


The episode explains why a widespread Microsoft outage matters so much: Microsoft underpins everyday work and public services, so failures cascade across email, meetings, and other cloud services. It argues the cause is likely internal—possibly a flawed software update—leading to server-side cascading failures (a “thundering herd” problem) rather than a cyberattack. It highlights Microsoft’s central role in cloud and SaaS, driven by convenience, frequent updates, and security benefits, but also creating single points of failure.
Roland Manthorpe, Sky News technology correspondent, who reports that Outlook servers failed and the issue spread to Teams and Copilot; he discusses “normal accidents” (Charles Perrow) and the need for redundancy beyond just switching vendors.
schools, hospitals, GP surgeries, airlines, and UK government email reliance; also air traffic control outages and the NHS fax/page redundancy.
Written by AI. May contain mistakes.
Discussion on the widespread impact of Microsoft service failures.
“You know the story, you remember the headlines, but what if you could hear from the people who were actually there?”
Exploring how deeply integrated Microsoft is in various sectors.
“Roland, just explain exactly what has stopped working this morning and what we know about why.”
Insight into Microsoft's cloud services and their operational issues.
“if you're not getting emails and can't get online.”
The risks associated with digital centralization in technology.
“Microsoft manages to resolve soon enough.”
Discussion on the implications of failures in large tech companies like Microsoft.
“How likely do you think another pandemic is?”
Explore the global ramifications if Microsoft were to fail.
“well, all these, you know, kind of, in inverted commas, cybersecurity experts and IT specialists telling us that this is the best way to go.”
Learn about the concept of normal accidents in complex systems.
“He studied the Three Mile Island nuclear accident extensively.”
Consider the necessity for alternatives and fail-safes for Microsoft services.
“because these systems are so unbelievably, unfathomably complicated.”
Discuss the broader implications of digital service outages.
“Yeah, I mean, in a way, I don't think we should blame ourselves, though.”
Explore strategies for hospitals and organizations to mitigate risks.
“that for various kind of cascading reasons causes a gigantic traffic jam.”
Analyze how people tend to forget about outages once resolved.
“I mean, this is part of the challenge here, isn't it?”

Get the gist, explore an idea or ask about something said.
On This Is Why, what are the key points of this episode?
Automatic transcript. May contain errors.
0:02Sky News, the full story first.
0:11There's only so much turning it off and on again you can do when the problem is Microsoft itself. And this is why.
0:21Whether you're building a roster or a business, great talent can come from anywhere. Pebble helps companies hire and support employees across 185-plus countries. Pebble believes you don't build a winning team by limiting where you look for talent. That belief is at the heart of their new partnership with Bay FC and the way Pebble helps companies build their own teams every day. With Pebble, companies can hire in over 185 countries, bringing everything together in a single platform that integrates with the tools teams already use. The result is a global team that's actually set up to succeed, with accurate pay, real benefits, and the support they need to thrive wherever they are.
0:57Bottom line, great teams can come from anywhere. Pebble helps companies build them. Special offer for podcast listeners. Pebble is normally$399 a month per employee. Already a no-brainer for what you get. But right now, there's a limited-time offer on their site that makes it even easier to get started. Go to hellopebble.com slash sports before it's gone. That's hellopebl.com slash sports. Terms and conditions apply. How likely do you think another pandemic is? It's right at the top of the agenda. It will happen. It will happen. You know the story, you remember the headlines, but what if you could hear from the people who were actually there?
1:31I saw a man who'd been hit in the face with two bullets, put my camera to my eye and he looked at me with all the pain in the world in those eyes. He was trying to move his head to say no, and I didn't take that picture. I'm Cathy Newman and I'm going back inside the stories that changed everything with the people at the very heart of them. Search Backstory and follow wherever you get your podcasts.
