Here’s a detailed look at the best-performing sector this year: security software. Even with last night’s big drop in U.S. stocks, this sector managed to buck the trend and stay strong. The logic behind security software’s strength lies in this: AI Agents replace some 'software execution seats,' but security software controls 'identity, permissions, data, traffic, runtime behavior, and recovery capabilities.' The more autonomous the Agent becomes, the more critical the control layer becomes. So, holding security SaaS stocks means three things are happening simultaneously: 1) AI Agents, cloud migration, and multimodal data are significantly expanding the attack surface (new risks like data breaches, model poisoning, prompt injection, etc.), driving the overall security market to continue growing, with AI bringing genuinely new budgets; 2) Budgets are shifting from traditional point tools to platforms, identity, cloud, data, and runtime; 3) Enterprises deploying AI must simultaneously strengthen 'zero trust,' cloud security, endpoint detection (EDR), and AI security platforms. AI isn’t just about adding a new security product—it’s about rebuilding an entire AI control system for the enterprise. This control system includes: identity and permissions, data access, cloud and workloads, networks and APIs, model and Agent runtime, SOC detection and automated response, business recovery, and rollback. Over the next three to five years, the broad AI-related security budget could reach hundreds of billions, even close to a trillion dollars. I was probably one of the earliest to point out that security software is the most undervalued software sector. Back in late January, I talked about it here: https://((x.com))/qinbafrank/status/2017528068324397416?s=46&t=k6rimWsEbo2D2tXolYcM-A In February, I detailed my analysis of promising security software companies here: https://((x.com))/qinbafrank/status/2021506837720498224?s=46&t=k6rimWsEbo2D2tXolYcM-A, and since then, most of them have doubled in value. Special note: Crwd did a stock split in late June (1-for-4). I also revisited this topic in late May via a quoted tweet.