Cloud native EDA tools & pre-optimized hardware platforms
Artificial Intelligence (AI) is dominating technology headlines (and for good reason). AI touches every industry, including supply chain management, gaming, mobile, consumer goods, and aerospace and defense. This new era of AI is making intelligence more accessible and applicable to nearly every segment.
The democratization of AI power is revolutionizing not only how we innovate, but also the speed at which we innovate. We haven’t seen this level of productivity-busting technology since at least the advent of the computer (and possibly even the industrial revolution!). But doing more, faster, and at the scale necessary to meet insatiable demands coupled with the staggering explosion of intelligent systems is forcing a reinvention of computing and, thus, the way we design chips.
This transformative technology is being applied successfully to the very tools used to optimize chip design and analyze vast amounts of data, building efficiencies throughout the chip development process. With all that has been done, however, there are still significant gaps in what designers want to achieve and what available resources and time allow. The is where the next phase of AI will come into play. Generative AI is already proving itself in areas like speech and image generation, and it is starting to show what is possible when applied to chip design.
This very topic is a key discussion point at this year’s Computex Forum in Taiwan themed “Let’s Talk Generative AI.” The conference will feature insights into the advancements of generative AI in various technology applications and industries across the globe by luminaries from Arm, AWS, Google, Nvidia, and Synopsys. Our very own Thomas Andersen, Vice President of AI and Machine Learning, will delve into the role of AI in chip optimization and data analytics, and discuss how generative AI is becoming a game-changer in chip design.
The explosion of data is pushing the limits of compute while also building the foundation for AI development. The complexities driving this exponential data growth necessitate the use of AI to collect, manage, and interpret data, as well as provide guidance on its application. The challenges of chip design, escalating development costs, shrinking market windows, and a talent deficit in engineering are all factors propelling the adoption of AI.
Even as AI becomes more prominent in chip design, we have to recognize the challenges that will continually need to be addressed. Although there is an explosion of data, much of that data isn’t public. This data scarcity, especially for training generative AI models, poses a significant challenge, as does the protection of proprietary data and intellectual property. Even after hurdling this data conundrum, designers still need to validate the output from these models. In his forthcoming talk at Computex Forum, "Unleashing the Power of GenAI for Chip Design," Thomas Andersen will address these challenges and propose an evolutionary approach to generative AI where we learn to crawl, walk, run, and even eventually teleport ourselves. Don't miss this insightful talk on Wednesday, June 5 at 11a.m. NST at the event.