Upwind raises $300M at $3.8B as runtime security extends to AI workloads

The company reported 900% year-over-year revenue growth and a doubled customer base heading into the round. Customers include Roku, Siemens, Nubank, Wix, Peloton and Nextdoor. Reported total funding is above $680 million, with annual recurring revenue expected at $50 million to $60 million. That implies a steep revenue multiple. The product uses Linux eBPF to collect kernel-level telemetry, and it generates AI bills of materials to flag risky components in AI models. The valuation has moved fast. It was about $800 million to $900 million in late 2024, then $1.5 billion in January, then $1.6 billion in a Salesforce investment. Reports tie the demand to a vacuum in the cloud security market after Google acquired Wiz. After the round, Upwind acquired Aegis and set up an AI research unit. The company has not said how it will spend the money. Stated plans are to scale the platform and add customers. For AI founders, security buyers now expect agents and model deployments to be monitored in production, not just scanned at build time. Runtime visibility is becoming a procurement checkbox. Startups selling into enterprises should expect security reviews to cover AI workloads, and independent vendors in this category are well funded.