July 19, 2024T1
The CrowdStrike Falcon Outage — The Largest IT Outage in History
On 19 July, a faulty update file (Channel File 291) distributed by the cybersecurity firm CrowdStrike's Falcon Sensor sent an estimated 8.5 million Windows machines worldwide into bootloop / blue-screen failure. Airline flights (Delta cancelled about 7,000 flights, with reported losses of US$500 million), hospitals, bank ATMs, emergency response systems, and broadcasters were simultaneously taken down. Total estimated economic loss exceeded US$10 billion, the largest of any single IT outage. Recovery required manual boot into Safe Mode; many organisations needed days to weeks. The incident exposed, at global scale, the supply-chain risk of SaaS security products and the weakness of software validation processes.

Metadata
- Date
- July 19, 2024
- Decade
- 2020s
- Tier
- T1
- Timelines
- A General History of Information Technology · A History of Cloud Computing · A History of Cybersecurity
- Sources
- 06
- Connections
- 01
The CrowdStrike Falcon Outage — The Largest IT Outage in History
On 19 July 2024, a faulty update file (Channel File 291) distributed by the cybersecurity firm CrowdStrike for its Falcon Sensor sent an estimated 8.5 million Windows machines worldwide into the blue screen of death (BSOD) at roughly the same instant. Total estimated economic loss exceeded US$10 billion, making it the largest single IT outage on record.
What Happened
At 04:09 UTC, CrowdStrike pushed Channel File 291—a content-configuration file for the Falcon Sensor. These files update threat-detection logic and were, by operational design, pushed directly to production without passing through customer validation stages.
The file contained data that triggered an out-of-bounds memory read. The Falcon Sensor's core is a driver (csagent.sys) running in Windows kernel space; an unhandled exception there halts the operating system immediately. Affected machines BSOD'd in waves, and on reboot loaded the same bad file and crashed again—locked into a bootloop.
Recovery was conceptually simple but manual: boot into Safe Mode, delete the offending file, reboot normally. On BitLocker-encrypted drives the recovery key was required, and large enterprises had to perform the procedure across thousands or tens of thousands of machines in parallel.
What Fell Together
By the hour after 06:00 UTC, mission-critical systems began going dark around the world simultaneously.
- Aviation: Delta, United, and American halted operations. Delta alone cancelled around 7,000 flights.
- Healthcare: Hospitals in the US and UK lost electronic records, appointments, and diagnostic systems; some surgeries were postponed.
- Finance: ATMs, branch terminals, and parts of securities-trading workstations went down.
- Emergency services: 911 systems in several US states became intermittently unreachable.
- Retail: POS, inventory management, online payments.
- Broadcast: Sky News, ABC (Australia), and other broadcasters went off-air.
Full recovery took days to weeks depending on the organisation, particularly painful for remote-work fleets and store terminals requiring physical access.
The Delta Lawsuit and the "Kernel Privilege" Debate
The fiercest legal fight was begun by Delta Air Lines. Delta claimed losses of over US$500 million on its own and sued both CrowdStrike and Microsoft in October 2024. The complaint's technical core was an accusation: CrowdStrike had "intentionally created and exploited an unauthorised door within the Microsoft OS" to push updates straight into the Windows kernel—in violation, Delta alleged, of Georgia's computer-crime statute.
CrowdStrike countersued, arguing that Delta was using litigation "as a smokescreen to hide its own IT and incident-response inadequacies". Several shareholder derivative suits and federal securities class actions were filed in parallel.
What the lawsuits exposed was the political dimension of "kernel privilege" in endpoint security. EDR (Endpoint Detection and Response) products operate in the OS kernel in order to detect threats, and a bug there can take down the entire operating system in an instant. The foundational question—"are we adding vulnerabilities in the name of security?"—finally arrived on the public agenda.
Microsoft's Response — The Windows Endpoint Security Summit
On 10 September 2024, Microsoft convened the Windows Endpoint Security Ecosystem Summit at its Washington headquarters. CrowdStrike, the major EDR vendors, and government representatives were invited, with a single agenda item: how to change Windows's security architecture so that "this never happens again".
Microsoft's vice-president of enterprise and OS security, David Weston, signalled a clear direction—Windows would gain new platform capabilities that operate outside the kernel. EDR products would no longer need to live in kernel space. Not an outright kernel ban, but a deliberate architectural offer that frees vendors from the requirement.
Why had Microsoft granted EDR vendors kernel access in the first place? The context goes back to a 2009 EU antitrust ruling. At the time the EU was concerned that Microsoft's own security product (Windows Defender) shut out third-party competitors, and Microsoft was required to grant third parties the same level of kernel access it gave itself. In 2024 the headline "The EU gave CrowdStrike the keys to the Windows kernel" (The Register) captured the bitter shape of the situation—a textbook case of an unintended regulatory consequence.
CrowdStrike's Stock and Earnings
On 19 July CrowdStrike shares fell -11%, and accumulated losses of roughly -45% over the 18 trading days that followed. The company introduced a Customer Commitment Package (CCP) discount programme to retain customers and absorb damages, which then trimmed roughly US$11 million off quarterly revenue for several consecutive quarters from H2 2024 into 2025. The CCP programme was ended in 2025, but the financial drag continued into 2026.
Yet CrowdStrike did not fall. The set of credible EDR alternatives is small, replacement effort enormous, and most customers kept their contracts. The share price recovered gradually over the following 18 months. What the industry demonstrated was a different risk altogether—the "switching difficulty" of a concentrated security stack.
What It Demonstrated
The CrowdStrike outage left three lessons.
1. SaaS security as software supply chain. Auto-updates that bypass validation can cause incidents indistinguishable from supply-chain attacks. The debate around mandatory staged (canary) rollouts moved sharply forward.
2. Re-evaluation of kernel privilege. The premise that "a security product must live at the deepest part of the OS to do its job" finally came under genuine scrutiny. Windows, macOS, and Linux are all now exploring technical paths to push EDR back into user space.
3. Regulatory intent versus consequence. What the EU required in 2009 in the name of fair competition made a global outage possible in 2024. It is now a canonical example of the gap between regulatory design and the technical reality fifteen years later.
8.5 million BSODs were not, in the end, a single company's QA failure or a single OS vulnerability. They were the architectural philosophy of the whole industry — concentrating privilege in security products — implemented one morning by one bad data file. That is the structural shape of the event.