韩媒专栏复盘 Palo Alto AI 安全逻辑，引用的均为数月前已公布的 FY2026Q4 财报与并购整合数据，无新增事实

As artificial intelligence evolves from conversational models into agents that carry out tasks on their own, the cybersecurity market is entering a new growth phase. Palo Alto Networks has identified three shifts reshaping the security market: the spread of AI agents, more sophisticated exploitation of vulnerabilities and the wider availability of open models. The company's view is that as more businesses adopt their own AI systems, the volume of data and systems requiring protection grows, lifting demand for security along with it.
The focus of attacks is moving from finding vulnerabilities to how quickly attackers can break in and defenders can respond. As AI agents multiply, companies must manage the identities and access rights not only of people but also of AI. That is why demand is rising for integrated security platforms that pull data from multiple systems to detect threats and respond automatically, rather than for standalone security products.
That trend is already showing up in earnings. Results for the fourth quarter of fiscal 2026, covering May through July, beat market expectations. Remaining performance obligations, or contracted revenue yet to be recognized, reached $21.2 billion, while next-generation security annual recurring revenue hit $9.1 billion — both record highs. Next-generation security ARR rose by $1 billion in a single quarter. The company also signed 220 new contracts bundling multiple security functions into a single platform, and revenue retention among those customers topped 120%.
AI-specific security products are growing quickly as well. Prisma AIRS surpassed $100 million in ARR within four quarters of its launch. Its customer count rose 2.7-fold to 800 from 300 in the previous quarter. The product combines controls on requests coming into AI services with monitoring of AI agents' identities and behavior on a single platform.
The effects of acquisitions are also becoming visible. ARR at Chronosphere, which monitors system performance in real time, has grown to more than $500 million from $85 million at the time of the acquisition. The integration of identity security firm CyberArk is running three to six months ahead of the original schedule. In effect, the company is successively securing the data monitoring and identity management capabilities needed in the AI era.
One variable is the strengthening of security features by AI developers themselves. Even as the reasoning ability of general-purpose AI improves, actual defense requires accumulated threat intelligence and internal corporate data. If the spread of AI expands not only attack capabilities but also demand for defense, Palo Alto's integrated security platform strategy is likely to serve as a medium- to long-term growth driver.