Partnering with Ricursive Intelligence: Unlocking the Future of Chip Design with AI
In today's AI-driven world, compute is the most valuable resource, powering everything from Nvidia's GPUs to Google's TPUs and Amazon's Trainium. But the race to design the most performant chips for AI workloads is a challenging one, with major hurdles to overcome.
The first hurdle is speed. Chip design is a slow process, taking 12-24 months at mature nodes and a staggering 18-36 months at the leading edge for 5nm or 3nm chips. The second hurdle is cost. Chip design is prohibitively expensive, with an average cost of $200-250 million for 7nm, $450-500 million for 5nm, and a mind-boggling $600-650 million for 3nm chips. And that's not all - a significant portion of this cost is attributed to human labor and Electronic Design Automation (EDA) tooling, which is dominated by Cadence and Synopsys, generating billions in annual revenue.
This is where AlphaChip comes in. AlphaChip, a pioneering effort in AI for chip design, has caught my attention for its potential to revolutionize the entire chip design process. By leveraging AI, AlphaChip has demonstrated the ability to cut the floorplanning step in physical design from months to hours, showcasing the power of AI to transform chip design.
But what if we could take this a step further? What if we could build AI to automate the entire chip design process, from architecture design to RTL and verification, all the way through physical design? Imagine if chip design took days instead of years, unlocking a world of possibilities and revenue potential.
Imagine a world where each chip design didn't cost hundreds of millions of dollars, and chip companies didn't need to rely on large human teams and clunky EDA tooling. And most excitingly, what if we could unlock novel chip designs that we might never have explored before?
AlphaChip revealed an important human bias in chip design - our tendency to think in Manhattan grid-like structures. However, AlphaChip's designs were different, more organic in shape, inspired by nature. So different, in fact, that humans initially rejected them. Yet, AlphaChip's impact was profound, shaping four generations of the TPU.
At Sequoia, we are thrilled to partner with Anna Goldie and Azalia Mirhoseini, the co-founders of Ricursive Intelligence. They have pioneered AI for chip design by creating and leading the AlphaChip effort, and they are at the forefront of this emerging AI for chip design ecosystem. Anna and Azalia are visionaries with a clear and ambitious vision, and they have assembled a team with the highest talent density in the field.
Ricursive Intelligence's core belief is that chip design is the compute bottleneck, and progress in AI, hardware, and infrastructure is limited by the speed and efficiency of silicon creation. In their own words, 'If we get this right, it's not just faster chip design cycles; it's a fundamental expansion of what's possible in hardware. Once chip design becomes fast and accessible, everyone will be able to customize, and the automation will unlock a flood of new hardware innovation.'
Anna and Azalia's vision for Ricursive is to redefine the chip design process, moving from 'fabless' to 'designless.' Fabless means designing chips without owning expensive fabs, outsourcing production to foundries. Designless means outsourcing not only manufacturing but the entire chip design process, taking an idea and converting it into a manufacturable design.
We envision a future where Ricursive helps any company design chips for its own workloads faster, more efficiently, and more creatively than ever before. By doing so, Ricursive can revolutionize the most valuable resource in our era: compute. We couldn't be more excited to be a part of building a true generational company in the making.
