san jose collage
San Jose
Serving San Jose, California (Santa Clara County)Vol. 1, No. 24
Technology

San José airport took a hit from the FAA grounding, and the cause is still unexplained

Planes sat on the ground at SJC and dozens of flights were canceled. People familiar with the investigation, as reported by CNBC, say engineers have reproduced the overload that grounded U.S. flights, but the FAA has not found its cause, Airbyte has been down since Tuesday, and a separate attack hit online checkout pages.

ShareFacebookXEmail
Two travelers look up at a wall of Delta departure screens at New York’s JFK Airport, where many flights are marked canceled in pink.
Travelers check a departures board full of cancellations at New York’s JFK Airport during the FAA’s nationwide ground stop on Nov. 13, 2026. — Photo: Sashimi-b / Wikimedia Commons, CC BY-SA 4.0

SAN JOSE — The outage week isn’t over. Engineers have reproduced the network overload that grounded every U.S. departure last Friday, people familiar with the investigation told CNBC, but the FAA still can’t say why a new router at an Arkansas site misbehaved, and it has frozen installs of that model. A data-pipeline company has been down since Tuesday, and a real hack hit online checkout pages.

San José airport: planes held, flights canceled

The FAA’s Nov. 13 grounding hit San José Mineta International Airport (SJC) directly. The FAA says it issued the stop at about 3:30 p.m. Eastern, which is 12:30 p.m. PT. KQED put the advisory shortly before 1 p.m. PT. At SJC, some planes were already boarded, and passengers told ABC7 they sat on them for hours before getting off. San José State’s football team was taxiing on its charter when the stop began and waited about three hours, then went back to campus. It planned to fly to Colorado Springs the next morning, ABC7 reported.

The FAA says air traffic resumed at 8:15 p.m. Eastern (5:15 p.m. PT). SJC and the other Bay Area airports told KQED the stop was lifted around 5:30 p.m. PT. SJC spokesperson Ana Maria State said delays would continue as airlines got back on schedule, and she told travelers to check with their airline.

How many SJC flights were hit? The counts depend on who is counting. ABC7 reported nearly two dozen SJC flights canceled by late afternoon Friday. KQED later cited more than 80 delays and over 30 cancellations at SJC as of 3:30 p.m. The flight tracker FlightAware counted 102 cancellations at SJC for all of Friday and 49 on Saturday. NBC Bay Area, citing Flightradar24, reported 41 cancellations and 180 delays at SJC in its Saturday update. By 9 a.m. Sunday, FlightAware showed 46 delays and 12 cancellations at SJC, NBC Bay Area reported. The FAA notes that not every cancellation was directly caused by the stop.

SJC told KQED on Saturday morning that its operations were “proceeding normally.” On Monday, SJC officials told NBC Bay Area that United’s separate outage caused no cancellations or major delays there. Early Thursday, the FAA’s airport status page listed SJC as “On Time.”

The FAA still lacks a root cause

According to the FAA’s Q&A page, the agency’s network became unstable when a flood of data hit it at a rate of 1,000 connection attempts every four to five seconds. The FAA traces the surge to two sites: a radio site in El Dorado, Arkansas, where contractor L3Harris had installed a new router, and a radar site in Amarillo, Texas. The FAA says it disabled the broadcast function on more than 60 other Juniper routers and has paused installing them.

Engineers have reproduced the overload in a Florida lab, people familiar with the investigation told CNBC on Wednesday, but they have not yet determined why the routers failed. CNBC’s sources spoke of two routers, while the FAA and Reuters describe a single new router at the El Dorado site, with a second site, in Amarillo, also under review. On a streaming show Wednesday, Transportation Secretary Sean Duffy said the stop might have been shorter if L3Harris had caught the problem sooner, The Associated Press reported. The FAA says it has no indication of a cyberattack.

A week of breakdowns

The grounding was part of a larger pileup. First, work email went dark. Then every departing flight in America sat on the ground. Then bank transfers stalled, Apple’s services stumbled and a classified National Security Agency network went down. No wonder people are asking: Is someone attacking us?

Officials say no. But they stop short of 100% certainty, and the companies haven’t finished their investigations. Their leading explanation is a twist: the damage came mostly from repairs.

AI tools are finding software flaws faster than ever, and companies are racing to fix them. “As you patch and upgrade your systems, taking them down and bringing them back up, there are things like this that happened,” Director of National Intelligence Jay Clayton told CNBC, according to ABC News. Below: the latest problems, a day-by-day look at each breakdown, then answers to the big questions.

