Welcome to the monopoly round-up. Lots of news, as usual. Nvidia CEO Jensen Huang’s downfall is the most important story of the week, by far, and we’re going to have a piece on that soon, taking a step back from the gossipy parlour intrigue and reframing this event in terms of the political and policy debates it is causing.

There was a lot of other important news this week, too, including a case to block a veterinary services merger that would affect every pet lover in the country, and some consequential leadership battles that could affect whether the new Congress tackles corporate power, as Senator Chris Van Hollen launches a bid to challenge Senator Schumer’s leadership.

But first, I want to discuss two major outages of key infrastructure that happened this week, the Microsoft Outlook collapse that caused a nationwide work stoppage on Thursday, followed the next day by the nationwide ground stop that prevented some 12,000 flights from taking off during a four hour period on Friday. As of this writing, thousands of flights are still being canceled, and stranded passengers are struggling to make it to their destinations.

When Will We Stop Tolerating the Inevitability of Failure?

The Microsoft and air traffic control outages hit like a one-two punch last week. And their scale was significant, because each of these systems are infrastructure for their respective clients and industries. Microsoft Outlook and the company’s suite of office products are staples of modern work, the technological infrastructure of government agencies, schools, for-profit and nonprofit corporations, the largest law firms in the country, and more. As of November 2026, there are about 450 million paid commercial subscribers to Office 365, and another 95 million paid consumer licenses.

A significant portion of those subscribers, over a half billion of them in total, saw their functions ground to a halt. One estimate put the cost of the outage, which lasted from 21-24 hours for various services, at $7 billion.

In this case, it was the failure of a single component of an automated health check that shut down one server, which redistributed that server’s traffic to other servers, which caused those servers to become overloaded and shut down, and so on and so on. Tech outages are not new or even rare, but the scale of interconnectedness and dependency means that the extent of harm caused by one missing puzzle piece is immense. It is a bit like we are in a giant bank run, except with industrial and technical systems.

Around 10 hours into the Microsoft outage, which became widespread around midday eastern time on Thursday, more than one group text I’m on had begun discussing what it would mean if Outlook just… never came back? That thought exercise is less about Outlook, our lives and work (inevitably) would continue, but about the possibility that every other system we rely on could be next.

Those discussions had hardly subsided when news broke that Federal Aviation Administration issued its ominous Friday the 13th order:

DUE TO EXTRAORDINARY CIRCUMSTANCES AND FOR REASONS OF SAFETY.
ATTENTION ALL AIRCRAFT OPERATORS, BY ORDER OF THE FEDERAL AVIATION
COMMAND CENTER ALL AIRPORTS ARE GROUND STOPPED.

For air traffic control, it was a failed router following a copper-to-fiber migration that triggered the flooding of other routers, imperiling radar and telecommunications systems. (Again, these are the current diagnoses, which may change as more information comes out.)

The Microsoft and air traffic control outages happened back to back and affected tens of millions of people across the country, so naturally there was speculation that the incidents were somehow related, that the United States had been the target of a ransomware or state-sponsored cyber attack, or that we were witnessing the early onset of broader societal disruption. The verdict may still be out on some of that.

Cascading Failures

As a technical matter, both the Microsoft Outlook outage and the FAA’s nationwide ground stop were attributed to “cascading failures” following technological upgrades. I like that term because it captures the domino-like phenomenon where a single glitch in the system can quickly spiral into an incapacitating social and economic crisis.

Cascading failures are what occur when systems are based on flawed and deeply interdependent technologies. The quintessential example is Chernobyl, where flawed systems and a culture of inadequate safety protocols created a hair trigger for one error to cause a disaster.

In his 1984 book “Normal Accidents: Living With High-Risk Technologies,” Charles Perrow introduced the idea that certain systems are so complex and so tightly interwoven that accidents are not merely possible, but inevitable. “Tightly-coupled” system design was how we got Three Mile Island, and even in 1984, Perrow saw how we were replicating high risk system design in the airline, maritime and space industries, with weaponry and medicine, and dams and mining. The obvious response is to create more resilience in our industrial infrastructure, to lessen the likelihood of such incidents and reduce the damage when they occur.

Despite Perrow’s diagnosis over 40 years ago, similar “interactive complexities” have become even more prevalent in the networked era of the last quarter century. It used to be that when you were having a tech problem, work would call in the IT guy to resolve some local server issue. Now we’ve constructed bespoke systems on top of remote servers and cloud-based infrastructure that are, predictably, concentrated choke points for systemic failure. Most businesses now outsource control over their own systems to large firms.

Indeed, our political economy architecture over the past forty five years encourages this kind of irresponsibility. First, there is a lack of liability for firms like Microsoft when they cause damage. Second, labor law doesn’t protect the engineers who know how to protect these systems. One clear root cause of cascading failures — the first domino of system collapse, which predates either of these incidents in isolation — is getting rid of workers who know best how to monitor for and mitigate risks.

Microsoft has laid off approximately 35,000 employees since October 2022, and in May 2025 laid off 3% of its entire workforce, disproportionately software engineers, while beefing up project management. Basically, the trend is toward replacing engineers, who understand the importance of reinforcing weaknesses in the foundation, with a bunch of well-paid efficiency hawks with a bias toward cutting corners on the path to rapid product delivery.

Meanwhile, Microsoft CEO Satya Nadella says that up to 30% of the company’s code is now written by software. That percentage is much higher at other companies, including Anthropic. Tech executives have recognized the risk associated with smaller teams having to scrutinize millions of lines of code they didn’t write, and products being shipped by managers who haven’t had time to review at all.

