In today’s AI-driven world, compute is the new gold, and the race to dominate it is fiercer than ever. From Nvidia’s GPUs to Google’s TPUs, Amazon’s Trainium, and even Elon Musk’s recent revelations about Tesla’s AI chips, one thing is clear: the future belongs to those who can design the most powerful chips for AI workloads. But here’s the kicker: the process of chip design is painfully slow and astronomically expensive, often taking years and costing hundreds of millions of dollars. And this is the part most people miss: it’s not just about money or time—it’s about the untapped potential of what we could achieve if these barriers were broken.
Let’s break it down. Designing a chip at the leading edge (think 5nm or 3nm) can take 18 to 36 months, with costs soaring to $600-650 million per design. A staggering 50-70% of that is human labor, while another 5-15% goes to Electronic Design Automation (EDA) tools, a market dominated by giants like Cadence and Synopsys. These tools, while essential, are clunky and often limit creativity. But what if AI could change all that?
Enter AlphaChip, the project that turned heads by slashing the floorplanning step in chip design from months to mere hours. It wasn’t just a time-saver—it was a glimpse into a future where AI could automate the entire chip design flow, from architecture to verification. Imagine if designing chips took days instead of years. The economic impact? Massive. Reports suggest that even a multi-month delay in chip production could cost companies over $10 billion in lost revenue. Now, flip that: what if chips could be designed faster and shipped earlier? The revenue potential is mind-boggling.
But here’s where it gets controversial: what if AI doesn’t just speed up the process, but fundamentally changes how we design chips? AlphaChip’s designs weren’t just faster—they were different. Instead of the traditional Manhattan grid-like structures, AlphaChip produced organic, nature-inspired shapes. Initially, humans were skeptical, even rejecting these designs. Yet, they went on to shape four generations of Google’s TPUs. This raises a bold question: Are we limiting innovation by relying too heavily on human intuition?
At Sequoia, we’re thrilled to partner with Anna Goldie and Azalia Mirhoseini, the visionary co-founders of Ricursive Intelligence, as they lead their first funding round. These two pioneers, who spearheaded AlphaChip, are now at the forefront of the AI-for-chip-design revolution. Their mission? To build a frontier AI lab that redefines the category. In just weeks, they’ve assembled a team with unparalleled talent density, united by a core belief: chip design is the bottleneck holding back AI, hardware, and infrastructure progress.
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. The automation here will unlock a flood of new hardware innovation.”
Their vision for Ricursive is nothing short of revolutionary: shifting the industry from “fabless” (designing chips without owning fabs) to “designless” (outsourcing the entire design process). Imagine a world where any company, regardless of size, can design chips tailored to their workloads—faster, cheaper, and more creatively than ever before. This isn’t just about improving efficiency; it’s about democratizing access to the most valuable resource of our time: compute.
But let’s not shy away from the debate: Is “designless” the future, or does it risk commodifying a highly specialized field? And what does this mean for the thousands of engineers currently driving chip design? We’d love to hear your thoughts in the comments.
At Sequoia, we’re not just excited—we’re honored to help build a generational company with Ricursive. Because if they succeed, they won’t just revolutionize chip design; they’ll reshape the very foundation of technological progress. The question is: Are we ready for what comes next?
