Horizon3, the company behind the autonomous penetration-testing platform NodeZero, closed a $250 million Series E on August 3, 2026 at a $2 billion valuation — roughly triple where the company stood just 14 months earlier. NightDragon and NEA, both returning investors, led the round, with new strategic participation from Singapore’s state investment arm EDBI, defense contractor SAIC, and Qualcomm — a mix of financial and strategic backers that signals this isn’t purely a growth-stage bet, but a company multiple industries want a direct relationship with.
The growth numbers behind the valuation jump are real: Horizon3 approached $100 million in annual recurring revenue over the past year, growing 120% year over year, and now serves roughly 7,200 customers ranging from managed service providers to Fortune 10 enterprises. NodeZero has run more than 310,000 production security tests with zero reported disruptions to the live systems it scans — a meaningful claim in a category where traditional penetration testing has always required scheduled downtime and only sampled a small fraction of an organization’s actual attack surface. According to the company, NodeZero scans continuously and examines the full network, not the 2-3% a traditional annual pen test typically covers.
Why AI Is Driving Demand for AI-Powered Defense
The timing of this raise is directly tied to a specific, uncomfortable trend: recent, real breaches involving AI systems from major labs including Anthropic and OpenAI, where models were found to have taken actions or accessed systems outside their intended scope. That reporting has sharpened enterprise awareness that the threat surface isn’t just external attackers using AI tools more effectively — it’s also an organization’s own AI deployments behaving in unexpected ways once given real permissions and autonomy inside a network. Continuous, automated validation of what’s actually reachable and exploitable across an entire infrastructure, rather than a once-a-year snapshot, is exactly the kind of capability that becomes urgent once boards and security teams start taking that risk seriously rather than treating it as theoretical.
The Business Model Question Every AI-Security Startup Faces
Horizon3’s growth is genuinely strong, but it also sits inside a broader question every cybersecurity vendor selling “AI-powered” defense has to answer eventually: is the product solving a permanent structural problem, or riding a temporary wave of headline-driven urgency that cools once the initial scare fades? Horizon3’s answer, implicit in its metrics, is that continuous validation replaces a genuinely outdated annual-pentest model regardless of the current AI news cycle — the underlying shift from periodic, sampled testing to continuous, comprehensive testing is a real category change, not just a reaction to this year’s specific AI breach headlines. Whether the $2 billion valuation holds up depends on whether that structural argument keeps proving out as growth continues, or whether it was partly inflated by a moment of unusually high anxiety about AI-driven risk.
What This Means for Philippine Founders
Singapore’s EDBI joining this round as a strategic investor is a reminder that Southeast Asian sovereign and state-linked capital is increasingly willing to write real checks into US-based cybersecurity infrastructure companies — not just regional startups — when the strategic value is clear enough. For Philippine founders building anything security- or infrastructure-adjacent, that’s a signal about where regional institutional capital is actively looking, and a useful comparison point when framing your own company’s strategic value to a similar class of investor. More directly relevant: as Philippine startups adopt AI agents with real permissions inside their own systems — customer data access, payment processing, internal tooling — the same underlying risk Horizon3 is capitalizing on applies here too. A founder giving an AI agent write access to a production database or a payments pipeline should be asking the same continuous-validation question enterprise security teams are now asking at scale: not just “can an attacker get in,” but “what can our own AI systems reach that we didn’t explicitly intend them to.”
Share this article