HiddenLayer's $100M Series B Signals AI Security Has Become a Real Market, Not a Bet

When HiddenLayer raised its $50 million Series A three years ago, the central question hanging over the deal was whether AI-specific threats existed in enough volume to justify a standalone security category. Real-world examples of attacks against AI systems at scale were hard to come by. That uncertainty has evaporated.

HiddenLayer, the Austin-based startup that protects AI models, agents, and workflows from adversarial attacks and malicious code injections, has now closed a $100 million Series B. Delta-v Capital led the round, with Ten Eleven Ventures, Morgan Stanley, Microsoft's M12, and Booz Allen Hamilton among the participants. The capital arrives as the underlying market has expanded fast enough to reshape the competitive landscape almost overnight.

Gartner estimates enterprises will spend $2.83 billion this year securing AI tools, an 83% jump from 2025. The research firm expects that figure to climb to nearly $4.78 billion in 2027. That kind of growth curve explains why cybersecurity vendors, both venture-backed startups and public company incumbents, are racing to stake out territory before the category matures and consolidates.

HiddenLayer's own numbers track that trajectory closely. CEO and co-founder Chris Sestito told TechCrunch that annual recurring revenue grew more than 10x over the past year, landing in the "tens of millions" of dollars. More than 90% of that growth came from new customer acquisition rather than expansion within the existing base, a distinction that matters. It suggests demand is broadening across the market rather than concentrating in a handful of accounts that happened to bet early on the company.

The Customer Base and Product Evolution

Financial services firms and large technology companies building AI products currently make up HiddenLayer's biggest verticals. The company also holds contracts with the Department of Defense and the intelligence community, a foothold that tends to signal durability given the procurement cycles involved. One customer, described only as a "leading frontier model provider" with more than 700 million weekly users, points strongly toward OpenAI or Anthropic, though HiddenLayer hasn't confirmed the identity.

The product itself hasn't pivoted so much as it has widened in scope. Sestito described the core offerings, discovery, runtime protection, attack simulation, and supply chain security, as largely unchanged since 2023. What's shifted is the threat surface they now need to cover: prompt injection, agent manipulation, and malicious tool use have joined the list alongside more traditional adversarial attacks on machine learning models.

"Inference is still inference," Sestito said. "So whether it's on a traditional machine learning model, whether it's GenAI, whether it's an agentic work stream, a lot of our technology still applied. So really, we haven't had to pivot, but we've had to grow our scope." He likened the company's runtime security approach to endpoint detection and response tools in traditional cybersecurity, just retooled for AI systems.

One growing concern Sestito flagged: open source and open-weight models being tampered with in ways that aren't obvious to the end user. HiddenLayer parses and scans roughly 50 different AI file frameworks to verify that a model is actually what it claims to be. "We're looking at things like models purporting to be one thing, but they're another, hidden models inside of models," he said. That's a narrow but genuinely novel risk vector, and one that didn't exist as a commercial concern when the company raised its Series A.

The new funding will go primarily toward sales and distribution, according to the company, alongside continued investment in engineering and research. HiddenLayer also plans to expand into Europe and EMEA, a logical move given that AI governance regulation is advancing faster on that side of the Atlantic than in the U.S.

Key Risks

The competitive picture is where this story gets more complicated for HiddenLayer's long-term prospects. Large cybersecurity incumbents, Cisco, Palo Alto Networks, and Check Point among them, have shown a clear preference for acquiring AI security capability rather than building it in-house. Cisco bought Robust Intelligence. Palo Alto Networks acquired Protect AI. Check Point picked up Lakera. That pattern suggests HiddenLayer's most likely exit path, and its most immediate competitive threat, may come from the same buyers currently sitting on the sidelines.

Startups working adjacent or overlapping ground have also raised significant capital recently. Noma Security pulled in $100 million to push adoption of AI agent security. Zenity raised $125 million in a Series C. AIR raised $50 million specifically to help companies vet the third-party skills and add-ons AI agents rely on. The category is getting crowded quickly, and differentiation will matter more as budgets get allocated.

There's also a platform risk worth watching. Sestito himself acknowledged that some slice of what HiddenLayer sells could eventually get bundled into the infrastructure layers built by Microsoft, OpenAI, and AWS. His bet is that those platform players will gravitate toward governance features, discovery, identity, and policy controls, rather than the deeper security tooling HiddenLayer specializes in. That's a reasonable distinction today. Whether it holds as hyperscalers expand their own security stacks is an open question.

The Bottom Line

HiddenLayer has converted early skepticism about market size into a demonstrated growth story, with 10x ARR expansion and a customer roster that includes frontier model providers and government agencies. The $100 million round gives it runway to scale sales motion and push into Europe before the market consolidates around a smaller number of winners. But the company's stated ambition, to "scale vertically alongside artificial intelligence" and eventually expand horizontally into AI-dependent cybersecurity, will require it to outrun both a wave of well-funded startups and acquisitive incumbents who have already shown they'd rather buy this capability than build it. The market has proven itself real. Whether HiddenLayer remains the category's independent leader, or becomes its next acquisition target, is the question this raise doesn't answer.