Between November 12 and November 17, six critical systems went down: Microsoft 365, the FAA's air traffic network, JPMorgan Chase, United Airlines, the NSA's classified network and Charles Schwab's website. None of the six has been blamed on a cyberattack.

Six critical systems failed between November 12 and 17. None was blamed on a cyberattack.
These weren't startups. They were some of the best-funded, most heavily engineered systems in the world, and they failed for ordinary reasons: an old upgrade, a software bug, a telecom fault.
That matters because your startup runs on a stack of systems like these, and almost none of them are yours.
The fix isn't more engineering. It's a map: which vendors you depend on, what breaks when each one fails, and what you'll do in the first hour.
Below is what happened, system by system, what the market was doing around it, and the Vendor Map Playbook to find your own weak points.
Inside you'll find:
1. The Backdrop: Three Weeks That Cost the Market Trillions
2. The Office: 24 Hours Without Outlook, Teams or Copilot
3. The Sky: One Telecom Fault, 12,462 Cancelled Flights
4. The Money: When the Bank Is the Outage
5. The Pattern: Six Failures, Zero Attackers
6. The Vendor Map Playbook: Map, Score, Prepare, Protect, Respond
1. The Backdrop: Three Weeks That Cost the Market Trillions
The outages didn't land in a calm market. They landed at the end of three of the most volatile weeks of the year.
On October 29, attackers drained about $6.3 billion from Binance's hot wallets, the largest crypto theft on record. On November 5, two days after the midterm elections, the Nasdaq Composite fell 5.52% and the S&P 500 fell 4.35%, their worst session since April 2025. About $3 trillion of value left the S&P 500 in one day, roughly $450 billion of it from Nvidia alone.

The November 5 sell-off: every major US index fell more than 4%.
Four forces stacked up:
AI economics: investors questioned the pace, cost and financing of the AI data-center build-out, and a weak forecast from Qualcomm added fuel.
Policy: the midterm results and calls in Congress to pause advanced AI put regulation back on the list of risks.
Rates: a hotter October inflation report on November 10 led traders to fully price a December rate hike.
Key-person risk: on November 12, Jensen Huang resigned as Nvidia's CEO after 33 years, and the stock fell as much as 14.3% during the session.
Even so, the S&P 500 ended the week of November 13 up about 2.5%, snapping a four-week losing streak.
→ When public valuations swing like this, private rounds usually feel it next. The founders who raise well in that market are the ones who can show investors exactly what could go wrong, and what they've already done about it.
2. The Office: 24 Hours Without Outlook, Teams or Copilot
The first failure hit the tool most startups open before their first coffee.
On Thursday, November 12, Microsoft 365 started failing in the Americas. Microsoft's own incident record dates the impact from 16:20 UTC on November 12 to 16:10 UTC on November 13: 23 hours and 50 minutes. Users reported problems in 74 countries and territories, and Downdetector logged 138,058 reports of Outlook problems in the US at the peak. Some businesses fell back on fax machines to confirm trades.

Microsoft 365 was down for almost a full day, the longest of the week's outages.
The detail founders should notice: the outage also broke the ChatGPT and Claude connectors that plug into Microsoft 365. If your AI workflows read your inbox, your calendar or your files, they went down with it.
Microsoft found no evidence of malicious activity. It traced the outage to a fault left behind by an earlier upgrade that had added capacity to its network. A change made to improve the system is what took it down.
→ Your productivity suite isn't a utility. It's a vendor, and every tool you connect to it inherits its downtime.
3. The Sky: One Telecom Fault, 12,462 Cancelled Flights
The next afternoon, the failure moved from offices to airports.
At 3:54 p.m. ET on Friday, November 13, the FAA ordered a nationwide ground stop after a telecommunications outage hit all 22 of its en route air traffic control centers. About 3,100 aircraft were in the air, bound for US airports. The FAA lifted the order around 8:15 p.m., after more than four hours. Operators cancelled about 12,462 flights during the disruption, although the FAA notes that not all of them were caused directly by the ground stop.

