Why I Backed OpenAI, AGI, & the Zero to One Team | Vinod Khosla

Video link :

Notes :

(00:00–02:56) Great founders constantly ask what they can improve. Vinod sees himself as a venture assistant, not a capitalist—helping startups where they lack experience. He stresses the importance of values alignment between founders and investors.
(03:30–05:09) The team you build is the company you build. Licensing tech doesn’t replace team-building. Example: convincing Andy Bechtolsheim and Bill Joy to join Sun Microsystems.
(08:32–10:22) Right teams matter more than picking the perfect initial market. The best teams will iterate toward the right answer.

(17:09–18:45) Early stages are for exploration—don't lock into a path too soon. Keep burn low and iterate until you find direction.
(18:45–20:48) “Don’t make a plan; plan to plan.” Update your plan every 3 months as you learn and discover what's important.

(20:48–22:27) Khosla critiques “founder-friendly” VCs who don’t challenge strategy. Founders need thoughtful criticism, not comfort.
(22:27–24:23) Set a culture with your investors that invites hard questions—focus on what’s not going well.

(25:34–31:31) Khosla invested in OpenAI in 2018 when it had no product, plan, or revenue—he saw exponential potential like early TCP/IP. Sent an apology letter to LPs acknowledging the unusual bet.
(31:31–36:13) Also backed Commonwealth Fusion and other radical tech bets. Believes most people underestimate exponentials.

(36:13–38:06) Founders should identify 3 things they uniquely believe, even if others disagree. Avoid following conventional VC advice too closely.
(38:06–40:07) Advocates for “safe math” as a method for AI safety, citing discussions with Yoshua Bengio.