2:18The frustrating thing is that you can have a perfectly good computer, a working internet connection, and still be unable to get on with your job. Because Microsoft doesn't just make words. It runs our emails, our meetings, our diaries, and powers the systems that other organisations depend upon. Businesses around the world are having a pretty lazy Friday. Schools, hospitals, GPs, airlines. Frankly, it would be easier to list what's not been affected. We have built an extraordinary amount of everyday life around one company's technology. When it stops working, we discover quite how difficult it is to do without.
2:56Roland Manthorpe, this guy's technology correspondent. Roland, just explain exactly what has stopped working this morning and what we know about why. Well, the first bit is definitely easier than the second bit, because loads of Microsoft products have stopped working, in particular Outlook. The issue seems to be around the servers that power Microsoft Outlook. But the contagion, I guess if you can call it that, has spread to other Microsoft services, Teams and Copilot in particular. So we know that. The why is harder to establish. I'll tell you what it probably isn't according to Microsoft. Almost certainly they say it's not a cyber attack.
3:37They say there's no evidence of malicious behavior. So therefore, the problem is inside Microsoft. And they haven't said exactly what it is. From reading between the lines and based on past experience of this sort of thing, it seems like a software update that has gone awry, that has introduced some kind of bug into the system, which has caused a kind of cascading failure, which then brings down server after server. So who exactly has been affected? I mean, quite clearly Sky has. I've had to decamp to our old radio studio to try and speak to you today to get to the bottom of this, but if it's Microsoft products, I mean, that's pretty much everyone, isn't it?
4:14It's not everyone, because not everyone uses Microsoft, but a lot of businesses and institutions do. So it's a real problem for work, for anyone trying to get work done on a Friday. I'm sure actually some people will be quite glad of this in some ways. It makes life a lot easier if you're not getting emails and can't get online. But yeah, I mean, the scale of Microsoft's influence in particular in work is enormous. I think the last time I remember seeing some of their stats, they had 450 million paid seats, as in, you know, they're the people paying for these services. And that's a pretty good indication of just the scale of what they're doing.
4:52And then we've heard reports, you know, we've been kind of following this on Sky News. We've heard reports from schools, from hospitals, lots of just people saying they can't manage to do their ordinary work. government in particular runs um central government runs a lot of that runs on microsoft outlook i'm sure they're having issues because yeah i mean like like you were saying there's not much you can really do if your main email communication system goes down but just in terms of of the effects of this where are we seeing it most keenly do you think because of course if we're talking about a kind of an internet slowdown or a break in major pieces of software you know immediately my mind goes to, I mean, what could be the worst possible outcome of this?
5:33I mean, plenty of people are going to be missing hospital appointments, but hopefully those can be rescheduled. Is there anything more dangerous or perhaps more insidious that we need to be thinking about? In a way, I think this actually kind of relates to the whole question that we're looking at, because it all depends on your capacity to root around and do other things and your resilience as a whole. You don't know what organisations have. So, I mean, you know, Sky runs on Microsoft products, but, But, you know, we're right now, we've found a system, we've made it work. But it's quite easy to imagine that you've got various things connected to Outlook or to Teams.
6:09And without that, the system as a whole goes down. But I think in general is that because ultimately we do still have internet access, people can find systems and find ways to get around things. Of course, then, you know, you get into different questions, like questions of data protection. To what extent is it acceptable to pour information over to your personal email address, to send an email if you're a head teacher or something like that? There are lots of naughty problems that I'm sure everyone is wrestling with at the moment. Yeah, I mean, right now, as we are talking, airlines are having problems.
6:46GP surgeries are having problems. Businesses, small, medium and large, will be having any number of problems. But there might be a few kind of smug people out there saying, look, I've got an iPhone, I use a Mac, I use Gmail, I don't touch that pesky Microsoft stuff. But presumably, even if they don't touch it, all of the institutions that they're trying to get in contact with today probably do. Yeah, yeah, exactly. And while there might be some people smugly saying, well, I use Gmail, there aren't that many people who are on something else, right? I mean, say when it comes to email, there tend to be two options.
