HiddenLayer has raised $100 million in Series B funding as enterprises spend more heavily on protecting AI models, agents and the tools connected to them.
Delta-v Capital led the round, joined by Ten Eleven Ventures, Morgan Stanley, Microsoft’s M12, Booz Allen Hamilton and other investors. The Austin-based company plans to put the capital into sales and distribution while continuing to expand engineering and research. It also intends to grow in Europe and the wider EMEA region.
The financing arrives three years after HiddenLayer’s $50 million Series A, when it was less clear whether attacks on AI systems would become frequent enough to support a large security market. The company’s latest growth claims suggest that buyers now see the problem as operational rather than hypothetical, although the figures reported by the company were not independently verified in TechCrunch’s article.
Enterprise adoption changes the security question
HiddenLayer co-founder and CEO Chris Sestito said annual recurring revenue increased more than tenfold over the past year. He declined to disclose the exact amount, placing it only in the tens of millions of dollars, and said new customers accounted for more than 90% of that growth.
Financial services companies and large technology businesses building AI products are its biggest customer groups. HiddenLayer also has contracts with the US Department of Defense and the intelligence community. Sestito referred to another customer as a leading frontier-model provider with more than 700 million weekly users, but did not name it. TechCrunch speculated about possible identities without confirming one.
The broader market is also expanding. Gartner estimates companies will spend $2.83 billion this year on products for securing AI tools, up 83% from 2025, and projects spending of almost $4.78 billion next year. Those are forecasts, not guaranteed market outcomes, but they explain why established security vendors and startups are pursuing the category.
Protection expands from models to agents
HiddenLayer still sells the core capabilities it offered in 2023: discovery, runtime protection, attack simulation and supply-chain security. What has changed is the scope. The company has extended those products to address prompt injection, agent manipulation and malicious use of tools connected to AI systems.
Sestito argued that inference remains the common technical activity whether the target is a traditional machine-learning model, a generative model or an agentic workflow. In his account, HiddenLayer did not need to abandon its original approach, but it did need to apply it across more kinds of deployment.
Runtime security has become a particular priority as businesses put AI into production. Sestito compared the emerging requirement with endpoint detection and response, except focused on AI behavior. The analogy expresses the company’s product strategy; TechCrunch does not provide comparative test results showing how well its controls perform.
Open-source models create another attack surface. HiddenLayer says it parses and scans roughly 50 AI file frameworks to detect models that are not what they claim to be, including concealed models inside other model files.
A growing market brings larger competitors
The new capital also reflects competitive pressure. Startups Noma and Zenity have each raised more than $100 million for adjacent or overlapping areas. Large cybersecurity companies including Cisco, Palo Alto Networks and Check Point often acquire technology rather than build every capability internally, according to TechCrunch.
Sestito acknowledged that platform providers such as Microsoft, OpenAI and AWS could eventually bundle parts of AI security into their own products. He expects those platforms to emphasize governance functions such as discovery, identity and policy controls, leaving room for HiddenLayer’s specialist protections.
That division is a company expectation, not a settled market boundary. HiddenLayer must still demonstrate that its rapid revenue growth can become a durable business before rivals absorb or reproduce its capabilities. The Series B gives it resources to pursue that test at a moment when agent security is moving from forecasts and demonstrations into production budgets.
