Fault Lines · 12 November 2026
At the source: Fault Lines Podcast ↗ · video
In this episode of Fault Lines, Augustin Sayer sits down with Kanu Gulati, partner at Khosla Ventures, who was speeding up chip design software on NVIDIA GPUs years before the AI wave. An NVIDIA graduate fellow during her PhD, she went from Intel's research labs to Harvard Business School, a cold email to Vinod Khosla and an early role at the GPU database startup now called Heavy.ai , before joining Khosla Ventures in 2017. They talk about how Khosla invests by theme and treats every startup as a sequence of risks to retire, and why Kanu believes in large language models but backs the companies fixing their limits: formal methods that make AI provably correct, and the data recipes behind robotics foundation models. She also shares how she judges deep tech founders, what she sees in European and Indian startups, from Nomagic's warehouse robots in Warsaw to Sarvam's sovereign AI and ten-minute grocery deliveries, why cold emails still work, and why robots will go where human labor can't.
As published with the episode at Fault Lines ↗
No summary of this episode yet.