Infisical Raises $40M at $600M Valuation to Secure Secrets for AI Agents
Key Takeaways
- Infisical $40M Series B, announced on 27 October 2026, values the San Francisco security company at $600 million.
- The round was led by investor Elad Gil, who also led the company’s $16 million Series A in June 2025, with Gradient and Y Combinator among the other investors.
- Infisical provides infrastructure for managing secrets such as API keys, passwords, and authentication tokens, plus certificates, encryption keys, and privileged access.
- The company says it serves more than 1,300 customers and 300,000 developers and secures 10 billion secrets every day.
- Its Agent Vault product, currently in preview, lets AI agents use enterprise systems without ever receiving the underlying credentials.
- The funding will go toward product development, hiring across engineering and go-to-market teams, and expanding security infrastructure for AI agents.
Infisical has raised $40 million in Series B funding at a $600 million valuation, betting that AI agents create a security problem most companies have not yet solved. Agents need access to source code, databases, cloud infrastructure, and other critical systems to do useful work. That access requires credentials: API keys, passwords, and authentication tokens. Handing those credentials to an agent is dangerous, because an agent can leak them in a breach, send them to a language model provider, or expose them through a connected tool. Infisical’s answer is to give agents access without giving them the keys.
The round was announced on 27 October 2026 and was led by Elad Gil, who previously invested in the company’s Series A, with Gradient and Y Combinator among its other investors. The company plans to use the capital to accelerate product development, expand its engineering and go-to-market teams, and build out security infrastructure for AI agents as enterprise adoption increases. The valuation is a sharp step up from the $16 million Series A in June 2025, also led by Gil, and it reflects how quickly agent security has moved from a niche concern to a funded category.
Why Agents Break Traditional Secrets Management
Secrets management is not a new problem. Developers have always needed to store API keys and passwords safely, rotate them when they leak, and control who can use them. Tools for that job already exist, and Infisical itself started there. Its platform manages sensitive credentials, keeps digital certificates renewed and trusted, and controls privileged access to enterprise infrastructure, including databases, cloud environments, and source code repositories.
AI agents change the problem in three ways. First, they act autonomously. A human developer uses a credential deliberately, for a specific task, and can be asked why. An agent may use dozens of credentials in a single workflow, at machine speed, without a human watching each step. Second, they operate at scale. A company might run hundreds or thousands of agents, each needing different access, and the number of credentials multiplies with them. Third, they talk to outside systems. An agent that sends a prompt to a language model provider, or data to a connected application, can carry credentials with it in ways a human never would. A password pasted into a prompt is a password disclosed to a third party.
Co-founder and chief executive Vlad Matsiiako states the principle simply. You would not give an employee unrestricted access to every system in your company, and AI agents should not be any different. Agents need access to company systems to do their jobs, but giving them actual passwords and API keys is a major security risk. The company says its Agent Vault lets agents get their job done safely without ever having those credentials in their hands.
That framing is useful because it reframes agent security as an identity problem. Agents are a new class of enterprise user. Like employees, they need accounts, permissions, and audit trails. Unlike employees, they cannot be trained, cannot be fired in the usual sense, and cannot be expected to notice when something looks wrong. The infrastructure that manages their access has to be stricter than the infrastructure for humans, not looser, because agents will make mistakes at a scale and speed that humans cannot match.
How Agent Vault Works
Agent Vault is the centrepiece of Infisical’s agent strategy. It allows AI agents to interact with APIs and enterprise services without directly receiving sensitive credentials. The system acts as a forward proxy: when an agent needs to call a service, the Vault applies the credential during the request itself, while keeping the secret out of the agent’s environment. Companies define which systems each agent can access, and the agent cannot leak credentials in a cyberattack or inadvertently send them to language model providers or other tools it communicates with, because it never holds them.
The design has two practical advantages. The first is compatibility. The company highlights that Agent Vault works with standard HTTPS proxy configurations, which means teams can adopt it without rebuilding their agent infrastructure. The second is least privilege by default. Instead of giving an agent a broad API key that works everywhere, a company can scope access to exactly the systems and actions that agent needs. If the agent is compromised, the damage is limited to that scope.
