The Argument in Favor of Tranched Rounds

Stephanie Palazzolo argues that multi-tranche rounds at different valuations, once unusual, are now standard in AI, with hot rounds having three or more tranches. Critics say big VCs push smaller firms into the pricier tranche and founders get a headline valuation even though much of the money came in cheaper. Defenders point to compute needs: one investor says a young startup renting about 2,500 Nvidia chips for two years may need $150M-200M, and tranching avoids either heavy dilution or a $1B+ price that deters return-seeking VCs. Corporates such as Nvidia often take more than 25% of the expensive tranche but avoid leading; one investor said their firm puts most money in the cheap tranche and 'leads' the pricey one with a few million in name only. The newsletter also notes DeepSeek closing a US$14.6B round and OpenAI aiming to close a $30B round at up to $1.25T pre-money, plus deals including Pathos AI ($237M Series E at $3B post, led by Bristol Myers Squibb) and Warp Sciences ($215M at $865M post, led by Lightspeed). Caveat: the sourcing on tranche practices is anonymous investors.