A sitting chief technology officer at one of the world's largest cybersecurity companies is walking away from a high-profile role to start a venture fund, betting that the entire security stack needs to be rebuilt from scratch as artificial intelligence agents become a new attack surface. Elia Zaitsev, who spent 13 years at CrowdStrike and served as the company's global CTO for the past three years, is launching Cognition, a new venture firm targeting $170 million to back early-stage cybersecurity startups designed specifically for what the industry calls the "agentic AI era".

This move is significant not because of the fund size, which is modest by 2026 venture standards, but because of what it signals about where technical talent believes the next wave of innovation will happen. When a company's own CTO votes with their career that the future of security lies outside the incumbent, it sends a powerful message to founders, investors, and the broader industry.

Why Is a CTO Leaving to Start a Venture Fund Such a Big Deal?

Zaitsev is not simply joining a venture firm as an advisor or taking a part-time role. He is stepping down from a high-visibility, highly compensated position at a public company with a market capitalization in the tens of billions to run a concentrated, hands-on seed and Series A fund. That represents a genuine opportunity cost and signals real conviction that the next generation of category-defining cybersecurity companies will not be built inside the incumbents.

The timing matters too. CrowdStrike's stock price declined on news of Zaitsev's departure, which public markets interpreted as a signal that technical leadership believes the frontier of value creation is moving outside the company. This is not routine executive churn; it is a statement about where the industry is heading.

Zaitsev is joining forces with Gur Talpaz and Tayler Sipperly, two former CrowdStrike corporate-development executives who already have a proven venture track record. Together, they founded Brightmind, an earlier venture fund that backed SGNL, an identity-security startup that CrowdStrike itself acquired for $740 million in January 2026. That exit, from a reported $105 million entry valuation, represents roughly a 7x return on a Series A investment and is being described as one of the largest Series A exits in cybersecurity history.

What Makes Cognition's Investment Strategy Different?

Cognition is not a spray-and-pray fund that makes dozens of small bets across the market. Instead, it is designed as a concentrated, thesis-driven vehicle with a specific playbook. The fund targets approximately $6 million average checks at the seed stage and $15 million average checks at Series A, which means it will make roughly three to four concentrated investments per year.

This approach allows the fund to take large ownership stakes and board seats in a handful of companies rather than build a diversified portfolio across many startups. That is the same playbook that made Brightmind's SGNL bet work, and according to the source material, it only pays off if Cognition's sourcing and diligence are as sharp as the team's resume suggests.

The fund's focus is specifically on cybersecurity platforms built for the agentic AI era. This thesis rests on the idea that AI agents and autonomous systems represent a platform-reset moment requiring genuinely new security architectures, not just incremental tooling or thin wrappers around large language models (LLMs, which are AI systems trained on vast amounts of text data) bolted onto existing security products.

What Value Does Zaitsev Bring Beyond Capital?

For founders building in agentic AI security, Cognition offers something scarcer than capital alone: a general partner who has sat on the buyer's side of exactly this market and knows what CrowdStrike and its peers will actually pay for versus what they will build in-house. Zaitsev spent 13 years at CrowdStrike with a background spanning data science, artificial intelligence, machine learning, and malware research. He was CTO during the period that included CrowdStrike's acquisition of SGNL, giving him first-hand visibility into how the company evaluates and prices identity-security and AI-adjacent acquisition targets.

This insider knowledge is a different kind of value-add than the standard "we will make introductions" pitch most seed funds offer. Founders in this space should view this as a meaningful advantage during their next fundraising round.

How Should Founders Evaluate Cognition as a Lead Investor?

• Ownership and Control: Cognition's concentrated check sizes mean the fund will want real ownership stakes and likely board representation, making this a fit for founders who want a hands-on, high-conviction lead investor rather than a passive check.
• Thesis Alignment: The fund is specifically targeting cybersecurity companies with a genuine thesis about how AI agents change the attack surface, not generic "AI-powered" features bolted onto existing security products.
• Exit Strategy Insight: The team's value-add is specific and unusual: direct, recent operating knowledge of what a top-three security platform vendor will pay for and what it cannot build fast enough internally.

If founders are not looking for a hands-on, high-conviction lead investor relationship at the seed or Series A stage, Cognition may not be the right fit.

What Are the Risks in This Concentrated Approach?

The primary risk with any concentrated, thesis-driven fund led by first-time institutional general partners is repeatability. One great exit proves pattern recognition once; a $170 million fund making three to four bets a year needs that pattern to hold across a genuinely new and still-forming category where the winning architecture for "agentic AI security" is not obvious yet, even to people who have spent a career on the defending side.

Additionally, there is an unavoidable conflict-of-interest angle worth noting. CrowdStrike was the acquirer of Brightmind's SGNL bet, and Zaitsev was CrowdStrike's CTO at the time that deal closed. Cognition's new fund is being seeded by three people with deep, recent knowledge of exactly what CrowdStrike will pay to acquire and what gaps its own roadmap cannot fill internally. That represents a real information edge for sourcing and exit planning, and limited partners backing this fund are, whether they say so explicitly or not, partly underwriting a bet on Zaitsev's professional network and his read on his former employer's mergers and acquisitions appetite.

The cybersecurity venture landscape has experienced an interesting dynamic in recent years: enormous public-market enthusiasm for the category alongside a private market that has become pickier about which "AI security" startups actually have differentiated technology versus a thin wrapper on an LLM API. Cognition's framing, that agentic AI represents a platform-reset moment requiring genuinely new security architectures, is the correct read of where the industry is heading, and it is a thesis that benefits enormously from having someone who ran security engineering at global scale making the investment calls.