A rare wave of outages is sweeping the United States! AI is tearing open systemic cracks in IT infrastructure, and Citi has identified six beneficiary sectors.

date
20:41 19/11/2026
avatar
GMT Eight
Over the past week, the United States has experienced a rare wave of concentrated, large-scale technical failures and system outages. Citi said that although most incidents have been preliminarily ruled out as being directly caused by malicious cyberattacks, they reflect the systemic fragility of the entire IT infrastructure under the wave of artificial intelligence (AI).
Over the past week, the United States has experienced a rare, concentrated wave of large-scale technical failures and system outages. Microsoft Corporation (MSFT.US), the Federal Aviation Administration (FAA), Apple Inc. (AAPL.US), United Airlines (UAL.US), the U.S. National Security Agency communications system, and multiple Bank of America Corp entities successively suffered service disruptions. There were also cases of state-backed actors launching cyberattacks against U.S. critical infrastructure. In response, Citi noted in its latest research report that although most incidents have been preliminarily ruled out as being directly caused by malicious cyberattacks, they reflect the systemic fragility of the entire IT infrastructure under the AI wave. Citi outlined six core observation areas and provided corresponding beneficiary sectors and listed company names. Based on the disclosed root causes of the incidents, most failures point to underlying issues at the systems engineering level. Microsoft Corporation attributed service disruptions to legacy software defects and abnormal server loads under high network traffic; JPMorgan Chase (JPM.US) failures stemmed from software bugs in local hardware environments. Citi analysts said that although direct evidence of malicious attacks is insufficient, non-negligible cybersecurity-related complexities remain behind the incidents, and the risks cannot be simply categorized as ordinary technical failures. Based on this series of cascading outage events, Citi put forward six core judgments. First, the "Vulnpocalypse" has arrived, the patch window has nearly disappeared, and enterprise IT operations pressure has surged. AI technology has dramatically increased the number and speed of software vulnerability (CVE) discovery, completely compressing the buffer cycles for testing, updating, and patching that enterprises had previously mastered. The manpower and time resources enterprises devote to vulnerability remediation can no longer keep pace with the explosive growth in vulnerability disclosures. Citi believes the optimal path to address this structural contradiction is to move toward automated penetration testingthat is, routine infrastructure "emergency drills"while automating patch testing and patch management. Companies with business advantages in this direction include CrowdStrike (CRWD.US, SafeMind, exposure surface management products, incident response business), Palo Alto Networks (PANW.US, Unit42 threat intelligence team), established infrastructure vendors like Microsoft Corporation, Cloudflare (NET.US, PHP2Shell protection free feature), and some traditional virtual machine security vendors. Second, OT/IoT security is receiving greater attention, and the long-standing underinvestment in industrial control systems has exposed shortcomings. OT operational technology and IoT environments have long suffered from historical underinvestment and unique architectural constraints from SCADA industrial control systems. Fortinet, Inc. (FTNT.US) has the highest exposure in this sector, with related business accounting for approximately 15-20% of its total revenue; ServiceNow (NOW.US, with its Armis) and Zscaler (ZS.US) are also relevant; there are also vertical private companies like Claroty and Dragos; traditional industrial giants such as Siemens and Honeywell International Inc. (HON.US) are also expected to play important acquisition and integration roles in this area. Third, what cannot be observed cannot be fixed, and the strategic position of observability tools is further elevated. Only by achieving fine-grained visibility into infrastructure and IT system architecture and clearly mapping the dependencies among technical components can failures be diagnosed and resolved. The large-scale deployment of AI is continuously changing infrastructure architecture and posing new challenges to overall system stability, which will drive up enterprise investment in observability products. Datadog (DDOG.US) and Dynatrace (DT.US) are in the most favorable position in this field. Fourth, the value of on-premises deployment is being re-examined, and workload "repatriation" may become a new trend. A series of outage events may prompt enterprises to reassess the business logic of migrating computing power, infrastructure, and AI inference workloads back to on-premises deployment. The advantage of on-premises environments is that, in the event of a failure, the risk impact scope can be confined within hardened, controllable boundaries. Citi observed that this trend has gradually emerged at the industry conference KubeCon, and multiple vendors' product strategies are also tilting in this direction. CrowdStrike, Zscaler, Rubrik (RBRK.US), and IBM (IBM.US) are all increasing R&D investment in sovereign data centers to respond to privacy, compliance, and data security demands in the AI era. Fifth, if outages become the norm, rapid system recovery capability will be the core competitiveness. The report puts forward an important hypothesis: future technical system disruptions may shift from occasional exceptions to a normalized phenomenon, in which case the speed of business recovery after a failure will become a key metric. This logic benefits vendors in the backup, disaster recovery, and business resilience sectors. Rubrik is a representative company, focused on helping enterprises quickly restore normal business operations. Citi further noted that the compatibility between business operation requirements in the AI era and the traditional technology stacks and disaster recovery systems built before AI became widespread is deteriorating, which will accelerate the modernization of technology stacks in this field, while the market is still occupied by a large number of traditional legacy vendors. The large-scale deployment of agent systems and automated workflows will further amplify outage risks, forcing enterprises to accelerate the upgrading of backup and recovery infrastructure. Sixth, if outages are increasingly driven by malicious attacks, cyberattacks will become more sophisticated and spread faster. If future failures gradually overlap with malicious cyberattacks, the valuation center of the entire cybersecurity sector is expected to rise, and enterprise cybersecurity budgets will face upward pressure. However, incremental budgets will not be distributed evenly. Platform vendors with multiple product lines and product coverage across multiple stages of the cyberattack kill chain (network, endpoint, cloud infrastructure control points) are best positioned to capture new budgets. On comprehensive assessment, Palo Alto Networks and CrowdStrike best fit this profile.