LangChain, whose software helps businesses build AI agents, has raised $300 million at a $3 billion value, it announced on Nov. 13. AI agents are AI that can carry out tasks on their own, such as using apps and websites, instead of only answering questions. CapitalG, the growth fund of Google's parent company Alphabet, led the round. CapitalG first invested in LangChain's previous round, and it was joined by new investors GIC and DFJ Growth and existing investors IVP and Sapphire Ventures. In October 2025, LangChain raised $125 million at a $1.25 billion value, so its value has more than doubled in about a year.
LangChain began in late 2022 as a side project of Harrison Chase, now its chief executive. It started as open-source code, free for anyone to use, study, change and share, which developers use to build programs on top of AI models. CapitalG says those open-source projects have passed 1 billion downloads and, counting a newer one called Deep Agents, now see more than 350 million a month. At the time of the last round, its two main packages had about 90 million a month combined. Alongside that free code, LangChain sells LangSmith, a platform it says lets companies run, improve and govern their agents, with a free plan for one person. LangChain says more than 7,000 organizations use LangSmith, including NVIDIA, Toyota and Salesforce, whose investment arm also joined this round. It says Workday used LangSmith to cut the time to put a new agent to work from five months to under a day.
CapitalG's case for LangChain starts from its view that an AI model on its own doesn't give any company an edge for long. CapitalG says that every company can now reach the same leading models through the same online services. In its view, what makes an agent good at a job is mostly what surrounds the model: how the agent plans and uses tools, the business knowledge it is given, the tests that show whether it does the job well, and what it remembers. CapitalG says LangSmith works with any AI maker's models and, for companies on its custom-priced Enterprise plan, can run inside a customer's own Amazon, Microsoft or Google cloud account. It argues that companies that can swap models without rebuilding their agents, and can show with tests that nothing got worse, get to use the best new model the day it ships.
Even LangChain's lead investor says the competition is strong. CapitalG says the companies that make AI models and the big cloud companies are all shipping their own tools for building agents, and that much of that work is very good. It says LangChain's customers tell it that businesses using several AI makers and clouds want one neutral place, under their own control, to manage all their agents. LangChain says it will show new LangSmith features at an event on Nov. 20.
LangChain has not said how much money it makes or whether it makes a profit, and the $3 billion figure is the price its investors agreed to in this private round, not a value set by a stock market. According to a news report, the company has 435 full-time staff, plans to grow to 850 by the end of 2027, and will spend the money on expanding LangSmith and hiring engineers and salespeople.