7:22When it comes to cloud, there's really three options. You know, I can't remember the exact statistics, but I think 60 % of cloud is on Microsoft, Amazon or Google. and on it goes. The internet was originally a very distributed system, but these days it's extremely centralised. In some ways for good reasons, but we're seeing one of the costs of that now. Just tell me a little bit more about that, because I think plenty of people listening and watching to this will understand Word. They will understand Outlook and email. They'll even understand Excel to an extent in terms of spreadsheets. But it's the stuff that Microsoft does behind the scenes, the providing the power to these companies that I think is going to be one of the major issues around this.
8:06Just explain exactly what it is they do in that sector. Do you mean what is the cloud and what is Microsoft? What is the cloud? Okay, yeah, well, it's the most misleading name that has ever existed because it's not a cloud. It's a bunch of computers sitting in warehouses. I mean, enormous warehouses and many, many, many thousands of computers. The cloud is data centers, And data centers are really just other people's computers that information is stored on. Microsoft runs one of the biggest, Microsoft Azure, and they have a large chunk of the market. And this is actually relevant to their problems right now, because as well as letting other people use their computers to do things, they're using them to power their own services.
8:50And it's clearly a server issue that has gone wrong here. Obviously, we don't know exactly what, but one thing that seems to be emerging is that it might actually be their own automated error systems that are causing the problem. Think of like a 10 lane motorway. So the first lane gets closed. And so the traffic starts to kind of go into the other lanes. And if you imagine a kind of control center seeing this and say, oh, hang on, there's too much traffic. Well, we better close down lane number nine. The traffic pours into lane number eight. The control centre says, oh, hang on, we better close lane number eight and so on and so on and so on until all the lanes are closed.
9:28And you get a massive traffic jam. It's that feedback loop with the automated system that checks on these things that is in a way responsible for the problem. But then, of course, just like in the traffic analogy, if you think, well, what happens then? Well, you have an unbelievable traffic jam piling up, waiting to pour in the moment the lanes are reopened. And this is the problem that Microsoft is facing now. There's just this enormous, in computing, it's called a thundering herd problem. So there's this thundering herd, this is what they call it. There's a thundering herd of traffic just waiting to pile back onto the servers.
10:03And every time Microsoft try and stand up these servers, well, the traffic just floods on and they go down again. And I think, yeah, so this seems to be the problem that they're experiencing. And it's with their own servers in these data centers in what is misleadingly called the cloud. Okay, so that is the problem that everyone is keeping fingers crossed. Microsoft manages to resolve soon enough. But the problem that the rest of us are experiencing is, well, in essence, that we've put all our eggs in one basket. How did we get to this situation, Roland, where Microsoft products are so ubiquitous in offices around the country that so much of e-business is run on applications, software and cloud storage provided by one firm?
10:45Oh man, well, therein lies a tale. I think... Bill Gates plays a part, doesn't he? Oh yeah, yes, of course, although I think we can sort of go... I mean, because Bill Gates, the founder of Microsoft, but really what we're talking about is a different generation of technology here. We're talking about the SaaS generation, the software as a service generation, where people, and this was all enabled by the cloud, because when you had the cloud, you could rent software that wasn't being run and you wouldn't have to do everything internally. And this has brought tremendous advantages in terms of productivity, efficiency, and access to top-level services.
11:28Because, I mean, if you're trying to build things yourself, well, you know, what are your chances if you own a small business or even a big business of building something that is as good as Microsoft's? So suddenly we can all get access to the very best software and it can be updated. One of the advantages of using the cloud is that they can send updates to every single piece of software all around the world pretty much at once. It's totally incredible. But you can sort of see from this that there's a centralizing force here, right? So instead of everyone doing it their own way, you tend to go to a few people because those people offer reliability, security.
12:07Security is a huge aspect of it. These days, if you try and do your own thing, it is extremely, extremely difficult. You are extremely vulnerable to cyber attacks. And we've seen the other side of this recently, say in this thing with PHP, we talked about that before, how there was a vulnerability in PHP, this programming language. But if you're running your own website and you are in control of that, that means you're having to do this quite complicated technical patch yourself. and having to keep on top of it. If you're in Microsoft, for instance, well, they've handled that all for you and it's completely taken care of.
