a16z Podcast - a16z Podcast: The $200 PC in the Enterprise

Episode Date: December 19, 2017

What capabilities do enterprise companies really want from their computers? Twenty years ago, those capabilities might've been bundled into a mainframe. Ten years ago, it might've been the PC. Today, as more and more businesses rely on devices that need only browsers/ internet connectivity, what will the "$200 box" sitting at an employee workstation look like? In this hallway-style episode of the a16z Podcast, Benedict Evans and Steven Sinofsky discuss how tech devices evolve for the enterprise -- and more broadly, what happens when the S-curve levels out??

Hi, and welcome to the A16Z podcast. We're here today with Benedict Evans and Stevensonovsky in another one of our hallway-style conversations. In this episode, they discuss what happens when the S-curve levels out, and especially as more and more businesses are relying on devices that need only browsers and internet connectivity, what will the $200 bucks sitting at an employee's workstation look like? Or, more broadly, how do tech devices evolve for the enterprise, from the mainframe to the PC, to the tablet, and the smartphone? So I'm Benedict Evans. I'm here with Steven Sinovsky. And we've been talking a fair bit up around the kind of fundamental platform shift that's going on in tech from PC, Wintel, mouse, Windows-based

computers to mobile operating systems. And we thought it was kind of interesting to talk about what happens to the PC in that environment. And maybe the kind of a good way to set up that conversation is to go back and think about what happened to the mainframe. Because in a previous generation, mainframes and mini computers and workstations lost their position at the center of the tech industry to the PC and to the client server and so on. But that didn't mean that the mainframe business just kind of evaporated the day after Windows 95 shipped. Instead, actually IBM's mainframe business grew really healthily throughout the 2000s. And so it's interesting now as we think, well, it's one thing to say that the center of gravity and tech

moves to mobile and moves to mobile operating systems and all these use cases will slowly move. But what happens to the other half of that story? What is the 5, 10, 20 year half life of the PC look like. Yeah, it's super interesting. I like to put some color on that because I think that a lot of people don't really realize this. But back in like 1993, 94, IBM was on the brink of bankruptcy. And the theory was that PCs were just ending IBM's reign. And disruption theory was still, you know, years away from being published in a book and an HBO article. But the idea was like, oh, they're just going to evaporate. Like, and we think about disruption and we think about the Blackberry and the company just going away overnight because of the iPhone and so. But it turns out

if you actually look at IBM, the next 20 years were incredibly profitable for them. And if you just use the stock price measure, which obviously has all the other variables associated with it. But it actually, IBM beat the S&P 500 by 7X over that 20 year time frame on the back of a dead mainframe. And I remember when the IBM Z series mainframe was announced. And this was like a big, this is going to reinvent it. But remember, the only people working on mainframes were IBM. And they came with this new mainframe and like, the big thing was, oh, it's going to run Apache.

And, like, everybody's, like, who cares? Like, everybody's running them on Intel-based servers. And all of a sudden, like, the mainframe world got very exciting, but it was this single source kind of dying innovation and last breath. But for 20 years, they profited from it and beat the S&P 500. So it's super, like, in a sense, what we tend to underestimate is this notion of how long the tail is on something in technology, particularly in the, enterprise space. Yeah. And the best example I've heard of this is that the UK VAT system,

the UK value-added tax system runs on VACs, created by DEC, which I think discontinued it before many of the people listening to this podcast were born. And yet that system is still sitting there running because, well, no one actually decided to spend the money to replace it. Yeah. And so we're in this space now where it could very well be that the PC sees its most profitable 20 years, but yet the least number of people thinking about it or worried about it or investing in it and the least amount of change. So if this is part of the story, I mean, you know, one of the things you see around sort of the chatter around Apple is people complaining that they don't do that much to make that to the max anymore. The max don't improve very

much. Frankly, I think you can see the same thing across the whole PC industry that the S curve has flattened out. And if you buy a new PC or a new Mac, it's not really very different from one that you might have bought four, five, six years ago. Well, you know, Intel has announced, like they're not even focused on PC chips. Like, it's about the data center and it's about IOT and VR and a bunch of stuff, but not like Core I7, the next generation. Yeah, and it seems to this sort of two or three parts to that. One of them is, you know, the center of gravity, the center of the excitement, center of

investment and volume and everything else in the tech industry has moved to the mobile platforms from the PC platform. and so if you were going to make a new widget or a new component or something, of course you would target mobile instead of targeting PCs. The other part of that, though, is that the PCs had sort of reached a point that there wasn't much more that you could really do with them. I mean, I wrote a blog post a year or two ago called Like the Best is the Last, talking about piston.

