# Two Outages Test U.S. Aviation and Cloud Resilience

Published: 2026-11-14

## Verdict

Friday's nationwide aviation stop and Microsoft's global workplace-software outage were not one event and there is no evidence that they shared a cause. They were, however, a concentrated demonstration of the same economic weakness: essential services depend on communications systems whose failure can spread faster than customers can switch to alternatives. The Federal Aviation Administration chose to stop departures rather than risk aircraft operating without reliable controller communications. Microsoft said a technical issue following a network change disrupted Outlook, Teams and Copilot across multiple regions. The immediate costs landed outside the organisations that operate the systems: passengers lost time, airlines lost schedule capacity, NHS teams reverted to paper and alternative systems, and small firms faced lost productivity that insurance may not cover. The policy question is no longer simply whether the two systems can be repaired. It is who pays for redundancy, who proves it works, and whether public procurement and regulation can make resilience a contractual requirement rather than a promise after an outage.

## The FAA chose a stop rather than an unsafe departure

At about 3:53–3:54 p.m. Eastern on Friday, 13 November, the FAA ordered a nationwide ground stop for departing flights. Its notice used the phrase “extraordinary circumstances and for reasons of safety.” Transportation Secretary [Sean Duffy](/people/sean-duffy) said a communications problem was affecting all 22 Air Route Traffic Control Centers, the facilities that coordinate aircraft between airports. The immediate consequence was deliberately blunt: aircraft already airborne could continue to land, but departures were held on the ground. The FAA said all flights already in the air had been accounted for and that inbound international flights would be handled case by case.

That distinction matters. A nationwide stop is not evidence that every aircraft or airport had lost every system. It means the agency could not be sufficiently confident that the communications chain needed for safe departures was available across the national network. Air traffic control depends on more than radar screens. Controllers must be able to communicate reliably with pilots, pass clearances, coordinate hand-offs between centres and maintain a common operational picture. If a communications link becomes unreliable, the safety margin is not to keep launching aircraft and hope the network recovers. It is to reduce traffic until controllers can establish safe operating conditions.

Homeland Security Secretary [Markwayne Mullin](/people/markwayne-mullin) said there was “no indication as of yet” of a malicious attack. That was a statement about the evidence available at the time, not a completed forensic finding. It did not prove that a cyber incident was impossible, and it did not identify the failed equipment or supplier. The responsible conclusion is narrower: officials described a telecommunications outage, technicians were investigating, and no public evidence had established a deliberate attack. The President had been briefed, but a briefing is not a technical explanation.

The order's status also needs careful wording. The [Associated Press](/organizations/associated-press) reported that the ground-stop order ran through at least 11 p.m. Eastern on Friday. The FAA advisory feed carried an effective window extending into early Saturday UTC. Neither statement, by itself, establishes the exact minute when every airport returned to normal schedules or when all delayed flights cleared. This analysis therefore does not claim that the nationwide stop remained active after its published window, nor does it claim that the system was fully normal at a particular later time. The operational fact is that the FAA imposed the stop, and the public record described a scheduled period while the technical investigation continued.

The event was exceptionally rare. The AP described it as the first nationwide ground stop since the January 2023 failure of the pilot-notification system, with the previous comparable nationwide halt following the September 11, 2001 attacks. The comparison is about the scale of the departure restriction, not about the causes or severity of the incidents. The 2023 NOTAM failure was a different system problem. The September 2001 shutdown was a response to a terrorist attack. Friday's event should not be treated as equivalent to either.

## The modernisation budget is now an accountability test

The communications failure landed while Washington was already arguing over the cost and timetable of replacing ageing aviation infrastructure. Congress approved $12.5 billion for air-traffic-control modernisation last year. On Friday, the FAA and [White House](/organizations/white-house) sought another $30.1 billion for a wider package of aviation improvements. Those numbers are not a bill for this one outage. The additional request covers a broader set of problems, and the Government Accountability Office has separately described the next phase of automation modernisation as needing roughly $10.2 billion without a firm start or completion date.

The distinction should prevent two opposite mistakes. The first is to argue that the new outage proves the entire modernisation programme has failed. A system replacement is a multi-year undertaking, and the incident's precise technical cause has not been publicly established. The second is to treat the funding request as self-justifying. A large appropriation does not demonstrate that the money is reaching the communications links, backup paths, software and operational procedures that failed or were unavailable on Friday.

The FAA's next public explanation should identify the affected network layer, the number of centres that lost reliable communications, the time needed to restore each function, the role of contractors and the availability of fallback links. Congress should require a costed schedule that connects each tranche of funding to a measurable reliability outcome. That means reporting not only how many systems have been purchased or installed, but how many centres can continue operating safely when a primary communications path fails. The GAO's warning about cost and schedule planning makes that a practical oversight question, not an argument against modernisation itself.

