The list. P0 Security made the 2026–2027 Fortune Cyber 60 in the early-stage group. The company and co-founder Nathan Brahms say it is P0’s fourth year in a row on the list. Fortune’s profile puts it in San Francisco with CEO Shashwat Sehgal, founded in 2022, 30 employees. Fortune reports $20 million in funding. The primary sources match: a $5 million seed in October 2023 led by Lightspeed Venture Partners and SV Angel, then a $15 million Series A led by SYN Ventures with Zscaler and Lightspeed. Valuation and revenue are Undisclosed.
Disclosure that matters. Lightspeed helps run the Cyber 60, and Lightspeed has been a P0 investor since the 2023 seed. Fortune lists Zscaler as an investor too. A big security vendor on the cap table is worth noting, because Zscaler sells to the same buyers.
The name tells you the founding story. P0 means “priority zero,” the most urgent ticket in engineering. 451 Research says the founders were tired of waiting for access to production during a live customer fire. So they built a way to grant that access fast, for a short time, and then take it away.
The problem in plain words. Most companies still have too much “standing privilege”: admin rights that sit there all day whether anyone needs them or not. Now AI agents act with those same rights. P0 makes the point directly: agents usually run on the credentials of the person or service account that started them. So if the human has too much access, so does every agent working for them. P0’s own demo shows a renewal agent doing exactly what it was asked and still pulling far more customer data than the support rep needed.
The one unique thing: one decision that blends four identities. P0’s Authz Control Plane, extended to agents in June, does not treat agent access as a static key. At the moment an agent acts, it checks four things together: the person who started the task, the agent itself, whether that tool is allowed, and what the target resource permits. If it passes, P0 grants just enough access, just in time, for that task. When the task ends, the access is revoked. If it fails, P0 can deny it or send it to a human for approval.
Think of a hotel key card printed for one room and one hour, instead of a master key left in the agent’s pocket. The plumbing is the clever part. 451 Research says P0 is agentless and API-native: no proxies, jump hosts or bastion servers. It checks the user through the identity provider the company already runs, like Okta or Entra ID, then creates the right role inside the target system and lets it expire. 451 also notes an MCP gateway that sits in front of MCP servers and enforces tool-level rules.
Why it is winning now. The Lightspeed Cyber60 Survey, quoted in other honorees’ releases, says agent identity, authorization and access governance is the top category CISOs plan to evaluate next year. P0 does not have to invent a budget line. “Zero standing privilege” projects already exist inside most identity teams. P0 just says an agent is one more identity to put on that program.
The CyberMerge view. This is the boring answer, and boring can win. Keycard, which this series covered, is building agent identity fresh with task-scoped tokens. P0 is extending a privileged access tool that already ships. For a CISO, “add agents to the program we fund” is an easier yes than “buy a new category.”
But look at the money. Neo and Ent each came out of stealth this year with $100 million. P0 has raised $20 million total, per Fortune. Four years on the list on that budget says capital efficiency. It also says P0 has to raise, or partner, before the giants close the gap. Palo Alto Networks now owns CyberArk, the biggest name in privileged access, and every platform will pitch just-in-time access for agents.
Bottom line. P0 is right that agent security is mostly an old access problem moving faster. Watch for named customers, the MCP gateway in real deployments, and the next round. If P0 can prove its runtime checks don’t slow agents down, it is a clean target for a platform that wants agent access control without building it.
