LangChain hits $3B valuation after $300M Series C round

LangChain, the San Francisco start-up behind an open-source framework for building AI agents, has raised a $300 million Series C at a $3 billion valuation, according to TechCrunch on 13 November 2026. TechCrunch reports that the round was led by CapitalG, with participation from Singapore's GIC, DFJ Growth, IVP and Sapphire Ventures, and that the company announced it on Friday.

According to TechCrunch, the new valuation is more than double the $1.25 billion at which LangChain raised $125 million about a year earlier, in a round led by IVP. Fortune reported that earlier round on 20 October 2025, describing it as a Series B.

What is LangChain, and what does LangSmith do?

An AI agent is a software system built on a large language model (LLM) that can reason, use tools and take actions on a user's behalf, instead of only generating text. According to Fortune, LangChain began in late 2022 as an open-source project by Harrison Chase, weeks after OpenAI released ChatGPT. TechCrunch explains that early LLMs struggled with tasks such as searching the web, calling APIs and interacting with databases, and that LangChain's framework handled them.

TechCrunch says the frontier AI labs now deliver harnesses that do those tasks competently, so LangChain turned its focus toward managing AI agents. Its agent creation and management software is called LangSmith. Unite.AI describes it as a platform for operating, improving and governing agents, whatever framework was used to build them.

How big is the customer base?

LangChain claims 7,000 LangSmith customers, according to TechCrunch. The start-ups it names are Harvey, Rippling and Clay, and the enterprises are Nvidia, Toyota and Salesforce. Unite.AI adds the start-up Listen Labs to that list. These are the company's own figures and customer lists, and the sources do not verify them independently.

Unite.AI reports two examples of use. At Clay, the platform tracks hundreds of millions of agent runs each month and supports agent improvement. Workday uses LangSmith across the organisation and, according to the same article, cut the deployment time for a new agent from five months to under a day.

On revenue, the sources contain no 2026 figure. In October 2025, a LangChain spokesperson told Fortune that a TechCrunch estimate from July 2025, putting annual recurring revenue between $12 million and $16 million, was low for where the company stood. Fortune added that the company was not profitable but described itself as fairly efficient in its spending.

What is inside the platform?

According to Unite.AI, LangSmith rests on three parts. Observability, mentioned below, means recording and inspecting what an agent does at each step.

• Agent Runtime covers the path from agent architecture to production infrastructure. Teams use Deep Agents, LangGraph and LangChain to keep model choice, and deploy through LangSmith Agent Server and Sandboxes, in the cloud or in the customer's own environment. LangSmith Managed Agents and Fleet offer a fully managed route.
• Agent Improvement pairs measurement with automated fixes. LangSmith Observability and Evaluation expose agent behaviour and assess quality, while LangSmith Engine, an in-platform agent, looks for errors, produces code and prompt corrections and checks production traffic for regressions. LangSmith Fine-Tuning turns agent traces into training data for specialised models.
• Agent Governance centralises control of models, tools and data. The LLM and Tool Gateways let teams set model budgets and fallback behaviour, safeguard sensitive information and manage tool access and authentication across teams and agents.

Who is investing, and who are the competitors?

Unite.AI says existing investor CapitalG led the round, with significant participation from new investors GIC and DFJ Growth and from existing investors IVP and Sapphire Ventures. Salesforce Ventures, Insight Venture Partners, QuantumLight and MongoDB Ventures also joined, while Sequoia, Amplify and Benchmark increased their participation.

The path to this round was fast. Fortune reports a $10 million seed round led by Benchmark in April 2023, then a $25 million Series A led by Sequoia that valued the company at $200 million. The company has since moved from a developer library to a paid platform.

The market is crowded. TechCrunch says plenty of other start-ups pitch similar advantages and are also doing well, citing newcomer Mintive. In 2025, Fortune named LlamaIndex and Haystack as rivals and noted that OpenAI, Anthropic and Google now offer built-in capabilities that were once LangChain's differentiators. Chase told Fortune that he expects most enterprises to use several agent platforms.

LangChain's pitch, according to TechCrunch, is to let businesses own their intelligence. The company says LangSmith does not lock users into any particular model, cloud or ecosystem, unlike the major clouds or frontier AI labs. That is a positioning claim by the company, not a verified result.

What comes next?

Unite.AI reports that LangChain has scheduled a release event for 20 November 2026, where it plans to share additional LangSmith capabilities. It also says LangChain expects agents built for specific domains to become standalone products and internal operational tools, differentiated by knowledge, data and user experience.

What does this change for a company?

For a business planning to deploy agents, the round suggests that investors now value the layer that monitors, evaluates and governs agents, not only the tools that build them. The Workday example, as reported, points to deployment speed as a real stake, but it is a single customer result supplied by the vendor.

Dependence on one model provider or one cloud is becoming a criterion to check when choosing an agent platform, since neutrality is precisely what LangChain sells. Before committing, a company can ask where its agents' logs and data live, and which budget and access controls it will have. For the broader picture of how agents fit together in an organisation, see our page on agentic systems.