LE PERSPECTIVE / THE STORY

Upwind has raised $400 million in a Series C to expand its cloud and AI security platform, another sign that investors see protecting live AI infrastructure as a major cybersecurity business.

Bessemer Venture Partners led the round, with existing backers including Greylock, Cyberstarts, Salesforce Ventures and TCV participating. Canapi Ventures joined as a new investor. Upwind says the financing lifts its total funding to more than $800 million.

It is a fast follow-on: the company announced a $250 million Series B in January 2026, less than nine months ago. The new capital is earmarked for product and engineering expansion, with a sharper focus on how enterprises secure AI systems.

The Real Bet Is Runtime
Scans show what might be vulnerable. Runtime shows what’s actually exposed. Upwind is betting the real value lies in knowing which workloads are live and which risks an attacker can reach right now. As AI agents change infrastructure faster than teams can review it, telling real threats from theoretical ones becomes the job.

Why live cloud activity matters

Traditional cloud security tools can flag configurations and vulnerabilities without always showing whether an attacker can actually reach them. Upwind's approach starts with what workloads are doing in production, then combines that activity with context about applications, identities and infrastructure.

The goal is straightforward: help security teams spend less time sorting through alerts and more time fixing exposures that matter. Bessemer highlighted that runtime-first approach when it backed Upwind's Series B earlier this year.

Upwind is building security agents

Upwind is extending beyond cloud infrastructure into endpoints and AI workflows. CEO Amiram Shachar says the company is building AI sensors for endpoints, scanners for agents and their skills, and a broader inventory graph connecting those systems.

It has also detailed three AI agents. Blue investigates signals across logs to build an attack narrative. Red tests whether reported exposures are exploitable. Green identifies ways to remediate security findings across resources. These are company-described capabilities, not independently measured results.

What the money needs to prove

The company has also launched an AI-focused cybersecurity lab covering AI security, agentic security, frontier models and advanced AI systems. Together, those investments point to a larger ambition: one platform spanning code, cloud infrastructure, endpoints and the AI agents operating across them.

There is a practical test ahead. Security teams will need proof that automated investigation and remediation cut real risk without causing disruption. Upwind now has substantial capital to develop that approach; its next challenge is demonstrating that it works at enterprise scale.

Get stories like this in your inbox every week. Subscribe to The Week in Perspective.