I started out talking about piston-powered aircraft in the 40s and 50s. And it's not that they stopped getting better because people stopped caring because jets were there. It's they stopped getting better because there just wasn't anything else you could do. You couldn't make them any faster or any more efficient. Well, I think that the key to add, you know, like a different, a new kind of view on this, too, is that what really also it matters is the API ecosystem that's being written to to provide capabilities for people. And in the business world, what really took much longer than anybody thought was the transition to HTML and the browser. And so you have on the one hand the mobile ecosystem where every business is busy building apps for direct. consumer. You know, no matter what retailer you are, no matter what airline you are, no matter

what insurance company, medical records company, you're building an app to interact with your consumers. And the question is, what are all of the people in your company using to do work with each other? And for many years, and for a very long tail, those were client server apps built on the Windows platform. And this is an area of the Mac missed. Like nobody ever built these on the Mac. And so part of what resurrected the Mac and brought it to... was the internet, was the internet, but it took like 20 years. So this was a famous kind of, don't get out of the Trojan horse before you're inside the city gates,

where Mark Andreessen said that Netsgate was going to turn Windows into a set of badly debug device drivers because all of the applications would be done inside the web browser. That was pretty meta of you to build like two sayings on top of the time. I do my best. But this was the point that all the applications would happen inside the web browser, and the web browser would be on the PC, and suddenly no one would care about Windows APIs anymore. And that was, again, supposedly going to happen with Java on the client, which tells you how long

ago this was. Again, people listening to this, you've never used Java on the client. Or Netscape. Or Netscape. Yeah, many people who've never used Netscape. But that was like the vision in 1995. And it took until, you know, arguably at least 2005 for things like Ajax to start making that at least vaguely plausible.

Ajax, pioneered by Microsoft, ironically. And so you get things like Salesforce founded in like 99, I think, or 2000. So just before the crash. Right. But it really took at least a. decade after that before that became something that even vaguely looked sensible. And of course, now it seems entirely normal that you can basically use PowerPoint written in HTML inside a web browser. And you can create all of these tools and applications in a browser. But it

took 20 years for it to really become a thing for all of those enterprise apps to me. And it's super, I mean, it's really surprising how long that that took. And part of it is because it takes, like most enterprise applications, you know, you talk to a CIO and they're managing a portfolio of a thousand, two thousand, three thousand, what they call applications. And It takes a very long time. Those aren't all resourced evenly. Like in an enterprise, you don't have 3,000 teams. So just to be clear, when you say they might have 1,000, that might be like 500 different

things built on the same Oracle thing or something. Oh, right, right. So there'll be like an expense reporting app, but then there'll be the manager version of the expense reporting app, or there'll be the reporting tool for the expense reporting, or there'll be the travel people who input receipts or anything like that. And then, you know, forget something like budgeting, which involves. dozens of departmental level things, vendor things, invoice approval, and this, so there'll be so many of these apps,

and many of them have, they're also baked into the workflows of the company, and so it takes a very long time, because unlike a company, you know, an enterprise is usually run with a pool of resources to build apps, and only the biggest apps have dedicated teams year-round. And other apps are star for resources until they need changes at the fiscal year boundary or tax law changes, or they add a new geo, or emerge, and acquisition integration. And so the tail is very long to get these done, but the demand was there immediately.

And so now what's happened is we've reached a point where most enterprises have moved all of their applications, except for very, very few to the browser. To a web interface. To a web interface. And you could see this every day. You go to a big metropolitan hospital, and there's an 85% chance they're using one of the two EMR systems from either Epic or CERner. And so the doctor or the nurse is looking at your lab results in a browser in that application.

Now, what's interesting is, as a hospital, they're probably even doing that through a Citrix client because they don't even want any state at all on the PC, but it's still a browser. Certainly when you're talking to... And five years ago, that would have been a Windows app. It absolutely would have been a Windows app. And you see the same thing within the, if you unfold a hospital, like looking at CTs or MRIs. Like, five years ago, that was a very complicated Java application. that didn't really work, that needed a firewall specially configured and stuff like that. The same goes for American Express travel agents or 401K managers or customer service reps.

