In early October 2026 I joined a threaded X conversation with @HarryStebbings, @venkyganesan, @dunkhippo33 and others about how we think about seed and pre‑seed investing. I argued that seed bets are best treated as option‑style investments and that true‑seed outliers should deliver 1,000× returns, which is far higher than the ~20× uplift we saw from Anthropic’s entry point.
- Seed vs. true seed – True‑seed / pre‑seed deals often have “basically no revenue” and provide little observable signal.
- Decision: Treat seed investments as options and only scale up positions when quantitative evidence shows a startup is likely to become an outlier.
Anthropic’s total uplift from entry point was ~20×, which I do not consider a true‑seed return.
A true‑seed outlier should generate 1000×+ returns.
I linked this thread to my earlier post x october 2026 posts part 1 for the full deal‑flow numbers and filtering pipeline details.
- Decision: Maintain distinct seed and late‑stage entities so LP exposure to high‑risk seed books stays separate.
- Zero correlation – “there is zero correlation w any name brand fund picking well at seed.”
- Option‑bet framing – “Every seed investment is an option bet.” I view each deal as buying a call option on a potential outlier.
- Portfolio construction – Build a broad set of seed bets to increase the probability of hitting a 1,000× outlier.
- Fund structure – Many funds run separate late‑stage vehicles for the same LPs to keep early‑stage risk isolated.