Upwind Security raises $400 million led by Bessemer

Upwind Security's rapid fundraising pace reflects investor appetite for security tools tailored to AI. The company argues that static, agentless security cannot keep up with AI agents and MCP servers that spin up on temporary compute and communicate before disappearing. Upwind's platform scans cloud infrastructure to inventory AI resources, including models and external AI services used outside approved controls. It then visualizes sensitive data and permissions AI systems can access, and monitors for prompt injection, data exfiltration, and agent misuse.

To help customers investigate and fix issues, Upwind uses its own AI agents: Blue Agent consolidates attack flows across logs and data, Red Agent safely tests whether exposures are actually exploitable, and Green Agent finds high-leverage fixes to resolve thousands of security findings at once. The company says hundreds of enterprises worldwide use its platform, with customer CISOs reporting that it changed where their teams spend time.