This milestone comes approximately five years after crossing $1 million ARR in September 2021, and reflects sustained growth as organisations adopt continuous, production-safe, autonomous penetration testing to prioritise and remediate truly exploitable risk amid accelerating AI-driven threats.

“The journey to $100M ARR required three things: investing in the channel early, methodically evolving from a single product into a platform, and navigating economic cycles via operational excellence that balanced growth and efficiency. $100M ARR is just the first mile marker in our marathon,” said Snehal Antani, co-founder and CEO of Horizon3.

The company began by solving one of the hardest problems in the domain: autonomous pentesting of internal production systems. Accomplishing this provided access to high-value cyber training data residing behind the firewall. From there, Horizon3 expanded attack surface testing to external, cloud, Kubernetes, and web applications, all conducted safely in production.

In parallel, it expanded into adjacent budgets affected by comprehensive pentesting, including auto-deploying deception based on identified attack paths, driving CTEM workflows, and now, agentic remediation workflows via blue-team agents. Deception is a resurging category given its effectiveness against agentic attackers.

Advancing attack-defend AI learning loops

“The exploit window is collapsing. Frontier models are turning newly disclosed weaknesses into working attack paths faster than traditional testing cycles can respond. In order to remain ahead of that shift, we are investing in three areas,” Antani continued. “First, we are expanding sovereign cyber AI regions so organisations where data residency is required can run the same high-frequency, in-scope testing that more than 7,000 customers already use. Second, we are building cyber resilience centres of excellence across critical infrastructure so the closed-loop, hack, fix, verify, repeat model operates at the scale and auditability those environments require. Third, we are advancing attack-defend AI learning loops so every production test improves the next one, turning the volume of real-world attack paths already generated across hundreds of thousands of tests into a durable advantage.”

Horizon3’s focus on enabling cybersecurity professionals, IT administrators, and network engineers to “hack themselves” has been critical to its growth. Customers can prioritise exploitable vulnerabilities, fix issues that truly matter, and verify the effectiveness of their security controls.

Continuous exposure management practice

Horizon3’s self-service experience and predictable subscription licensing model has enabled approximately 80% of customers to shift from annual to monthly pentesting, with 40% shifting to weekly. This frequency of use accelerates their transition to continuous exposure management practices, enabling customers to prove their cyber resilience over time.

“The goal is simple: continuous, customer-controlled offensive capability that matches the speed of AI-driven threats without sacrificing the safety and observability production systems demand. Our ability to safely pentest large-scale production systems in a repeatable, auditable, and observable way, while operating within customer-defined scope and guardrails, sets us apart from every other solution and lab in the market. As frontier models advance, they will create a massive tailwind for us and accelerate the shift to continuous, comprehensive pentesting,” Antani added.

Horizon3’s Nodezero platform has been used to conduct more than 350,000 production-safe tests across more than 7,000 organisations globally, including Fortune 10 enterprises, multinational banks, major healthcare networks, and government agencies. The company is FedRAMP High authorised and helps customers meet a range of regulatory requirements.