The incident also gives the FAA's critics a reason to ask whether the programme is being managed as one integrated national system or as a collection of projects. Air traffic control crosses state lines and airline networks. A local outage can become national when the affected component is a shared dependency. Resilience therefore depends on the architecture between facilities, not simply on replacing equipment inside each one.

## Microsoft's outage showed the same concentration risk in business software

The Microsoft failure began at 16:17 GMT on Thursday, according to the cyber-risk estimate published by The Insurer. By Friday it had entered a second day for some users. Outlook, Teams and the Copilot assistant were among the affected Microsoft 365 services. Downdetector received reports from 74 countries and territories within an hour of the first problems. That is a measure of where users reported trouble, not a verified count of countries in which every Microsoft service failed or a count of people affected. Microsoft has previously said it has more than 450 million paid commercial subscriptions to Microsoft 365, but that number is not an outage total.

Microsoft said the technical issue arose during an upgrade following a recent network change. It described servers becoming unexpectedly unavailable during automated checks and said recovery was gradual. The company said there was no indication of malicious activity and no evidence that the incident was a cyberattack. It apologised and warned that some users could continue to experience intermittent access problems or delayed email delivery even as services recovered. That is a more useful account than calling the outage a cyberattack or implying that every Microsoft product was down everywhere.

The economic mechanism is straightforward. A company may have moved email, calendars, document sharing, meetings and identity management into one cloud ecosystem because the integration reduces administrative cost. The same integration means that a fault in a shared service can disable several business processes at once. Employees may still have laptops, local files and phones, but they may not have the authentication, shared mailbox, meeting or workflow data needed to complete a task. A company that has a second provider in theory may discover that its backup is not configured, its users have not been trained, or its data cannot be moved quickly enough to use it.

The British health service showed how quickly that can become a public-service problem. [NHS England](/organizations/nhs-england) said local services were using alternative IT systems or paper records to maintain essential care and appointments. Rachel Harrison, the GMB union's public-services head, said staff receiving emergency calls could not refer patient details in the normal way, creating a serious backlog. She added that emergency and life-threatening call-outs appeared unaffected. That qualification matters: the union described a disruption to the information and referral process, not a collapse of the emergency service. The NHS urged patients to continue using emergency numbers and local services, while the National Pharmacy Association said its members were having trouble communicating with doctors' offices about prescriptions.

The same distinction applies to small firms. A large employer may have an IT recovery team, secondary communications channels and a contractual service manager. A small clinic, supplier or accountancy practice may have one Microsoft tenant, a few staff and no tested alternative. Its direct loss can be a missed appointment, a late invoice, an order that cannot be processed or hours spent reconstructing a work queue. The largest loss is not necessarily the value of the unavailable software. It is the value of the business process that depended on it.

## Insurance will not pay for every hour of lost work

Morningstar DBRS initially estimated that insured payouts from the Microsoft outage could total $300 million to $500 million, compared with an estimated $400 million to $1.5 billion in insured losses from the 2024 CrowdStrike outage. Those figures should not be confused with total economic damage. The Insurer reported a separate preliminary estimate from cyber-risk firm Kovrr of $300 million to $1.4 billion in insured losses and about $16 billion in lost productivity. Kovrr described its calculation as a back-of-the-envelope estimate and assumed that the outage would be resolved by 7 p.m. GMT on Friday. It was not a final claims tally.

The gap between lost productivity and insured loss is the important point for small employers. Some cyber policies cover malicious attacks but exclude accidental technology-provider failures. Business-interruption cover may require a waiting period of eight to twelve hours before it responds. A business that lost a working day may therefore have a real economic loss but no payable claim. Even where cover exists, a claimant must demonstrate the interruption and its financial effect. The insurer's exposure is not automatically equal to the value of every delayed email or missed appointment.

For insurers, the outage is a test of policy wording and accumulation risk. A single provider can create correlated claims across hundreds of otherwise unrelated customers. That makes the event different from a conventional local equipment failure. It also gives insurers a reason to ask clients whether they have tested offline procedures, alternative communications and recovery plans. The cost may ultimately appear in premiums, exclusions or higher deductibles rather than a single large payout.

## Who gains leverage and who bears the cost

Passengers and airlines pay first when departures stop. Travellers absorb missed connections, hotel costs, lost work and uncertainty about rebooking. Airlines must reorganise aircraft, crews and passenger flows, often after schedules have already been disrupted by the first restriction. Airport workers face irregular shifts and crowded terminals. Those costs are not all borne by the airline that caused no technical fault; they are spread across travellers, contractors and local businesses around airports.

