LangChain Raises $300M Series C at $3B Valuation

Harrison Chase and Ankush Gola built LangChain's commercial platform to accept agents made with other frameworks. That decision gives the San Francisco company's financing story a commercial dimension beyond the popularity of its own open-source code: LangSmith can earn a place in an engineering budget even when another framework gets the developer's vote.

On November 13, 2026, LangChain announced a $300M Series C at a $3B valuation, led by existing investor CapitalG. GIC and DFJ Growth joined returning investors IVP and Sapphire Ventures. The financing backs a platform organized around running, improving and governing AI agents, as enterprises work out how to retain control of systems built with models they buy from somebody else.

LangChain's Series C expands an existing investor relationship

CapitalG first invested in LangChain's Series B and is returning as the Series C lead. Its investment thesis centers on an independent place for companies to manage agents across model providers and clouds. An enterprise may change its preferred model several times while continuing to need a consistent account of what its agents did and how well they performed.

LangChain also named Salesforce Ventures, Insight Venture Partners, QuantumLight and MongoDB Ventures as new investors, with Sequoia, Amplify and Benchmark investing again. Salesforce Ventures separately confirmed its participation and identified the founders as Harrison Chase, CEO, and Ankush Gola, CTO. The investor's interest puts the operating record created around agents inside the commercial argument for backing the company.

The financing follows a $125M Series B announced in October 2025, led by IVP at a $1.25B valuation. LangChain previously announced a $25M Series A led by Sequoia in February 2024 and a $10M seed led by Benchmark in April 2023. Those 4 announced amounts add to $460M, a sum of disclosed rounds rather than a separate claim about the company's cash balance.

A framework can introduce a customer without owning the whole stack

LangChain began as Harrison Chase's open-source side project in late 2022; Harrison Chase and Ankush Gola started the company in early 2023. Its developer offerings expanded through LangGraph and Deep Agents, while LangSmith became the commercial platform. The distinction matters to a buyer deciding which parts of an agent system to assemble internally and which responsibilities to give a vendor.

A framework helps engineers construct the agent. A commercial operating platform has to remain useful after that agent starts encountering real requests, unexpected inputs and changes to the underlying model. The engineering team needs a record it can inspect, examples it can test again and a way to compare behavior across revisions. That work continues after the first implementation decision has been made.

LangSmith's framework-agnostic approach gives LangChain a route to customers who chose different development tools. The business can meet those teams at the point where they need to understand and improve agent behavior. For an open-source company, that is a meaningful expansion of the commercial conversation: familiarity with the code can start a relationship, while responsibility for operating the system can deepen it.

Workday shows where shared infrastructure earns its place

LangChain's November 10 Workday case study describes an internal platform using LangGraph for orchestration and LangSmith for tracing and evaluation. Workday's team wrapped those building blocks in its own SDK and internal services, keeping the surrounding security, data access and governance requirements inside its architecture. Product teams could concentrate more directly on the behavior of the agent they wanted to build.

The case study says the time required to get an agent prototype into staging fell from 4-5 months to under a day. That is a vendor-published account of a specific stage in development. Keeping the staging detail attached to the number makes the result useful to another operator, who still has to account for the approvals and evidence required to release an agent into production.

The commercial significance sits in the repeated work that a platform team can take off another team's desk. Authentication, infrastructure and evaluation become shared capabilities, reducing how much each application team has to assemble before trying a useful agent. A buyer considering LangSmith can ask where that shared responsibility belongs in its own organization, and how much of it the buyer wants a supplier to carry.

Model freedom gives evaluations more commercial weight

LangChain says more than 7,000 organizations rely on LangSmith. That adoption figure describes organizations using the platform; it supplies no independent measure of revenue or retention. The funding case also rests on a continuing operating need: a team changing an agent's model must decide whether the resulting behavior is acceptable for the job.

A more capable model can alter cost, response time and the way an agent chooses tools. Evaluations give the engineering and product teams a way to examine the change against examples that matter to their business. Production traces can supply some of those examples, turning an observed problem into material for the next release decision. The customer has to keep defining what a good result means as the work evolves.

That creates an interesting relationship between model choice and platform commitment. A company may want freedom to change its model supplier while continuing to invest in the evaluation record it has built elsewhere. Salesforce Ventures explicitly points to accumulated traces and learning in its investment argument. LangChain's opportunity grows with the usefulness of that record, including the judgment people put into deciding which failures deserve attention.

The Series C gives Harrison Chase and Ankush Gola more resources to develop the platform around those decisions, though the company has not published a detailed allocation of the proceeds. Inside the customer's engineering organization, another agent will bring another release discussion, with somebody responsible for explaining why its next model, prompt or permission change belongs in production.