Oasis Security Lands $120M to Secure AI Agents

There are 82 non-human identities for every human user in the average enterprise. That's not a typo. It's a security crisis hiding in plain sight — and Oasis Security just raised $120 million to fix it.

The Israeli cybersecurity startup announced its Series B round led by Craft Ventures, with heavyweights Sequoia Capital, Accel, and Cyberstarts doubling down. Total funding now sits at $195 million. Annual recurring revenue has grown 5x year-over-year. A majority of Fortune 500 companies are already customers.

Two years ago, "non-human identity security" wasn't a category. Now it's arguably the most important gap in enterprise infrastructure. The math is simple and terrifying: every AI agent, every automated workflow, every bot spinning up in your cloud environment is an identity. It needs credentials. It accumulates permissions. And almost nobody is managing them properly.

The Identity Problem Nobody Planned For

Here's what happened. Enterprises sprinted to deploy AI agents across IT operations, customer support, finance, and engineering. They gave these agents API keys, service accounts, and broad permissions to get things done fast. The productivity gains were real. The security posture was a disaster.

The numbers tell the story. According to research cited across the industry, 90% of AI agents are over-permissioned, often holding 10x more privileges than they actually need. AI agents move 16x more data than human users. Only 20% of organizations have formal processes for revoking API keys. And 49% of organizations don't even track how many disconnected apps they have running.

This isn't theoretical risk. New attack vectors like "tool misuse" — where an attacker leverages an AI agent's excessive permissions to pivot into sensitive data — are already being documented. An agent designed to list orders gets tricked into using an invoicing tool to exfiltrate vendor data. The agent had the permissions. Nobody thought to scope them.

Every AI agent is an identity. Every identity is an attack surface. And the ratio is 82 to 1.

What Oasis Actually Built

Founded in 2022 by Danny Brickman and Amit Zimerman — both veterans of Israeli Defense Forces cybersecurity — Oasis emerged from stealth in January 2024 with a $40 million Series A. They followed that with a $35 million extension just months later. The speed of fundraising reflects the speed at which the problem exploded.

Their platform, Oasis Agentic Access Management (AAM), takes a fundamentally different approach from traditional identity and access management. Legacy IAM platforms were built for humans clicking through login screens. They weren't designed for thousands of machine identities spinning up and down dynamically, requesting access in real time, operating across IaaS, SaaS, PaaS, and on-premise environments simultaneously.

Oasis's approach centers on three principles:

• Just-in-time access: Agents get permissions when they need them, scoped to what they need, and those permissions expire automatically.
• Intent-based control: The system evaluates what an agent is trying to do, not just what credentials it holds, enabling smarter policy enforcement.
• Infrastructure-agnostic deployment: Works across AWS, Azure, GitHub, ChatGPT, Salesforce — wherever agents operate.

This is the right architecture. The old model of provisioning static credentials and reviewing them quarterly is laughable when you're dealing with agents that can spin up, execute tasks, and accumulate entitlements in seconds.

Why This Round Matters Beyond the Dollar Amount

Craft Ventures leading this round is notable. David Sacks' firm doesn't chase hype — they chase categories with clear enterprise pain and defensible market positions. Their bet here signals conviction that non-human identity management isn't a feature inside existing security platforms. It's a standalone category.

And they're probably right. CyberArk, Saviynt, and others are bolting on NHI capabilities, but the problem is too big and too different from human identity management to be an afterthought module. The semantics of machine identity — ephemeral credentials, programmatic access patterns, multi-system orchestration chains — require purpose-built infrastructure.

Gartner projects that by 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI. By 2026, 40% of enterprise applications will feature embedded task-specific agents, up from less than 5% in early 2025. The World Economic Forum has flagged NHIs as a "new frontier of cybersecurity risk." The economic impact of AI agents is estimated between $2.6 trillion and $4.4 trillion.

Where that much value flows, attackers follow.

The Competitive Landscape Is Getting Crowded

Oasis isn't alone. Entro Security is building a platform for unifying security across AI agents, NHIs, and secrets. Token Security focuses on centralized visibility and continuous least privilege enforcement. The Cloud Security Alliance has published frameworks for enterprise NHI risk mitigation. An entire ecosystem is forming.

But Oasis has first-mover advantage with Fortune 500 scale, a 5x ARR growth trajectory, and nearly $200 million in the bank. In enterprise security, incumbency with large customers is a moat. These aren't SMBs you can poach with a slick demo. Fortune 500 security teams that have deployed Oasis across their agent infrastructure aren't ripping it out anytime soon.

The Bottom Line

We've entered an era where the majority of "users" in enterprise systems aren't human. They're bots, agents, automated workflows, and API-driven services. They move faster, touch more data, and accumulate more permissions than any human employee. And until very recently, almost no one was treating them as the security risk they are.

Oasis Security's $120 million raise isn't just a big number. It's a market acknowledgment that identity security needs to be completely rethought for a world where machines outnumber people 82 to 1. The companies that figure this out will own a foundational layer of AI-era infrastructure. The companies that don't will be the ones in the breach headlines.

Place your bets accordingly.

Want more analysis on the companies building AI's security infrastructure? Follow Ultrathink for daily coverage of the funding rounds, product launches, and strategic moves shaping the future of enterprise AI.