Attack or accident, down is down
Between November 12 and 17, Microsoft 365, the FAA, Workday, Chase and Zelle, United, the National Weather Service, the NSA and Apple went down. Everyone asked who did it. The better question is what a day down costs you, and whether your protection matches it.
- Eight outages in six days, November 12 to 17, and every stated cause was a bug, a change or maintenance. For the people locked out, whether it was an attack made no difference.
- No attacker, real money: Parametrix estimated CrowdStrike's 2024 outage cost the Fortune 500 $5.4 billion, with no more than 10 to 20 percent covered by cyber insurance.
- Price a day down by category: lost sales, SLA credits, refunds, idle payroll, fines, lost customers and, for contractors, CPARS. Then size protection to that number.
Attack or accident, down is down. Between November 12 and 17, eight major systems failed, and none of the stated causes was an attack: a dormant bug, a contractor’s router work, a disaster-recovery test, maintenance, an internal update. The week’s question was whether America was under a coordinated cyberattack. The more useful one is what a day down costs you, and whether your protection is sized to it.
Eight outages in six days
Microsoft 365 email went down for about a day. The FAA issued a nationwide ground stop. Workday payroll and timekeeping went offline for some customers. Chase transactions slowed and some Zelle payments were delayed. United departures were delayed. On Tuesday the National Weather Service, the NSA and Apple reported disruptions. Our earlier piece, Bug, insider or attack?, covered the first five; this one adds the Weather Service, the NSA and Apple through November 17.
The day each disruption began, with the cause the organization or officials gave.
- Nov 12Microsoft 365A latent bug from a July update
- Nov 13FAAData flooding during a contractor's router work
- Nov 13WorkdayA recovery check that reached production
- Nov 16Chase and ZelleA software bug, Chase said
- Nov 16UnitedA technology issue
- Nov 17Weather ServiceRoutine network maintenance
- Nov 17NSAScheduled routine maintenance
- Nov 17AppleAn internal update
Data
| When | Milestone | Detail |
|---|---|---|
| Nov 12 | Microsoft 365 | A latent bug from a July update |
| Nov 13 | FAA | Data flooding during a contractor's router work |
| Nov 13 | Workday | A recovery check that reached production |
| Nov 16 | Chase and Zelle | A software bug, Chase said |
| Nov 16 | United | A technology issue |
| Nov 17 | Weather Service | Routine network maintenance |
| Nov 17 | NSA | Scheduled routine maintenance |
| Nov 17 | Apple | An internal update |
That is the wrong question
Secretary of State Marco Rubio said officials have “no evidence” the incidents are related to each other and “zero evidence” that any of them has anything to do with “nefarious actors” (White House transcript; the Associated Press). Director of National Intelligence Jay Clayton pointed to patching: “Lots of institutions, because these vulnerabilities have been identified, are patching. They’re taking their systems down” (White House transcript; first reported by Fox News).
Maybe every stated cause holds up. Maybe an investigation finds something else. For the employee who couldn’t check pay or the traveler stuck at the gate, it made no difference at all.
A vulnerability doesn’t care who finds it. An adversary, a bad change ticket or an untested failover all end in the same place: down.
Eight failures, eight exposed weak points
Every one is a weakness an adversary would use, and every one took a system down with no adversary at all
| Who | Stated cause, as reported | The weakness it exposed (our read) |
|---|---|---|
| Microsoft 365 Nov 12 | A latent bug from a July update, hit by extra load after a Teams failover while nearly 14% of capacity was out (Microsoft’s preliminary report, via BleepingComputer) | Containment and recovery built for partial failure. More than 80% of gateway servers stopped responding. |
| FAA Nov 13 | Data flooding its telecom network; contractor L3Harris had been installing a new router at a remote site (FAA) | One change at one remote site destabilized 17 of 22 en route centers. |
| Workday Nov 13 | During a disaster-recovery check, recovery-site services “unexpectedly connected to Production databases” (Workday status notice, via UC Today) | The safety net and the live system were not walled off from each other. |
| Chase and Zelle Nov 16 | Chase: “a software bug in an on-premise hardware system,” not cyber-related. Zelle said its own network was operating normally (Al Jazeera) | One bank’s processing problem delayed payments for customers of other banks. |
| United Nov 16 | A technology issue; pilots told controllers it involved the system that works out a plane’s final weight (AP) | Departures waiting on one shared system. |
| National Weather Service Nov 17 | Routine network maintenance; about 40 minutes of website and data disruption (NWS, via Fox News and WINK News) | Maintenance that reached public data services. NOAA Weather Radio was not affected. |
| NSA Nov 17 | “Temporary degradation during scheduled routine maintenance” (NSA). Routine system patching, CBS News reported, citing sources (CBS News) | Maintenance on a core network reached the workforce: staggered arrivals. |
| Apple Nov 17 | “An internal update” (Apple, via The New York Times) | One change reaching payments, sign-in and the App Store. |
Each stated cause is what the organization or officials said, as reported; several are preliminary, and Microsoft’s final report is due November 20. The right column is our read, not the organization’s finding. We take no position on whether any outage was an attack.
