Skip to content
NewA calmer, faster ZeroHour

AI security took more than one in three cyber seed deals

Original title: AI security took more than one in three cybersecurity seed funding deals

Help Net Securityby Mirko Zorz ()2 min read
AI summary · x-ai/grok-4.7Summaries by Vani AI

U.S. cyber seed volume slipped in Q3 2026 even as broader venture capital had one of its strongest quarters since 2021.

DataTribe, using PitchBook data, says U.S. cybersecurity seed deals fell slightly in the third quarter of 2026, the second decline in a row. AI security accounted for more than one in three of those deals, mostly companies securing AI agents, with AI SecOps and identity next. Across all sectors, $187.25 billion went into seed through Series E rounds through three quarters, in roughly 4,700 deals. DataTribe also says the gap to Series A has widened and a strong Series A is now closer to $3 million in annual recurring revenue.

Unconfirmed: DataTribe says the two-quarter cyber seed decline may be ordinary variation.

Key facts
  • Twelve companies each raised rounds of more than $1 billion.
  • The median U.S. Series A rose from $13.7 million in 2025 to $17.4 million.
  • Series A pre-money valuations were nearly 4.6 times the seed median in Q3.
  • Pinpoint ranked identity the most active cyber segment, with 16 rounds and seven acquisitions.
  • Cyera’s Oasis Security deal was valued at $1 billion, and Okta agreed to buy Permiso.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

From the original article

2 min read at helpnetsecurity.com

Seed deals for cybersecurity startups dipped slightly in the third quarter of 2026, the second decline in a row, while the wider U. S. venture market had one of its best quarters since 2021.

Help Net Security

By Mirko Zorz · helpnetsecurity.com

Read the full article at Help Net Security

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.