How one telecom outage turned into more than 12,000 cancelled flights.
It was only the second nationwide ground stop since 9/11. It also came hours after the White House and the Department of Transportation asked Congress for an extra $30.1 billion to modernize aviation infrastructure.
Three days later, on Monday, November 16, United Airlines had an outage of its own. It hit several applications pilots rely on, including crew messaging and the system that provides the final weight of each departing aircraft. AP put the disruption at about 90 minutes; a United spokesperson told CNN the outage itself lasted about 30 minutes and affected fewer than 150 flights. The FAA made a point of saying it was not an FAA equipment issue.
→ Some dependencies can't be replaced. There is no second FAA. For those, the plan isn't a backup vendor: it's knowing in advance how your business keeps running while they're down.
At startup scale, it looks like this. Picture an events startup with a 300-person conference on Saturday and a third of its speakers flying in on Friday afternoon. It can't fix the FAA. It can decide in advance which sessions move online, who calls each speaker, what attendees are told and when, and which costs it covers. Written down on Monday, that's a 30-minute decision on Friday. Improvised on Friday, it's a lost weekend and a pile of refunds.
4. The Money: When the Bank Is the Outage
On Monday, November 16, the failures reached the place where startups keep their cash.
Starting Monday morning, a software bug in an on-premise hardware system at JPMorgan Chase slowed online banking, branch transactions and Zelle payments. Downdetector logged nearly 5,000 reports for Chase at the peak. Customers of Bank of America, Wells Fargo, U.S. Bank, Capital One and Truist reported problems too, because their transfers ran through Chase. The bank said customers kept access to their accounts and funds, and that transactions were posting normally again by about 6:30 p.m.
The next morning, Charles Schwab's website started returning a server error. Schwab called it a brief webpage loading issue, resolved around 10:38 a.m. ET. The useful part is what still worked: trading remained available, and clients could reach their accounts through a separate login page and the Schwab mobile app. A second door is a plan B.
→ A bank doesn't have to fail to hurt you. One day of delayed payments can mean a late payroll or a supplier who stops shipping.
The extreme version of this risk has already happened to startups. When Silicon Valley Bank collapsed in March 2023, Roku disclosed that about $487 million, 26% of its cash, sat there, largely uninsured. Roku could still tell investors it had enough to operate for the next twelve months, because the other $1.4 billion was spread across several other large banks.

Roku had 26% of its cash at SVB. The other 74%, at other banks, is what kept it operating.
5. The Pattern: Six Failures, Zero Attackers
On Tuesday, November 17, the list reached the intelligence community. NSA Net, the classified network the National Security Agency uses for internal communications, went offline, and the agency staggered staff arrivals at its headquarters while it worked out the scope of the problem. Officials said there was no evidence of a cyberattack, and four people familiar with the matter told Nextgov it may be linked to a series of software upgrades.
Put the six failures side by side and a pattern appears.

The reported cause of each outage. The only attack of the month, Binance, came through one laptop.
Most founders picture risk as an attacker. This week, none of the six failures was blamed on one. Three were linked to upgrades or software bugs. The risk was change.
The one real attack of the month makes the same point from the other side. Binance said it found no evidence that a wallet key had leaked; its first findings pointed to one engineer's laptop instead. And it survived because it had money set aside: it reserved more than $6.34 billion of its own funds to cover users.
→ The question for your startup isn't "could we be hacked?" It's "which single vendor, person or device could stop us, and what would we do in the first hour?"
Three questions worth answering before your next round:
Which vendors could stop your product from working, and for how long?
Where is your cash, and how much of it sits in one bank?
Who has access to production, and what happens if that person is gone for two weeks?
If you can't answer each in one sentence, the Vendor Map Playbook below is built for you.
What's next: the Vendor Map Playbook, a 14-day plan to find, score and protect the dependencies that could stop your startup. It includes a 60-second self-test, a map of eight dependency categories with what founders usually forget in each, a scoring formula and an outage-cost calculation with a worked example, a contract check for your vendors' SLAs, if/then plans with real cases from SVB, CrowdStrike and AWS, an access checklist, a first-hour incident runbook, six tools to start this week, five copy-paste prompts and the Vendor Map Kit to run it all.
Subscribe to CLUB to read the rest.
Become a paying subscriber of CLUB to get access to this post and other subscriber-only content.
UpgradeAlready a subscriber? Sign in
A subscription gets you:
- Everything included in Premium.
- Early access to new tools and the specialized AI and product-market fit newsletter.
- One-on-one feedback on your business, with actionable guidance on strategy, positioning, and growth
- Masterclasses with Pablo Fernandez Alvarez