Down is down
No attacker still means real money. CrowdStrike’s faulty July 2024 update was “not a security incident or cyberattack,” its CEO said (TechCrunch). Parametrix estimated it hit a quarter of the Fortune 500 and cost those companies, excluding Microsoft, $5.4 billion in direct losses, with no more than 10 to 20 percent covered by cyber insurance.
Attackers use outages too. On November 13, Microsoft Threat Intelligence saw phishing emails posing as outage notices that cited the real incident number (SecurityWeek). And in AA24-038A, CISA, the NSA and the FBI warned that China-backed Volt Typhoon actors were pre-positioning on U.S. critical infrastructure networks, some for at least five years. A week like this shows them where the single points of failure are.
Your vendors’ weak points are your weak points
Workday’s website says more than 11,500 companies use it (FOX 35). None made the change that locked their employees out. Zelle’s owner is itself owned by banks including JPMorgan Chase, and some Zelle users reported delays when Chase’s processing slowed (Al Jazeera). Over 10 years, the Uptime Institute has found third-party providers behind a majority of publicly reported outages.
Breaches follow the same path. Verizon’s 2026 Data Breach Investigations Report found a third party involved in 48 percent of breaches, up 60 percent in a year; the 2025 report had it rising from 15 to 30 percent. Only 23 percent of third-party organizations fully fixed missing or improperly secured MFA on their cloud accounts. An annual questionnaire shows a vendor on the day it filled out the form. A control verified last quarter is a memory. Real-time beats point-in-time.
Protect it like it’s worth what it’s worth
Our read: many organizations buy security and IT support as a line item, benchmarked against last year’s budget or a peer’s. Size it instead to what it protects: the revenue that stops when systems stop, the data that can’t be recreated, the mission that can’t wait.
If a day down costs millions and your resilience budget is a rounding error, you haven’t managed the risk. You’ve accepted it blind.
What does a day down cost you?
Break it into pieces. The surveys say they are large: 57 percent of operators in the Uptime Institute’s latest survey said their most recent major outage cost more than $100,000, and one in five said more than $1 million. Splunk and Oxford Economics put downtime at about $200 million a year per Global 2000 company.
What a day down costs, by category
| Cost | What it looks like | What the data says |
|---|---|---|
| Lost sales | Orders and billable work that stop | Lost revenue averages $49M a year; CFOs say revenue takes 75 days to recover (Splunk and Oxford Economics) |
| Lost customers | Customers who leave or buy less | 29% of respondents say they have lost customers to downtime; 40% of CMOs say it cuts customer lifetime value (Splunk) |
| SLA breach and service credits | Credits owed for a missed availability promise | Missed SLA penalties average $16M a year (Splunk). Your own credits depend on your contracts: price them yourself |
| Money owed back | Refunds, settlements, liquidated damages | Settlement and legal costs average $15M a year (Splunk). Federal liquidated damages are set per contract as “a reasonable forecast of just compensation” (FAR 11.501); no public average |
| Contract compliance | Missing a measured performance standard | Performance-based contracts carry measurable standards and a surveillance plan (FAR 37.601 and 37.604). No public number: read your contract and price it yourself |
| Regulatory fines and reporting | Fines, and reporting clocks that run during recovery | Regulatory fines average $22M a year (Splunk). Cyber incidents go to DoD within 72 hours of discovery (DFARS 252.204-7012); FedRAMP cloud providers report security incidents within one hour (FedRAMP) |
| Credibility and brand | Share price, reputation, past-performance record | Stock can fall up to 9% after one incident (2.5% on average) and take 79 days to recover; brand trust campaigns average $14M (Splunk). CPARS ratings inform source selection for three years (FAR 42.1503) |
| Idle payroll | Staff paid but unable to work, plus overtime | Lost productivity averages $12M and overtime wages $11M a year (Splunk). Your payroll: price it yourself |
Splunk figures are 2024 annual averages per Global 2000 company, modeled from a survey of 2,000 executives, not a per-outage cost.
For a federal contractor an outage can also start a paper trail. Schedule performance goes into CPARS, which agencies use in source selection for three years after the work ends (FAR 42.15). A failure that endangers performance can bring a 10-day cure notice, then termination for default (FAR 49.402-3). Defaults are recorded in FAPIIS.