They've all got these customers. This is just an aside here. People talk about getting this conversation over and over again, getting work done on PCs and you need these apps and you need to do your stuff on a PC. The vast majority of work that's getting done on a PC is not actually happening constructing complex financial models in Excel or doing stuff in Indesign or in Xcode or in AutoCAD. It's, you know, you've got a Windows box that cost your CIO $300 and you're sitting there putting stuff into a SAP client

or into a hospital management program or something you've got like... Or you're just doing email. Yeah, but you've got a bunch of those sort of specific vertical apps that need no CPU power. Right, right. Or anything like that. They just frontends the data in some way. They're not, they don't really require multiple monitors.

They have no device connectivity. And now those apps, like they don't even run locally. They require no. So it's actually really interesting. like to go through the U.S. workforce and to just sort of take a quick look at like the 156 million or so people with jobs in the U.S. So this is just the full employment of the U.S. And you could get these on the Bureau of Labor Statistics. It's super interesting data.

I used to use it all the time to figure out addressable market. Yeah, I use that to count up to do a chart of how many elevator attendants there were in the USA since 1900. Okay, that's a good one. Which is actually a category. It still is, yeah. But there's like 20 or so top level categories. And, of course, the top one is Office of Administrative Support at $23 million. But if you go all the way to the bottom, which is farming, fishing, and forestry at about a million.

But in between, it's a pretty straight, linear curve. And very quickly, if you just put them in buckets, about $100 million, basically don't need PCs at all and just could use their phone for everything. Like a really good example of this would be construction. So, you know, five years ago, like if you were to do a construction project, the first thing that they would do is, is come in and say we need a phone line and you need to lease a PC in order to remodel your kitchen because we're going to have the PC for looking up the project and looking up parts supply lists and stuff

the whole time and we need a phone line to connect. And now they all just show up with their phones. And they're not even like issued by the construction company because most construction workers are vendors. And so you can't even have a job without your phone. And so all of a sudden like the whole use case sort of goes away. And just to put a number on it, there's about 600 million business computers in the whole world.

And so if you look at the U.S., there's probably about 100 million. And so then you go, wow, and if 50 million of the people need PCs, there's about 50 million that are controlling cash registers, controlling elevators, controlling robot arms that do machines, that are in ATMs and a whole bunch of stuff. And then the question is, of the 50 million that have jobs that are sitting in front of a computer, what happens to them? And this is the thing we kind of push away at

that some portion of those are And people always tend to react to this by saying Well, I need to run Xcode I need an IDE, I need to build a model in Excel Okay, fine That is running your payroll on a mainframe That you can't switch to a PC Because it needs to do stuff that any mainframe can do

Fine, that's some portion, that's 5, 10% of the base Then there's all the other stuff Which is stuff we've been talking about What is that PC at the nurses station in the hospital? What are they doing with that? And at what point does that become a browser? And at what does that platform mean? What is that device?

And this is sort of, you know, you get to the kind of the $200 device, the $200 PC here. And there's sort of, there's one narrative that says, well, of course, that should be an iPad. Or it should be a cheap Android tablet, you know, barring arguments about Android and so on. There's another argument that says, well, really this should be, I mean, coming back to, you know, the browser as a platform from Mark from 20 years ago, So there was the parallel Larry Andreessen fantasy that you were going to have a Java. What was it called the NetPC? Yeah, well, that was Larry Elson and the network computer. But yeah, the same thing.

We're just interesting all the things that made me cringe in the 90s. Well, you know, there's a difference between early. There's no difference between being early and being wrong. Yeah, it was wrong at the time. But now, to me, a Chromebook is exactly what Larry Ellison was trying to do 20 years ago. I mean, technically the architecture might be different. It's exactly the same thing. It's a web browser that runs applications over the network.

And, you know, this notion of having a device... It's a terminal. It's a terminal. It literally is an IBM 3270 terminal, excepting connected to a single computer by a wire. It's connected to all of the Internet. And that's actually a huge, huge difference. And so it's very interesting to think about, like, what are the alternatives? Because you could have a Windows PC that has a browser in it.

