AWS Launches Frontier Agents for DevOps and Security — Autonomous AI That Runs 24/7, Pen Tests for $50/Hour, and Already Works Across Azure

AWS DevOps Agent and Security Agent reached general availability on March 31, 2026, marking the first hyperscaler launch of autonomous AI agents for cloud operations and penetration testing. DevOps Agent runs incident investigation across AWS, Azure, and on-prem environments — preview customers report 75% lower MTTR and 94% root cause accuracy. Security Agent performs autonomous penetration testing at $50/task-hour, compressing weeks-long pen test cycles into hours. Together they represent a new class AWS calls 'frontier agents' — systems that run for hours or days without human oversight, handle multiple tasks simultaneously, and operate continuously. The competitive landscape is heating up: Azure's SRE Agent went GA on March 10 with 35,000+ incidents mitigated internally, while Google Cloud sells the platform (Agent Development Kit) for customers to build their own. This analysis covers the agent capabilities, pricing, enterprise adoption results, governance risks, and what autonomous AI operations mean for DevOps and security teams.