AI Security Becomes Billion-Dollar Market, Attracts $100 Million

The race to protect artificial intelligence models from attacks has just gained another significant chapter. HiddenLayer, an Austin-based startup focused on security for AI models, agents, and workflows, has closed a Series B round of $100 million led by Delta-v Capital, with participation from names like Morgan Stanley, M12 (Microsoft's venture arm), and Booz Allen Hamilton.

This investment reflects something that data has already been signaling. According to estimates from Gartner, companies are expected to spend $2.83 billion on security products for AI tools by 2026, an 83% jump from the previous year. The expectation is that this figure will approach $4.78 billion by 2027.

Three years ago, when HiddenLayer raised its Series A of $50 million, the big question was whether the risks against AI systems would actually materialize on a scale sufficient to justify a dedicated market. The answer came faster than many expected.

HiddenLayer's CEO and co-founder, Chris Sestito, revealed that the startup's annual recurring revenue has multiplied by more than ten times in the last 12 months. Although he did not disclose the exact number, he stated that the ARR is now in the "tens of millions of dollars." More than 90% of this growth came from new clients who signed contracts in the past year.

The company's largest verticals are financial institutions and large tech companies that are building AI products. The startup also maintains contracts with the Department of Defense and the U.S. intelligence community. Among its clients is a "leading provider of frontier models" with "over 700 million weekly users," pointing to one of the big names in the industry.

For those following the technology ecosystem, this growth signals that AI security has ceased to be a speculative niche and has solidified as essential infrastructure.

HiddenLayer's portfolio still revolves around four pillars: model discovery, runtime protection, attack simulation, and supply chain security. The difference from 2023 is the scope. The startup needed to expand each of these products to cover threats such as prompt injection, manipulation of autonomous agents, and malicious use of tools connected to models.

"Inference remains inference. Whether in a traditional machine learning model, in generative AI, or in an agentic workflow, much of our technology still applies. We don't need to pivot, but we need to expand our scope," Sestito explained.

Runtime protection, specifically, has become a priority. Sestito compared the solution to traditional endpoint detection and response (EDR) tools, but applied specifically to the realm of AI. It's an analogy that makes sense: just as endpoints needed dedicated protection in the cloud computing era, AI models require constant vigilance against manipulations and unexpected behaviors in production.

One of the most relevant points raised by HiddenLayer concerns open-source models. The startup analyzes and scans about 50 different AI file frameworks to ensure that the tool the developer is using is exactly what it claims to be.

"We are seeing models that present themselves as one thing but are another. Models hidden within models," Sestito stated. This type of threat is particularly relevant because the adoption of open-source models has grown rapidly. Platforms like Hugging Face host thousands of models that any developer can download and integrate into their systems.

The risk is not hypothetical. As companies integrate third-party models into critical business flows, verifying the integrity of these models becomes as important as code auditing in traditional software libraries. It's an analogous problem to what the smart contract ecosystem in DeFi faces with dependencies on unaudited code.

HiddenLayer is not alone in this market. Startups like Noma and Zenity have each raised over $100 million to tackle adjacent or overlapping areas of AI security. At the same time, cybersecurity giants like Cisco, Palo Alto Networks, and Check Point historically prefer to acquire smaller companies rather than build solutions from scratch.

Sestito acknowledged that some features that HiddenLayer offers may eventually be incorporated into the platforms of companies like Microsoft, OpenAI, and AWS. However, he believes that the AI infrastructure of these platforms will lean more towards governance, identity, and policy control features than the type of runtime protection that his company builds.

The Series B funds will primarily be directed towards sales and distribution, as well as expansion into Europe and the EMEA region. The engineering and research team is also expected to grow, but the priority now is to commercially scale what already works.

The surge in investments in AI security reflects a known pattern in technology adoption cycles. First comes the race to implement. Then, the race to protect. This happened with cloud computing, mobile devices, and blockchain infrastructure. Now it is happening with AI.

The difference is the speed. The cloud security market took nearly a decade to mature. The AI security market is forming in two to three years, driven by accelerated corporate adoption of generative models and autonomous agents.

For investors and technology managers, the message is clear: the cost of security is not optional. It comes along with deployment. And whoever solves this problem first, at scale, will capture a significant share of a market that is expected to exceed $4.7 billion by 2027.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.