Airbyte stays down for a third day

Airbyte, whose cloud service moves data between business systems, first reported trouble at 3:22 a.m. PT Tuesday, its status page shows. By early Thursday it was still in a partial outage: syncs were resuming for more customers, but the web app and API were not yet back. The page gives no cause.

A real attack on checkout pages

Unlike the outages, this one was an attack. Optimizely, which sells website-experimentation software, says an unauthorized third party exploited a flaw in its Web Experimentation signup flow on Nov. 16 and planted malicious code in some customers' sites, according to its Nov. 17 incident report. The company says the code was meant to capture card details typed into site forms and to swap cryptocurrency addresses, and that it removed the code by 2:24 p.m. PT that day. It says no customer passwords were stolen. The report doesn’t link the attack to the week’s outages, and Optimizely’s investigation, with outside firm Mandiant, continues.

Smaller glitches Wednesday night and Thursday

DigitalOcean, a cloud host, reported increased error rates in its SFO3 region starting at 6:53 p.m. PT Wednesday. The company blamed a network routing problem after a load balancer node became unavailable and said it was operating normally by 9:46 p.m., per its status page. Cloudflare said a faulty Cloudflare Access release caused sign-in failures for some apps from 10:48 p.m. Wednesday to about 12:56 a.m. Thursday, according to its status history. Neither company has tied its glitch to the other events.

Customers also reported trouble Wednesday with T-Mobile, Spectrum, Wells Fargo and Outlook, The Economic Times and the Washington Examiner reported. This desk found no explanation from those companies, so those reports are treated as unconfirmed here.

Nov. 12: Microsoft 365 goes dark

Outlook, Teams and other Microsoft work tools failed for about a day for businesses, schools and governments worldwide. Microsoft says a hidden bug in a fix from July finally surfaced. Extra traffic hit servers that were already missing nearly 14% of their capacity. At the worst point, more than 80% of the affected servers stopped responding. Microsoft’s final report on the outage has not turned up yet.

Nov. 13: The FAA grounds the nation

The Federal Aviation Administration stopped every departing flight in the country for more than four hours. It was the longest nationwide ground stop since 9/11, Axios reported. Duffy said a single new router flooded the FAA’s network while a contractor swapped old copper lines for fiber. The failure hit 17 of the FAA’s 22 regional air traffic centers. The FAA says operators canceled about 12,462 flights on Friday alone, according to its Q&A page, while Reuters reported more than 18,500 cancellations by Monday morning, based on FlightAware data. The counts cover different days and scopes.

Nov. 16: Chase stalls

JPMorgan Chase customers struggled to move money for hours. The bank blamed a “software bug in on-premise hardware systems,” meaning equipment in its own buildings, CNBC reported. Zelle payments between Chase and other banks were delayed, too.

Nov. 16: United delays flights

United Airlines said a “technology-related issue” held up departures. About 150 flights were delayed, Reuters reported.

Nov. 17: The NSA stumbles

The NSA said its network “experienced temporary degradation during scheduled routine maintenance” but that “mission-critical capabilities remain fully operational.” Clayton said patches for a newly found flaw in PHP, a common programming language, kept some systems from coming back up.

Nov. 17: Apple goes down

Apple said an internal update knocked out services including iCloud sign-in, and users reported Apple Pay problems. Apple told CBS News everything was working again by about 2:46 p.m. PT.

Close to home

The Microsoft outage also reached local campuses and city offices. Stanford University told its community that “most users” could not reach Outlook email and calendars. San Jose Clean Energy, a City of San José department, posted that an email outage on Nov. 12 might delay its replies to customers. Apple, based in Cupertino, said an internal update caused its Nov. 17 outage; no source this desk opened says which Bay Area stores or offices were affected.

Investors dumped local software stocks Tuesday as the NSA news broke. Salesforce fell 9.45% and Intuit 7.17%, though nothing links either company to the outages. And Mountain View’s Google says three-quarters of its new code is now written by AI.

Are officials sure AI hackers weren’t involved?

No one claims total certainty. Listen to the hedges. “I can tell you 98% certainty this isn’t a cyberattack,” Homeland Security Secretary Markwayne Mullin told CNBC, Reuters reported. “Looking very diligently across the landscape, we are not seeing this as a cyber attack incident,” Clayton said on Fox Business. Secretary of State Marco Rubio said there was “zero evidence” of “nefarious actors,” Reuters reported.