A day down, as a formula to fill in: (revenue per day × share lost) + SLA credits and penalties owed + refunds and liquidated damages + (idle staff hours × loaded hourly cost) + overtime to recover + expected fines + (customers lost × their annual value). Add what you can’t price well, brand and past performance, then put your spending on support, testing and redundancy next to the total.
Our forecast
Our forecast: by June 30, 2027, Verizon’s 2027 Data Breach Investigations Report will put third-party involvement at 40 percent of breaches or higher. If it lands lower, we will say so here.
Our question: what does a day down cost you, and what do you spend each year on support, testing and redundancy to prevent one?
What we’d do first: a right-sized stack
- IT support that knows your environment. Monitoring, incident response and people on shift around the clock who can tell a bad change from an intrusion fast. Our 2027 forecast makes the case for one 24/7 desk that watches outages and attacks together.
- Penetration testing and independent assessment. Outside eyes find what your team can’t see. Test against an attacker and against your own failures.
- Redundancy that has actually been tested. Failover, backups and recovery sites walled off from production, plus rehearsed manual workarounds for payroll, payments and communications.
- Change control with a safety catch. Staged rollouts and tested rollback, yours and your vendor’s. Microsoft says its normal rollback could not quickly revert a months-old bug.
- Vendor risk scored all year. Know which vendors can stop your mission, ask for change and recovery evidence, and put notification rights in the contract.
- Security that switches on when things break. Treat every outage as a possible security event: warn about phishing within the hour and tighten payment approvals until systems are back.
The bottom line
Nobody had to prove these outages were attacks for the damage to be real. The next one might be a nation-state, a contractor’s router or a bug that’s been sleeping since summer. You won’t get to choose. You can choose whether your protection matches what you stand to lose.
That is the business we’re in: making risk manageable. Our team has deep experience running Tier 1 through Tier 3 support around the clock and delivering security operations center as a service, along with incident response, penetration testing and independent security assessment; AccelSecure™ is the framework behind our 90 to 180 day ATO target, and our Next-Gen GRC work, including AusperRiskIQ™, is in beta. Our work has supported the intelligence community, the Department of Homeland Security (DHS), the Department of War (DOW) and Google, among others. If you can’t say what a day down costs you, start a conversation with us at contacts@auspertech.com.
Common questions
Were any of the November 12 to 17 outages cyberattacks?
How much does an outage with no attacker cost?
What can an outage trigger for a federal contractor?
How should we size spending on IT support and resilience?
Sources
- Microsoft, Preliminary Post Incident Report for Microsoft 365, EX1550543/MO1550821 (report date Nov 15, 2026), copy published by NextMicrosoft's own preliminary PIR: three causes, 80% of servers unresponsive, bulkheads did not stop the cascade, rollback could not quickly revert, final PIR due 11/20/2026. The admin-center original is not public. Accessed Nov 18, 2026.
- BleepingComputer, Microsoft says months-old bug caused global Microsoft 365 outage (Nov 16, 2026)Reports the preliminary PIR; final report expected later that week. Accessed Nov 18, 2026.
- FAA, Nationwide Ground Stop: Questions and Answers (Nov 15, 2026)Nationwide ground stop; data flooding the FTI network; L3Harris router work; 17 of 22 centers; no indication of a cyberattack. Accessed Nov 18, 2026.
- FOX 35 Orlando (LiveNOW from FOX), Workday outage means employees across thousands of businesses can't access accounts (Nov 2026)Workday spokesperson statement; Workday's website says more than 11,500 companies use it. Accessed Nov 18, 2026.
- UC Today, Workday's backup test took payroll and timekeeping offline (Nov 2026)Quotes Workday's Community Status Page notice (via Washington University IT) and Workday's restoration statement to ABC10. Accessed Nov 18, 2026.
- Al Jazeera, US bank outage: What caused Chase problems and Zelle payment delays? (Nov 17, 2026)Chase spokesperson statement; Zelle network operating normally; Early Warning Services ownership. Accessed Nov 18, 2026.
- HuffPost (Associated Press), United Airlines resolves tech issue after widespread system outage (Nov 16, 2026)United statement; pilots' account of the weight system. Accessed Nov 18, 2026.
- Fox News, Tech outages live updates (Nov 17, 2026)NWS statement on routine network maintenance (11:05 to 11:45 a.m. ET; NOAA Weather Radio not affected); Rubio's remarks; DNI Clayton on patching (Fox Business); lawmakers' requests for briefings. Accessed Nov 18, 2026.