Agent Vault is currently in preview, which is an important qualification. The broader Infisical platform is established, with more than 1,300 customers and 300,000 developers, and it secures 10 billion secrets every day, according to the company. Named customers include Airbus, Volkswagen, Hugging Face, OpenRouter, Lucid, First American Financial Corporation, Wise, and Clay. Its open-source project has more than 30,000 GitHub stars. But Agent Vault itself is newer, and its production track record is shorter. Buyers evaluating it should ask for deployment details and incident history, not just architecture diagrams.
That caution applies to the whole agent security category. The idea is sound, and the threat is real, but the products are young. A proxy that holds credentials is itself a high-value target, and its security matters as much as the security of the agents it protects. Infisical’s pitch is that centralizing secrets in a hardened vault is safer than scattering them across agent environments. That is a reasonable argument, and it is the same argument that made secrets managers standard for human developers. Whether it holds for agents at scale is something the next year of deployments will test.
A Funded Category, Not a Single Bet
Infisical is not alone in this market, and the funding data shows it. Dealroom’s coverage of the round notes that identity startup Oak emerged with $60 million, while NewCore raised $66 million to manage agent credentials, as part of a broader wave that pushed cybersecurity funding to $6.8 billion in the third quarter of 2026. Infisical’s $600 million valuation puts it among the larger bets on secrets management as core infrastructure for the agent era.
That context cuts both ways. On one hand, multiple funded competitors validate the problem. When several well-backed companies attack the same gap at the same time, the gap is usually real. On the other hand, competition means Infisical must differentiate. Its advantages are an existing secrets platform with real customers, an open-source community of significant size, and a product that extends that platform to agents rather than starting from zero. Its challenge is that agent identity is attracting startups, incumbents, and cloud providers simultaneously, and the eventual standard may come from any of them.
For readers who want the background on why agent access is so sensitive, our article on 1Password’s AI agent login explains how agents sign in without ever seeing a password, which is the same problem from the authentication side. Our explainer on what AI agents are and how they work covers the basics of agent autonomy, and our piece on Daytona’s agent cloud looks at the infrastructure agents run on. Infisical sits at the intersection of all three: agents need a place to run, a way to sign in, and a way to use credentials without holding them.
What the Funding Signals
The round’s structure is informative. Elad Gil led both the Series A and the Series B, which signals conviction from an investor who has watched the company operate for over a year. Gradient and Y Combinator, both existing investors, also participated. That pattern, insiders leading a follow-on round at a much higher valuation, usually means the company’s metrics supported the markup. The company says it serves more than 1,300 customers and 300,000 developers and secures 10 billion secrets daily. Those are the numbers behind the valuation, and they are company-reported figures, not audited results. Still, a customer list that includes Airbus, Volkswagen, and Hugging Face is not easy to fabricate, and it suggests the platform is genuinely deployed in serious environments.
The use of funds is also telling. Product development, hiring across engineering and go-to-market, and expanding agent security infrastructure is the standard formula for a company moving from a proven product to a new product line. The risk is execution: building Agent Vault into a production-grade product while maintaining the core secrets platform, and selling it to enterprises that are still figuring out their own agent strategies. Many of those enterprises do not yet have agent governance policies, which means Infisical must sell both the product and the problem it solves.
There is a larger point here about where value accrues in the agent era. Much of the attention goes to the models themselves and to the agents built on top of them. But agents that cannot be trusted with access cannot be deployed in real enterprises, no matter how capable they are. Security infrastructure is what turns a demo into a deployment. Secrets management, identity, permissions, and audit trails are unglamorous, but they are the difference between an agent that a company can actually use and one that stays in a pilot forever. Infisical’s $600 million valuation is a bet that this layer matters as much as the models it protects.
The honest summary is that Infisical has raised serious money, from serious investors, to solve a problem that is clearly real and growing. Its platform has scale, its Agent Vault has a sound design, and its market has momentum. The open questions are whether Agent Vault proves itself in production, whether the company can stay ahead of a crowded field, and whether enterprises adopt agent governance fast enough to buy what it sells. Those questions will be answered by deployments, not by valuations, and the next year will be revealing.