Chris Krebs, a former director of CISA, the federal Cybersecurity and Infrastructure Security Agency, was just as careful on NBC’s TODAY show. “As of now, there’s no evidence to suggest that,” he said of an attack. “And there’s no evidence to suggest it’s AI either.”

So an AI attack is unproven, but not ruled out with 100% certainty. AI does seem to play an indirect role: it is finding the flaws that set off the patching rush. The final word will come from root-cause reports, and most aren’t out yet.

How fixing flaws breaks things

A patch is a software fix for a security hole. Installing one often means shutting a system down and starting it back up. Do that to thousands of linked systems at once, and some won’t come back cleanly.

The number of holes is exploding. The CVE Program, the official catalog of known security flaws, logged 36,962 new entries from July through September. That’s up 78.6% from the spring, and the program credited “AI-driven vulnerability discovery.”

Attackers are speeding up, too. Google’s threat team reported Nov. 16 that criminals now run a fully automated system that studies a patch, finds the hole it fixes and attacks it “without human intervention.” That puts pressure on defenders to patch fast.

“We’re trying to stop bad things from happening. And of course, anytime you introduce a patch, you risk destabilizing a system,” Krebs said.

Why one glitch can stop a country

Modern systems are tightly linked. When one part fails, its load shifts to the next part, which can fail, too. That’s what happened at Microsoft. Frozen servers pushed users to healthy servers, which then froze. “In less than two hours, the previously unknown defect was encountered across our fleet,” Microsoft executive Charles Lamanna wrote.

Age makes it worse. In 2024, the Government Accountability Office, Congress’s watchdog, found that more than three-quarters of the FAA’s 138 air traffic systems were unsustainable or potentially unsustainable. “They’re pushing new technology on top of really old computer systems, and it just seems to overwhelm the system,” Mullin told CNBC.

Is AI-written code to blame?

There’s no proof it caused these failures. Mullin suggested that AI systems that write code could be contributing, Axios reported, but no company has said so. “Nothing to do with writing code, nothing to do with shipping code with AI,” Liz Fong-Jones, a technical fellow at the software company Honeycomb, told Axios of the recent outages.

Researchers do see a risk. Google’s 2025 DORA report found AI use has “a negative relationship with software delivery stability,” meaning changes are more likely to cause problems. In a 2023 Stanford study, programmers using an AI assistant wrote less secure code, yet felt more confident in it.

Does the $50 billion FAA request make sense?

The Transportation Department asked Congress for $50.1 billion on Nov. 17. Only $10.1 billion is for core technology: controller tools, sturdier telecom links and traffic-flow software. The other $40 billion is split evenly among airport upgrades, terminals, five new regional control centers and up to 140 control towers.

Congress already gave the FAA $12.5 billion in July 2025. In September, the GAO found the FAA had no full cost estimate or master schedule for its rebuild. That raises “the likelihood of FAA repeating its history of cost overruns, delays, and unmet performance targets,” the GAO warned. It noted the FAA has tried to modernize since the 1980s with “limited results.”

Critics question the number. Top House Appropriations Democrats say $9.1 billion of the money Congress approved for air traffic upgrades remains unspent. “Recent events are not the result of a lack of funding or support from this Congress,” said Rep. James Clyburn of South Carolina. Duffy answered that Congress earmarked the money for specific items. “I’ve got money for radar, I’ve got money for radios. But I have no money for a new network,” he said on CNN’s “State of the Union,” Roll Call reported. Asked why his request jumped from $30 billion to $50 billion, Duffy said he wanted “a number that I think the Congress will give me,” The Washington Post reported.

The verdict so far: everyone agrees the FAA needs new equipment, and the GAO says speed is critical. But the GAO has seen no full cost estimate, and most of the $50 billion is for airports, towers and control centers, not the kind of network that failed.

Was Trump praised for his response?

Mostly, no. Asked in Manila whether he was worried, Trump said, “It’s life. When you’re big and powerful and have all the money, things like that happen — it’s fine,” according to a White House post of his remarks.

Critics included his allies. “If this is all a coincidence, it’s like the greatest coincidence in the history of the world,” said Sen. Josh Hawley, a Missouri Republican, Reuters reported. “Americans aren’t stupid,” posted Tricia McLaughlin, a former Trump Homeland Security spokeswoman, the Associated Press reported.

Praise was thin. Karoline Leavitt, a former White House press secretary now at Fox News, said “transparency is key,” then praised the White House for sending Clayton on TV to explain, AP reported. White House spokeswoman Lauren Bis said, “There is no crisis or anything like that for Americans to be worried about,” Reuters reported.

