CrowdStrike Bets Nearly $500 Million on Securing AI Agents

CrowdStrike has moved aggressively into the race to secure artificial intelligence, announcing on 3 November that it has agreed to buy two AI security start-ups, Fiddler AI and Dreadnode, in deals reported to be worth about $495 million combined.

The twin acquisitions were unveiled at Fal.Con Europe, the company’s conference in Barcelona, and they point to where the cybersecurity industry believes its next big market lies: not in protecting laptops and servers alone, but in watching over AI agents — software systems that can take actions inside a company, from answering customers to moving data between applications.

As businesses roll out agents more quickly than their security policies can keep up, CrowdStrike is betting that customers will pay for tools that record what agents do, test them before they go live, and stop them when they go wrong.

Two deals, two halves of the same problem

The larger of the two deals is for Fiddler AI, a Palo Alto company founded in 2018 that builds observability software for AI models and agents. In plain terms, it keeps a record of what an agent decided, which tools it called and whether it stayed within the rules set for it. CrowdStrike plans to sell that capability as Falcon Observe, with general availability scheduled for the fourth quarter of its fiscal 2027.

The Wall Street Journal reported that CrowdStrike is paying about $400 million for Fiddler AI, according to reporting cited by TechCrunch. Fiddler AI has raised about $99 million to date, according to PitchBook figures cited in the same coverage. CrowdStrike itself did not disclose a price.

The second deal is for Dreadnode, founded in 2023, which automates security testing for AI systems. Rather than waiting for a real attacker to find a weakness, Dreadnode attacks a company’s AI models, applications and agents before deployment, using more than 100 known attack techniques, to expose flaws while they can still be fixed. Its technology is set to become part of Falcon Guardian, CrowdStrike’s controls for AI systems already in use, and its team will join CrowdStrike’s Cyber Superintelligence Lab once the deal closes.

The reported price for Dreadnode is about $95 million. The company has raised $14 million so far, according to PitchBook. As with Fiddler AI, CrowdStrike did not confirm the figure, and neither transaction has closed. Both are expected to complete in CrowdStrike’s fiscal fourth quarter of 2027, which ends on 31 January 2027, subject to customary closing conditions. Both deals are to be paid predominantly in cash, with a portion in stock subject to vesting.

A kill switch for rogue agents

Alongside the acquisitions, CrowdStrike launched an AI Kill Switch at the same Barcelona event. The feature is designed to let security teams cut off an AI agent that has gone off course, severing its connections to AI models through CrowdStrike’s gateway. The gateway version is available now, with an endpoint-level shutdown function to follow.

The combination is deliberate. The kill switch gives customers an emergency brake today, while the two acquisitions are meant to supply the longer-term machinery: testing before launch, monitoring while agents run, and evidence afterwards about what happened and why.

That sequence matters because AI agents fail in ways that traditional software does not. An agent can misunderstand an instruction, call the wrong tool, leak sensitive data into a prompt or be manipulated by hostile content hidden in the data it reads. Security teams often discover such problems only after the damage is done, because they have no reliable record of the agent’s decisions.

Why the agent security market is heating up

CrowdStrike is far from alone in chasing this market. As companies move from chatbots that merely answer questions to agents that act on their behalf, the attack surface grows with every new permission an agent is given. Industry forecasts cited in coverage of the deals put the market for securing AI at nearly $5 billion in 2027.

The acquisitions also fit a broader pattern of consolidation in cybersecurity, where large platform vendors are buying specialist start-ups rather than building every capability from scratch. For CrowdStrike, folding both companies into Falcon, rather than selling them as standalone products, is intended to make agent security a standard part of the platform its customers already use.

There are, however, limits to what the announcement proves. The prices are reported, not confirmed by CrowdStrike. The deals have not closed, and the new Falcon Observe and automated testing products are scheduled, not yet shipping. Customers will judge the strategy on whether the integrations arrive on time and work as promised — and on whether the testing and monitoring genuinely catch failures before they reach production.

For now, the signal is clear enough: the company that built its name stopping breaches on endpoints now sees AI agents as the next frontier to defend, and it is willing to spend close to half a billion dollars to get there first.

Related coverage: Marvell-Lumentum AI optics deal · OpenAI compute campuses · Apple retaking most valuable company. Sources: TechCrunch · PYMNTS.