LangChain CEO Harrison Chase on AI Agent Economics and Funding
This episode of 20VC features LangChain Co-founder and CEO Harrison Chase, discussing the difficulties of building AI agents, whether model labs will eat the agent stack, the business value of model routing, and the future of the inference market. Chase also shares views on whether founders should take valuation discounts for top-tier VCs and his hypothetical choice between investing in OpenAI or Anthropic.
ImportanceMaterialEvidenceE2 unreplicatedWrite-upQuick
This episode of The Twenty Minute VC (20VC) features Harrison Chase, co-founder and CEO of LangChain. LangChain helps developers build, test, and deploy AI agents and is backed by Sequoia Capital, Benchmark, and IVP, with its last round valuing the company at $1.25 billion. Chase delves into several core controversies in the current AI infrastructure landscape.
According to the agenda, Chase first analyzes why it remains so hard to build reliable AI agents and discusses whether large model labs will consume the entire agent stack. Addressing the market's focus on 'Model Routing,' he explores whether this area represents a real business with increasing value or if it will ultimately become a commodity. He also touches on why AI memory has not yet been solved and why AI bills are exploding even as token prices drop.
On the investment and strategy front, Chase addresses concerns about competition from Chinese AI models and shares where he would not be investing today. Regarding startup funding, he specifically discusses whether founders should accept valuation discounts to secure backing from top-tier firms like Benchmark or Sequoia. Finally, the episode covers a hypothetical topic: how investors might choose between OpenAI at a $1.5T valuation and Anthropic at $2T.