What about Sam Altman?

The OpenAI CEO hasn’t been tied to the outages. But his views on AI risk came up in October on Politico’s Decoded podcast. Asked how OpenAI differs from rival Anthropic on regulation, he said “we believe that the world should accept some bad things happening for the benefits of this technology and people having the agency.”

The bad things he named were hacks, misuse and scams. “I wouldn’t take a trade of saying like, ‘We’ll make sure there’s no major hacks, there’s no kind of, like, misuse of this technology, there’s zero scams’ … because I think people will do tremendously — orders of magnitude more — good stuff than bad stuff,” he said. He added that catastrophic risks, like losing control of AI, should not be accepted.

The remark “sparked an instant backlash from critics,” The Guardian reported. Florida Gov. Ron DeSantis said he had “no dice” with letting “a handful of tech oligarchs” make safety calls for everyone. AI critic Gary Marcus, a professor emeritus at New York University, said Altman was “saying the quiet part out loud.”

Is a week of outages normal?

No. Downdetector, a site where people report outages, logged nearly 6 million U.S. problem reports in the seven days ending Nov. 16. That was its highest weekly total in a year, Ookla analyst Luke Kehoe told Axios.

You may also have seen that the past 30 days brought only three “major outages,” versus a typical five, also per Downdetector. That count ended Nov. 16, so it missed the NSA and Apple, and Axios did not say how “major” was defined. Downdetector also tracks user reports about apps and services, so it may miss events like an FAA ground stop or a secret NSA network.

Counts also treat every outage the same, and they shouldn’t. The Uptime Institute rates outages from 1, negligible, to 5, severe. In its latest survey, only about 10% of data center operators’ worst outages were serious or severe. A nationwide ground stop that strands hundreds of thousands of travelers outweighs a handful of short app glitches.

Over the long run, outages per site are falling, Uptime found, though progress is slowing. And clusters can fool us. “Humans see patterns in everything, regardless of whether they’re there, and generally interpret coincidences as conspiracies,” security expert Bruce Schneier told Axios.

Could a bank outage or attack wipe out your money?

A delay is not a loss. During its outage, Chase reassured customers that they still had access to their accounts and funds, NBC’s TODAY reported.

If a bank fails, the FDIC insures deposits up to $250,000 per depositor, per bank, for each ownership category. “Since 1934, no depositor has lost a penny of their FDIC-insured funds,” the agency says. But FDIC insurance does not cover theft or fraud, or stocks, bonds, mutual funds and crypto.

Theft is covered by other rules. If your debit card is lost or stolen, report it within two business days, and your loss is capped at $50, the Consumer Financial Protection Bureau says. Wait longer and it can reach $500. Report any unauthorized withdrawal on your statement within 60 days.

Bank tech meltdowns have happened before. In 2018, a botched computer upgrade at Britain’s TSB disrupted online, phone and branch banking. British regulators fined the bank £48.65 million, and it paid customers £32.7 million in compensation.

Could hackers set off weapons or cause crashes?

Officials have reported nothing like that from the outages so far. The FAA’s ground stop was a safety step: when equipment “disrupted aircraft communications and radar feeds,” the FAA held departures, the Transportation Department said.

Officials do treat these as real risks. In 2018, the GAO found that Pentagon testers posing as attackers “were able to take control of systems relatively easily” in weapons still being developed. The GAO said the Pentagon had started updating its policies and doing more testing.

Cars and ships get scrutiny, too. In 2015, Fiat Chrysler recalled about 1.4 million vehicles after researchers remotely reached a Jeep’s controls through its radio, its recall filing shows. A software update fixed it. In 2025, the Coast Guard issued its first cybersecurity rules for U.S.-flagged ships and port facilities.

Are outages a weapon other countries could use?

U.S. agencies say rivals are preparing for exactly that. CISA, the NSA and the FBI say hackers linked to China’s government, including a group called Volt Typhoon, are positioning themselves inside U.S. networks. That “allows them to disrupt critical functions at a time of their choosing,” CISA says.

It has happened elsewhere. In 2015, hackers cut power to about 225,000 customers in Ukraine, CISA reported. In 2018, the White House blamed Russia’s military for the 2017 NotPetya attack, which it called “the most destructive and costly cyber-attack in history.”

That is why leaders stay alert even when an outage is an accident. “It’s fair to say there are a lot of bad actors out there who would love to do damage to the United States,” Senate Majority Leader John Thune said, The Sydney Morning Herald reported.