12:49And they're fighting off these cyber attacks on your behalf. The downside of it is, is that we are building in centralisation, which makes our digital world as a whole more vulnerable in some ways.
13:07How likely do you think another pandemic is? It's right at the top of the agenda. It will happen. It will happen. You know the story, you remember the headlines, but what if you could hear from the people who were actually there? I saw a man who'd been hit in the face with two bullets, put my camera to my eye and he looked at me with all the pain in the world in those eyes. He was trying to move his head to say no, and I didn't take that picture. I'm Cathy Newman, and I'm going back inside the stories that changed everything with the people at the very heart of them. Search Backstory and follow wherever you get your podcasts.
13:46I still don't quite understand how a company like Microsoft, which presumably has so much backup infrastructure, or it's coming out of their ears, can still have a failure across several services at once. And that if that sort of cataclysmic failure is possible, well, all these, you know, kind of, in inverted commas, cybersecurity experts and IT specialists telling us that this is the best way to go. I mean, it just feels today, if Microsoft falls over, half the world at least gets a skint knee. There are a couple of things there. How can this happen to Microsoft? It seems kind of insane, doesn't it?
14:23the way that and and without knowing all the details the the way that i understand it is that they are operating on such an enormous scale that a lot of this is done with automated checks it has to be there is simply no other way to operate it and when these automated checks start to interact in a in unhelpful way then you can get these cascading failures there's a sociologist charles Pero talks about this term, he calls it normal accidents. He studied the Three Mile Island nuclear accident extensively. That was an accident, a nuclear accident in 1979, which to this day is still the worst commercial nuclear power plant accident in US history.
15:06What he found in that was that often the cause of accidents is tightly coupled systems. When systems are too tightly coupled so that something going wrong here immediately impacts something over here. And say in a nuclear power plant, there might be valves, but in a data center, it might be a series of automated checks that are triggered and set off this cascading effect. Obviously, Microsoft are aware of this problem, but it's very hard to prevent completely, given the scale at which they're operating. because these systems are so unbelievably, unfathomably complicated. You simply cannot predict the outcomes of everything.
15:50So this is why Charles Perra called it that you get into a situation where accidents become normal. Do we now need to consider, you know, do institutions, you know, the public sector, the private sector, do they now need to consider having a backup to Microsoft, having a fail-safe plan for events such as this? You know, an event which I'm sure plenty of people would never have imagined ever taking place. It's happened now. We perhaps need to guard against it again. Yeah, well, I mean, it's not just Microsoft. It's the internet in general. It's other services. We've seen that this is a really important part of just being able to keep your services up.
16:29You know, in the UK, air traffic control reliably kind of goes down for some internal reason about once a year. And every time it causes these gigantic queues, that's a good example of a system that doesn't seem to have managed to find a totally resilient plan and a totally resilient alternative to whatever their existing software is. So is this our fault then? Is this our fault? Have we all just been daft, been seduced by the cult of Microsoft? It's great software, it's really easy to use, we're all familiar with it, blah, blah, blah. But in doing so, we've set ourselves up for precisely a moment like this, haven't we?
17:08Yeah, I mean, in a way, I don't think we should blame ourselves, though. There's, you know, using these services, they're fantastically convenient, they're useful. And let's be honest, they're unbelievably reliable. I mean, the real miracle in a way is that these incredibly complicated services stay up as much as they do. because it's simply mind-boggling the complexity involved. I'm sorry, just to jump in there, because I'm going to challenge you slightly because in my head, the analogy that you're giving me there is, oh, look, it's a road closure, right? We use the roads every day and eventually sometimes there's going to be a road closure and we just have to deal with the effects of it.
