The SaaS Destruction Map showed what dies. But destruction is only half the story. The same force that kills the application layer makes the infrastructure layer structurally more valuable — through four self-reinforcing loops.
Every agent action begins with a data query. A human might check a dashboard once a day. An agent queries the underlying data layer thousands of times per hour. Snowflake’s growth re-accelerated to 29-32% YoY in 2025 precisely because AI workloads are driving consumption — Cortex AI is now responsible for 50% of new customer acquisitions.
The more agents get deployed, the more data gets queried, stored, and governed. Data infrastructure isn’t a cost center that agents compress. It’s a consumption engine that agents accelerate.
Loop 2: The Attack Surface Expansion
Every agent-to-agent connection is a new vulnerability. Every MCP protocol handshake is a new attack vector. Every autonomous action without human review is a new risk surface.
When 100 humans used Salesforce, you needed to secure 100 endpoints. When 10 agents handle the same workload through APIs, you need to secure every API call, every data access, every action chain — and you need to do it in real-time without human supervision.
CrowdStrike’s 51% stock gain in 2025 wasn’t sentiment. It was the market recognizing that security TAM expands with every agent deployed.
Loop 3: The Observability Imperative
Here’s the paradox of autonomous systems: the less human involvement in a workflow, the more critical it becomes to monitor what’s happening.
When a human runs a process, they notice when something looks wrong. When an agent runs it, nobody notices until the damage propagates. Datadog’s entire thesis — monitoring digital ecosystems for issues, vulnerabilities, and performance — becomes exponentially more important when those ecosystems operate without human observation.
As Davidson analysts captured it precisely: without data lakes and observability, companies cannot achieve their AI goals.
Loop 4: The Identity and Compliance Cascade
This is the expansion mechanic that creates entirely new value rather than redirecting existing value.
Every AI agent needs an identity. Every identity needs permissions. Every permission needs an audit trail. Every audit trail needs compliance verification. None of this existed before agents.
Okta’s repositioning as the “Identity Security Fabric” for agentic AI isn’t marketing. It’s a structural bet that the identity layer for agents will be as sticky and essential as Active Directory was in the PC era. Agent-to-agent authentication is an entirely new product category.
The Structural Logic
These four loops share a property the destruction categories lack: positive feedback loops with agent adoption.
When agents kill a Tier 1 content tool, the destruction is one-time. Jasper loses a customer. That’s it.
When agents drive expansion, the growth compounds. Deploy one agent → it queries Snowflake (data revenue up) → it authenticates through Okta (identity revenue up) → it gets monitored by Datadog (observability revenue up) → it gets secured by CrowdStrike (security revenue up). Deploy ten agents and multiply every line.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