If AI caused it, would we be told the truth?

Not always right away. Public companies must report a “material” cybersecurity incident to the Securities and Exchange Commission within four business days after deciding it matters to investors. The U.S. attorney general can delay that for national security. A 2022 law will require critical infrastructure companies to report major cyber incidents to CISA within 72 hours, but CISA hasn’t finished that rule.

Those rules cover attacks, not honest mistakes. None of them requires a company to say whether AI wrote faulty code.

First explanations can also change. In 2023, Microsoft said Chinese hackers had likely stolen a key from a “crash dump,” a file saved when software fails. A federal review board later found that Microsoft realized soon after that the claim was inaccurate, but did not publicly correct it for more than six months.

This time, Microsoft says a final review is coming, and FAA and FBI teams are examining the router, NBC’s TODAY reported. Those reports will be the real test.

What’s at risk beyond email

Your health. When ransomware hits a hospital, deaths among patients already admitted rise 34% to 38%, University of Minnesota researchers found. About half of Americans take a prescription medicine every day, and an emergency can make refills hard, according to Ready.gov, the federal preparedness site.

Reaching 911. On Jan. 14, a Verizon “software issue” cut wireless service for about 10 hours. New York City emergency managers warned that some customers might not reach 911, BleepingComputer reported.

Money and goods. Late transfers can mean late rent. Grounded flights carried “critical medication and supplies, and our nation’s goods,” said Rep. Rosa DeLauro of Connecticut, the top Democrat on the House Appropriations Committee.

Trust and scams. Some public figures claimed a cover-up without evidence, AP reported. Criminals also exploit confusion. During the 2024 CrowdStrike outage, CISA warned of phishing and urged people to “only follow instructions from legitimate sources.”

Should you stock up on food?

No need to hoard. Vice President JD Vance said friends had asked whether to build a bunker or stock up on food and water. “My answer to that is no, we do not see some crisis coming down but you have to be prepared for everything,” he said on Fox News, Reuters reported.

Federal guidance asks for a modest kit, the same one you’d want for an earthquake. Ready.gov suggests at least a several-day supply of non-perishable food and one gallon of water per person per day for several days. It also recommends:

  • Some cash in small bills, since ATMs and card readers may not work.
  • A supply of your prescriptions, plus a plan for medicines that need cold storage and devices that need power.
  • Copies of key documents and account numbers, stored safely or saved electronically.
  • A backup way to reach people. The FCC says to text when calls fail, wait 10 seconds before redialing, and keep a charged backup battery.
  • Caution with “fix-it” messages. Follow only official sources, CISA says, and keep updating your devices and using multifactor authentication.

In Santa Clara County, sign up for AlertSCC, the county’s free emergency alert system. PG&E customers who rely on powered medical devices can apply for its Medical Baseline program for extra warnings before planned power shutoffs.

Is this the new normal?

For a while, probably. “I’m confident that we will get through this. But look, the patching is not done. The vulnerability discovery is not done. The next 12 to 18 months are going to be a bumpy ride. I think everybody needs to be prepared for that,” Krebs said.

Homeland Security’s chief agrees. “I believe we’re going to see more of this as new technology rolls out,” Mullin said.

Fixes are coming, too. Krebs said teams are “going to get better every time,” and CISA now tells federal agencies to fix the most dangerous flaws first and spread out the rest.

What this means

The outage week was a warning, not an invasion, so far as officials can tell. Most of it looks like a pileup of accidents, though Optimizely shows real attacks still happen. Old, tightly linked systems are being patched faster than ever, partly because AI keeps finding holes. Some fixes break things, and one failure can spread fast.

Expect more bumps, not collapse. Insured deposits are protected, outages pass, and a few days of supplies cover most of what could go wrong. The harder questions belong to Washington and Silicon Valley: how to fix holes without breaking things, and how to keep the next accident from becoming an attack.

Source: Microsoft root-cause post, Nov. 15, 2026; NSA statement, Nov. 17, 2026; White House Rapid Response, Nov. 17, 2026; U.S. DOT $50.1 billion request, Nov. 17, 2026; GAO-26-107992; FDIC; CFPB; CISA on Volt Typhoon; CVE Program Q3 2026; Uptime Institute 2026; Ready.gov; NBC’s TODAY, Krebs interview; Politico’s Decoded, Altman interview; Reuters; AP via ABC News. Full source list below.

Drafted by AI from the linked public source. Approved by a Daily News Editor.

Sources and documents