Hospitals, pharmacies and small firms bear a different cost from Microsoft's outage. They may have to switch to paper or alternative systems, rebuild records and explain delays to customers. Microsoft loses trust and may face tougher procurement conditions. Competitors gain negotiating leverage because buyers can now ask for proof of portability, independent identity systems and credible fallback arrangements. That does not mean a rival provider automatically wins Microsoft's customers or revenue. Switching a large organisation is expensive, and alternative services can depend on the same underlying network, identity or cloud infrastructure.

Regulators and public purchasers gain leverage to demand better resilience evidence. The UK parliamentary science and technology committee has already warned about dependence on a small number of US technology providers. Friday's incident gives that concern a concrete operational example. The right response is not to ban a supplier because of one technical failure. It is to ask whether the buyer can continue critical work if the supplier is unavailable, whether the supplier reports incidents quickly, and whether contracts require tested recovery objectives.

The FAA modernisation debate is the public-sector version of the same argument. Supporters of new funding can point to the ground stop as evidence that old infrastructure has a cost. Congress can also use the incident to insist that the money be tied to reliability, redundancy and independent testing. The agency gains a stronger case for investment, but it also inherits a stronger burden to explain what failed and what each dollar will change.

## What to watch next

For aviation, the next decisive evidence is the FAA's technical account: the failed component, the contractor or network involved, the duration of degraded service and the fallback capacity available at each centre. Congress should ask for a schedule that distinguishes communications upgrades from broader airport spending. A new appropriation should be judged against measurable uptime and recovery targets, not the size of the headline.

For Microsoft, the useful follow-up is a post-incident report that explains the network change, why automated checks did not prevent the disruption, how recovery was sequenced and what safeguards will stop a similar failure from spreading. Buyers should look for service-level changes and tested alternatives, not just assurances that the system is stable again. NHS and other public services should report whether their paper and alternative-system procedures were fast enough to preserve patient information and avoid backlogs.

The wider indicator is not a one-day movement in consumer prices. The public spending side belongs in the [Fiscal-Policy Sustainability](https://factrail.ai/drivers/fiscal-policy-sustainability) discussion: more resilient systems may cost more upfront, but repeated failures also create a bill that is paid through lost output, emergency workarounds and compensation.

The conclusion is not that the FAA and Microsoft have the same technical problem. They do not. It is that modern economies have built essential functions around shared networks, and the cost of a failure is distributed far beyond the operator. Resilience should be priced and tested before the next outage, not improvised after passengers, patients and small employers have already paid.

## Evidence boundaries

Established: the FAA ordered a nationwide departure stop after communications problems were reported across 22 centres; Microsoft 365 services were disrupted across multiple regions; NHS services used contingency procedures. Strong indication: both events exposed concentration and recovery weaknesses in essential infrastructure. Not established: a shared cause, a malicious attack, a final insurance-loss total, or a direct measurable change in consumer inflation.

## Sources

- [Reuters — FAA issues national ground stop due to extraordinary circumstances and safety reasons, 13 November 2026](https://www.reuters.com/world/us/faa-issues-national-ground-stop-due-extraordinary-circumstance-reasons-safety-2026-11-13/)
- [Associated Press — The latest: FAA grounds all flights nationwide due to communications problems, 13 November 2026](https://apnews.com/article/6aaa6bd4991ef22eb0c3baa6961a91d1)
- [Federal Aviation Administration — ATCSCC Advisory 111, National Ground Stop, 13 November 2026](https://www.fly.faa.gov/adv/adv_otherdis?adv_date=11132026&advn=111)
- [Government Accountability Office — Air Traffic Control Systems: Ambitious New Modernization Effort Needs Cost and Schedule Planning, 2026](https://www.gao.gov/assets/gao-26-107992.pdf)
- [Associated Press — Global Microsoft outage hits companies and governments, 13 November 2026](https://apnews.com/article/d92547a821930290db2964388016979a)
- [Microsoft 365 Status — service recovery update, 13 November 2026](https://x.com/MSFT365Status/status/2121266712884416537)
- [AFP — England’s health service triggers contingency after Microsoft outage, 13 November 2026](https://www.arabnews.com/world/englands-health-service-triggers-contingency-after-microsoft-outage-3011662)
- [The Wall Street Journal — Microsoft outage unlikely to cost insurers as much as CrowdStrike’s did, 13 November 2026](https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-11-13-2026/card/microsoft-outage-unlikely-to-cost-insurers-as-much-as-crowdstrike-s-did-ZwOqlIgFFEalc94ON5rL)
- [The Insurer — Microsoft outage could generate $700 million in insured losses: Kovrr, 13 November 2026](https://www.theinsurer.com/ti/news/microsoft-outage-could-generate-700-million-in-insured-losses-kovrr-2026-11-13/)

## Methodology

https://factrail.ai/methodology

Canonical: https://factrail.ai/articles/faa-microsoft-outages-aviation-cloud-resilience-november-2026