Another cause is weak antitrust law, which facilitates the creation of chokepoints. As BIG asked years ago in explaining the shortages during the post-Covid era, “What else is a monopoly but a business model designed to secure or create a bottleneck?”

It wasn’t necessarily done in bad faith; policymakers followed the lead of economists, who believed dominant firms were dominant because they were efficient. This faith in efficiency over all else meant that the public structuring of markets to force resiliency - aka regulation - was illegitimate. Alas, this philosophy has led to a series of bottlenecks in our supply chains, which are now global.

I wish we had learned from the post-Covid era, but it’s clear that we have still not re-architected key social infrastructure to become more resilient. Since 2021, it has gotten worse, with AI and cloud computing facilitating more centralization. Here’s investor Roger McNamee making that point:

Can Companies Manage Their Own Systems?

Five hours into its outage, Microsoft issued an optimistic update: “We're returning a portion of affected servers to the service, and they remain stable so far,” but followed it one hour later with, “The servers we returned to service can become unresponsive again once they receive traffic.”

These opaque updates and subsequent partisan finger-pointing suggested no one really knew what was going on. Microsoft itself explained to investors the nature of the problem. Highly interdependent systems magnify reliability issues, it said, while admitting that they are rolling out technology without adequate processes to detect defects:

According to Microsoft’s July 2026 SEC 10-K filing, interdependencies magnify reliability issues, and pre-release defects may be undetectable.

There will be studies, the obligatory reports and investigations, but even if we arrive at some official technological consensus, the overriding perception is that the most important and dominant corporations in our society are deploying critical infrastructure without really knowing how it works.

Here’s another example, from Anthropic’s pre-IPO filing, which plainly states their models are developing at such a rate they may soon become incapable of monitoring for harmful conduct and misalignment:

In Anthopic’s pre-IPO S-1, they admit that “we may deploy models whose behavior we cannot adequately verify.” Is this okay?

When you bet the entire house on technology that even its makers admit is unreliable, unsafe, and impervious to risk mitigation, and then something goes wrong, an outside observer might fairly ask, “What did you expect?”

But the problem isn’t isolated to private large corporations. We have ripped out the capacity of public institutions as well. Earlier this year, the FAA reduced its target number of air traffic controllers by 2,000 personnel, even though since 2015 the amount of air traffic has increased by 10% and the number of controllers has declined by 6%. Many controllers now work six day weeks with compulsory overtime. Nearly 11,000 air traffic controllers are still due months of back pay for a 2.8% raise granted by Congress, which the FAA refuses to pay.

Air traffic controllers are understaffed, overworked, and underpaid. We know this, yet are moving in the direction of making that problem worse. That’s the broader context in which the FAA announced, just three days prior to the outage, the nationwide rollout of its AI-assisted SMART system — despite concerns from Congress as to its safety and over the objections of air traffic control unions. I’m sure the timing is just a coincidence, nothing to see here, move right along, right? Or maybe, just maybe, it’s worth asking once again, What did you expect?

Broad systemic failures are avoidable, yet we’ve made them inevitable. It’s hard to imagine that changing any time soon, because we can’t expect to govern when we have given up on understanding how things work. And while this time it’s “just” Outlook, or “just” an annoyance isolated to people with the misfortune of having to travel this weekend, sometime soon it will be a much larger problem. The parallels to Chernobyl aren’t just theoretical. Nuclear power plants use Outlook, too.

There’s also just the Microsoft of it all. Microsoft is one of the enduring monopolies of the past 50 years, ever since it acquired the operating system called DOS in 1981 and wielded it to achieve monopoly control over personal computers. There’s a colloquial wisdom that Microsoft’s power has waned amid competition with the likes of Google and Apple, but it’s more accurate to say that Microsoft has burrowed itself into core markets — like enterprise communications software — through sticky procurement contracts that produce moral hazards through vendor lock-in.

If Microsoft’s refusal to improve the basic search indexing function of Outlook is a reflection of its insulation from competitive pressure, which it clearly is, then this week’s outage is even less surprising as a function of the lack of competitive pressure it faces to make its systems resilient.

The larger question here is whether policymakers have over-indexed on optimizing the efficiency of individual systems, which increases the vulnerability of larger, interdependent systems. That’s the way to think about cascading failures. It’s a critical concern not just for work emails and aviation, but for financial markets, telecommunications, energy, healthcare, and cloud computing.

One big lesson of the pandemic was single points of failure in supply chains were dangerous. They resulted in nationwide shortages of baby formula, medical contrast dye used in CT scans, and semiconductors used in automobile manufacturing. A society that prioritizes efficiency and scale, at the expense of resiliency and redundancy, is doomed to the inevitable stream of shortages, outages, and eventually even greater system collapse. In other words, we are going to restructure our corporate systems, it’s just a question of whether we do it deliberately, hiring the necessary people to manage the risk, or whether we do it after a collapse has harmed a lot more people.

Or to put it differently, we’ll know America is on the right path when Microsoft makes Outlook search actually work.

Now, on to the Good and Bad news of the week. Democrats are organizing around an agenda for the new Congress, one which centers economic populism, but not if leadership can’t create the pathways to govern. A shakeup is coming for the FTC, the Fox acquisition of Roku appears tainted by corruption, Chuck Schumer is facing a real challenge for Senate leadership, and much more below.

Good News

Bad News

That’s all I’ve got. Let me know what’s on your mind!