How Ricursive Intelligence’s Founders are Using AI to Shape The Future of Chip Design

Watch on YouTube ↗  |  January 14, 2026 at 20:36  |  37:00  |  Sequoia Capital
Speakers
Azalia Mirhoseini — Co-founder, Recursive Intelligence
Anna Goldie — Co-founder, Recursive Intelligence

Summary

Anna Goldie and Azalia Mirhoseini, founders of Ricursive Intelligence, discuss how chip design has become a bottleneck for AI progress and how AI can automate and accelerate the entire chip design process. They describe their AlphaChip work at Google, which designed multiple TPU generations, and their vision of a 'designless' industry where more companies can create custom silicon. The conversation covers recursive self-improvement, competition with EDA incumbents, and the potential for a Cambrian explosion of custom chips.

  • Chip design is described as a critical bottleneck for AI progress.
  • AlphaChip used reinforcement learning for chip floorplanning and was adopted across TPU generations.
  • Ricursive Intelligence aims to apply AI to end-to-end chip design.
  • The founders envision a shift from fabless to 'designless' custom silicon.
  • AI-designed chips could accelerate AI in a recursive self-improvement loop.
  • Incumbent EDA players Cadence and Synopsys are adding AI, but Ricursive targets end-to-end reimagination.
  • The company plans its first product within a year.
  • The discussion includes optimistic long-term AGI and productivity implications.
Ideas
Anna Goldie Co-founder, Recursive Intelligence 13:45
AlphaChip boosts Google TPU design.
AlphaChip's reinforcement-learning floorplanning was adopted across multiple generations of Google TPUs, expanding into more chip area and achieving increasingly superhuman performance, which demonstrates Google's custom silicon design advantage.
GOOG MED
Azalia Mirhoseini Co-founder, Recursive Intelligence 16:17
AI design enables custom silicon explosion.
Chip design is the bottleneck for AI progress because chip design cycles are much slower than model cycles. AI-driven design can compress the process and enable a 'designless' paradigm where companies with large workloads create custom silicon without huge in-house design teams, leading to a Cambrian explosion of custom chips and more efficient AI compute.
SMH HIGH
Azalia Mirhoseini Co-founder, Recursive Intelligence 21:54
AI-native rivals threaten EDA duopoly.
Incumbent EDA duopoly Cadence and Synopsys are adding AI as point solutions, but an AI-native frontier lab like Ricursive aims to reimagine and co-optimize the entire chip design flow end-to-end. This creates a competitive disruption setup for the incumbents that investors should monitor.
CDNS SNPS MED
Up Next
4:40
OBBB taxes and deregulation are the big drivers of market breadth, says Morgan Stanley's Jim Caron
CNBC · Jan 14
1:52
Jon Stewart Jokes About Fed in Surveillance Shout-Out
Bloomberg Markets · Jan 14
4:36
Blistering Metals Rally Sends Gold, Silver and Copper to Records
Bloomberg Markets · Jan 14
5:43
Big Banks See Record Lending Haul as Low Yield Assets Roll Off
Bloomberg Markets · Jan 14
4:57
Charlotte Mayor: All about making sure community and business are aligned to be a place for everyone
CNBC · Jan 14

This Sequoia Capital video, published January 14, 2026, features Anna Goldie, Azalia Mirhoseini discussing GOOG, SMH, CDNS, SNPS. 3 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Anna Goldie, Azalia Mirhoseini  · Tickers: GOOG, SMH, CDNS, SNPS