- The White House, “No Indication of a Cyberattack”: DNI Clayton and Secretary Rubio on Recent Tech Outages (Nov 18, 2026)Transcript excerpts: Rubio's “no evidence” and “zero evidence” remarks; Clayton's “Lots of institutions ... are patching” remarks. Accessed Nov 18, 2026.
- ABC News (Associated Press), Rubio says 'zero' information that US outages are linked or caused by a malign actor (Nov 18, 2026)Rubio's full “zero evidence” quote, said to reporters in Manila. Accessed Nov 18, 2026.
- WINK News, National Weather Service websites and data services restored after interruption (Nov 17, 2026)The NWS post: routine network maintenance; resolved. Accessed Nov 18, 2026.
- CBS News, NSA's classified internal network goes down, sources say (Nov 17, 2026)NSA spokesperson statement; sources on routine system patching; staggered arrivals. Accessed Nov 18, 2026.
- Honolulu Star-Advertiser (The New York Times), Apple services disrupted after internal update (Nov 17, 2026)Apple's statement; Apple Pay and iCloud problems reported. Accessed Nov 18, 2026.
- TechCrunch, Faulty CrowdStrike update causes major global IT outage (Jul 19, 2024)Accessed Nov 18, 2026.
- Parametrix, CrowdStrike to Cost Fortune 500 $5.4b; Insured Loss Range of $0.54b to $1.08b (Jul 2024)A quarter of the Fortune 500 (125 companies); excludes Microsoft; no more than 10% to 20% insured. Accessed Nov 18, 2026.
- SecurityWeek, Microsoft rules out cyberattack in outage, warns of phishing campaign (Nov 16, 2026)Microsoft Threat Intelligence: phishing citing EX1550543, messages detected November 13. Accessed Nov 18, 2026.
- CISA, NSA, FBI and partners, AA24-038A: PRC State-Sponsored Actors Compromise and Maintain Persistent Access to U.S. Critical Infrastructure (Feb 7, 2024)Copy on the FBI's IC3 site; cisa.gov refused automated requests. Accessed Nov 18, 2026.
- Verizon, 2026 Data Breach Investigations Report, Executive SummaryThird-party involvement 48% of breaches, up 60%; 23% of third parties fully remediated cloud MFA gaps. Accessed Nov 18, 2026.
- Verizon, 2025 Data Breach Investigations ReportThird-party involvement doubled from 15% to 30%. Accessed Nov 18, 2026.
- Uptime Institute, Annual Outage Analysis 2026 (May 2026)57% say their most recent major outage cost more than $100,000; one in five more than $1 million; third-party providers a majority of publicly reported outages over 10 years. Accessed Nov 18, 2026.
- Splunk and Oxford Economics, The Hidden Costs of Downtime (2024)Per Global 2000 company per year: lost revenue $49M, regulatory fines $22M, SLA penalties $16M, settlement and legal $15M, brand trust campaigns $14M, lost productivity $12M, overtime $11M; 29% lost customers; stock down up to 9% (2.5% average), 79 days to recover. Accessed Nov 18, 2026.
- Cisco newsroom, Splunk report shows downtime costs Global 2000 companies $400B annually (Jun 11, 2024)Accessed Nov 18, 2026.
- Acquisition.gov, FAR Subpart 42.15, Contractor Performance InformationCPARS; past performance used in source selection within three years of completion; FAPIIS terminations for default. Accessed Nov 18, 2026.
- Acquisition.gov, FAR 49.402-3, Procedure for defaultWritten notice with 10 days to cure before termination for default. Accessed Nov 18, 2026.
- Acquisition.gov, FAR Subpart 37.6, Performance-Based AcquisitionMeasurable performance standards, incentives, quality assurance surveillance plans. Accessed Nov 18, 2026.
- Acquisition.gov, FAR Subpart 11.5, Liquidated DamagesAccessed Nov 18, 2026.
- Acquisition.gov, DFARS 252.204-7012, Safeguarding Covered Defense Information and Cyber Incident Reporting'Rapidly report' means within 72 hours of discovery. Accessed Nov 18, 2026.
- FedRAMP, Incident Communications Procedures (playbook)Report suspected and confirmed incidents within one hour of identification. Accessed Nov 18, 2026.
Updates
- Nov 18, 2026. Checked the Rubio and Clayton quotes against the White House transcript and the Associated Press, and the Weather Service row against the agency's own post. Every attribution stands. Rubio is now quoted more exactly, and those sources are added.
Share on LinkedInFollow AusperSubscribe by RSS
Related reading
Working on this now?
Tell us where your program stands and we'll tell you what we would do next.
Talk to Ausper