17:47Isn't this rather more akin to you're driving across a bridge and the bridge collapses rather than it being a minor inconvenience? This is a huge problem. This is a huge amount of financial damage to companies. There are going to be people who suffer because they've missed out on hospital appointments today and all the rest of it. Isn't it of a grander scale, given how ubiquitous this Microsoft stuff is? I think maybe we're both right here in that, I mean, you're absolutely right. The scale is the problem. Ultimately, the way that I see it is that, yes, it's much more, it's more akin to an accident on the motorway that for various kind of cascading reasons causes a gigantic traffic jam.
18:28The thing is, is that whereas that on a motorway network, it can't spread and spread and spread and it will ultimately remain localized, on our current digital systems, the spread ends up being enormous and millions and millions of people, hundreds of millions of people end up being affected. I guess I feel like is that we have arrived at this place for a reason and it will be extremely hard to unwind from here. But I do think that in general, we need to think really carefully about having more redundancy in our systems without ending up piling on so many costs to people that it becomes, that we end up making it even harder, even harder to do business if you're a small business.
19:14This I think is the, or if you're a school or, you know, something like that. I mean, this is part of the challenge here, isn't it? So if you were, all right, let's just role play this for a second, Roland. I mean, if you were running a hospital trust right now, what would you want in place before the next outage? What a good question. I'm not sure that I would necessarily be thinking that the answer is an alternative digital system. I think you need to be thinking about what happens if you can't use digital systems at all. That seems to me to be really important. so I would be looking at some kind of well very much like the government has been looking at recently with with its its kind of you know plans in in case of the the internet going out what would you use and it's quite funny isn't it because we've many times over the years I've heard politicians say how outrageous it is that the NHS is still using fax machines and and pages in this day and age but actually you know the page the page system is a really good example of a system that survives for a reason.
20:13And because it has some kind of convenience that is actually really, really useful for clinicians. But it also does give you that redundancy that you need. So because to me, this isn't really a question of it's not just like, okay, we need to protect ourselves against Microsoft in particular. I'm not sure that running, you know, Google Cloud in parallel would be worth the cost when actually the real worry has to be not being able to get any of these services at all through a ransomware attack or, you know, who knows, some kind of like AI disaster that brings down a chunk of the internet for a while.
20:51Look, Roland, given that we're doing everything a little bit differently today, we're going to conclude this interview in a different fashion where I'm going to tell you what I think is what is going to happen next and you are going to tell me why I am wrong. Once those emails start flowing again, everyone is going to forget about this and cross their fingers that it doesn't happen again. That's it. I think that you're half right, because my experience of this is that people don't even cross their fingers. Everyone completely memory wipes what has happened and just assumes it will never happen again.
21:22Maybe, I mean, you hope that CISOs and CTOs in hospitals or large organisations are thinking about it. That's their job. But for most people, I mean, look, But how often have you thought of the Google cloud outage, which is about three weeks ago since then? Yep, me neither. Sorry. I haven't thought about it at all. I have carried on using those products. You're a technology correspondent, though. You should be thinking about it. I know. That's why I've got so much empathy for all the people. All the people just memory wiping it. It's amazing, isn't it? But I think that is because ultimately, almost always, these things are actually resolved relatively quickly.
21:59they go away and then you get another year of good service and it's it's like it's just there for you isn't it i think it's surprisingly hard to really reckon with what if this actually went away roland always pleasure to talk to you thanks very much mate and that's your lot for today on this is why it is usually at this point in the episode that i ask you to email in your thoughts why at sky.uk is the email have a go who knows if it will work have a lovely weekend we'll see you the other side.
22:58that changed everything with the people at the very heart of them. Search Backstory and follow wherever you get your podcasts.
When Microsoft goes down, we are given a stark reminder of how much life is dependent on a handful of tech giants.
But have we handed too much power and trust to too few companies? And what happens when the systems we rely on suddenly fail?
In this episode of This is Why, Niall Paterson speaks to Sky’s Technology Correspondent, Rowland Manthorpe, to answer this question and consider whether it’s time to rethink our relationship with big tech in the future.
Have you got a question for Niall? Email us at [email protected]