses of AI and the potential for AI to drive new drug development for personalized medicine. He added that the next generation can drive some of this more personalized medicine, as a lot of that data is already available and can be used to try some of these new applications.
Paul began his SkyTalk, “Entering a New Era with EDA 2.0 and AI-Driven Electronic System Design,” by expressing his intention to discuss AI in the context of Cadence and its relevance to the broader community, emphasizing the golden era for technology with chips becoming ubiquitous in devices despite economic challenges.
He talked about the increasing complexity of chip design, projecting a 100X increase in complexity by 2030. Paul suggested focusing on task abstraction and integrating AI into the process instead of following the traditional approach. He believes AI can raise engineers' task abstraction levels, leading to a 10X improvement in productivity and reducing challenges.
He introduced a simple visualization comparing human and AI capabilities, emphasizing that while humans possess deep thinking and intuition, AI is infinitely scalable in processing information. He sees an opportunity in AI's ability to handle large volumes of data and perform initial classifications. Paul stressed the importance of accelerating the human-AI intersection and catalyzing engineers' work rather than replacing them. He believes this approach should be a collaborative effort within the industry, aiming for "EDA 2.0."
He mentioned the need to move from design abstraction to task abstraction powered by AI, which is EDA 2.0. EDA tools are now required to run multiple engines and tools simultaneously, leading to new technologies and task abstraction.
Paul emphasized that creating a data platform for EDA tool input and output is critical for developing new AI-driven design optimization technologies that analyze multiple design aspects across different tools and engines. He discussed the Cadence Joint Enterprise and Data (JedAI) Platform and mentioned some of the products built on top of it—including Cadence Cerebrus Intelligent Chip Explorer, Allegro X AI technology, and Verisium AI-Driven Verification Platform—to deliver multi-engine, multi-run AI-driven technologies for various design aspects.
Paul also mentioned the concept of "AI inside," where AI techniques are embedded within individual EDA tools to improve performance and efficiency. He provided examples of AI applications in constraint solving and formal proof solving.
Brain Bailey of Semiconductor Engineering moderated the panel, “The Industry 4.0 Revolution of Semiconductor Design Panel Discussion,” and KT Moore, VP of Corporate Marketing, was among the panelists.
The panel discussion explored the impact of Industry 4.0 on semiconductor design and development, assessing the industry's readiness for these innovations. It explored integrating value chain stages to bring testing and lifetime operation closer to the design phase for improved collaboration and learning. Embracing machine learning, cloud analytics, remote diagnostics, and predictive modeling can empower the semiconductor sector to make more efficient decisions, potentially boosting revenue.
Additionally, the discussion examined the obstacles hindering Industry 4.0's full implementation, including data management, security, and interoperability challenges. Comprising experts from diverse semiconductor segments, the panel aimed to offer valuable insights into the industry's current state and the opportunities and hurdles posed by Industry 4.0. Adopting Industry 4.0 technologies can revolutionize industries, leading to heightened productivity, cost reduction, product quality enhancement, and improved customer experiences. The question remains: Is the semiconductor industry prepared for this transformative shift?
In his keynote, “Corsi e Ricorsi: Here We Go Again,” Alberto Sangiovanni Vincentelli discussed the evolution of EDA as a prime example of AI. He clarified that AI encompasses the "theory and development of computer systems able to perform tasks that normally require human intelligence." He also differentiated between the hype and reality of machine learning applied to EDA. He assured the community that design automation engineering is still one of the world's most intelligent professions.
Vincentelli argued that semiconductor and system design has evolved through a pattern of intuitive approaches, great insights, and rigorous methodologies, which he called "corsi e ricorsi," based on Giambattista Vico's principles. Over the past six decades, this phenomenon has fueled advancements in technologies and computational software, freeing designers from the constraints of choice and empowering them to tackle 18 orders of magnitude of design complexity. As a result, the possibility horizon continues to expand.
Vincentelli highlighted the importance of 3D-ICs to solve single-chip limitations. He explained how heterogeneous architectures can overcome these boundaries. Vincentelli also addressed the challenges faced by EDA when developing 3D integrated circuits and placed them in the context of the packaging methods used over the past four decades. However, his main concern was the lack of talent in the field. He stated that the scarce resource of the future is talent, and many people are focusing on studying AI rather than designing the necessary chips to implement it.