You could have a tablet, like from an iPad or an Android tablet that has a browser. that has a browser and apps in it, or you could have a Chromebook, which has a browser, and now with Pixel and the most current one, also runs Android apps. And so I think it's super interesting from an IT perspective,

because, you know, really what's on our mind is just, like, this is a cost containment problem for IT, because now they've, what IT is very clever. Like when PCs came out, what happened was you would buy a PC for like $1,500, but it turns out that all the profit went to Intel and Microsoft,

for that, and then all the cost went to IT. So they've seen this before. So now they're very smart. They like, tell everybody you have to have a phone to join our company and you can't, you have to have phone, and they publish your mobile phone and the listing, but then they put their own monitoring

software on it so they can wipe your phone and remote manage it, but they're not paying for connectivity, they're not paying for the device, and it's sort of a compromise you make. It's sort of like using your car to commute to work. And if you break it, you have to take it to the Apple store yourself. Right, right. And so they've really pushed in that direction.

And so the question is they don't want a device that brings them costs. To put a number on it, in the late 90s, Gartner coined this phrase, the total cost of ownership of a PC. And you would buy a computer. If Apple loved that chart. Yeah, yeah. Well, and we at Microsoft at the time really hated it because you would buy like a $1,500 computer. And Gartner would tell you it costs like $40,000 a year to give that computer to an employee. And it turns out like, wow, it was the greatest shell game in the world to push all

the cost of the customer. It's not unlike on-prem computing does the same thing. People complain about AWS pricing, but the flip side of it is like running a data center yourself is 10 times as expensive. And the same goes for having a client device. So the question is, given the choices of client device. So as you're sitting and looking at that nurses station, you've got a Windows PC there that has management cost. Do you replace it with another device that has management cost, or do you replace it something with something that gives you some fundamental advantage? Right. And I think that that is just super interesting. Because, like, you could take the doctor station and the nurse station.

You could even take, like, the bankers is my favorite. Because I think if I look at the numbers, there's 8 million people in the U.S. that work in business and financial operations. And, you know, okay, so it's not 100% of them that are doing Excel modeling. In fact, the vast majority of them aren't doing, like, anything like what we think of as sophisticated Excel use. In fact, the other variable that's really changed is the rise of the browser and the rise of SaaS applications has made it

such that the amount of ad hoc analysis that you have to do is greatly reduced because the tools themselves, this is what Salesforce has done, the tools themselves have added The tools are doing the work. The tools are doing the work because the PC was not just about moving where

computing happened, but it was, the data was on a mainframe and the format that you could get the data was fixed. And so if you wanted to see sales by country and the report was only sales by region, you had to get all the regions and add them up yourself. You would download everything into Excel and manually edit it.

All the features of Excel around list management and pivoting were all designed. They're all unbundling a mainframe. They're all in bundling the mainframe. And so then this notion that Excel is a universal tool came to be because it was the only tool where you could just be in control of what you wanted to do. And now what's happening is it turns out if you're a customer support agent, you want to understand the flow of the data, well, whether you're using Zendesk or some other CEM solution or something, like the tools

have bundled in the capability to do all that in a way that doesn't tell the business oh, you're just going to be a cookie cutter business by using our software. It's in fact, just what SAP did. You can be a supply chain and be your own supply chain if you use our software. And so

that to me, this SaaS plus the browser has really changed the dynamic of what IT is willing to support as a computer. So there is SaaS and then there is a browser. And then there is a device itself and what do you have to deal with? Right, right.

And as you go to a sandbox model of iOS or to a lesser extent Android, you take a whole step change in the management cost of having how you deal with that device. And if you go to a pure terminal, quote, unquote, device like a Chromebox or the network PC or 20 years ago, then you get arguably another step change, which is incidentally is why the added putting Android apps onto Chromebook is kind of paradoxical because you turn it back and suddenly you don't have the advantage that Chrome has over Android anymore. Yeah, I have to say, I found, like, to dive in on that one, I found that one, you know, somewhat puzzling to begin with, because the whole value proposition of a Chromebook is it's stateless. It's less hassle than Android.

It's less hassle than Android. And it's just a browser. And I've been using a pixel for, you know, three weeks now, and it's a great, great device. It's a little big for what I carry around, but it works super well. But the Android apps, on top of the security management, you know, there's a store. now. You have to think about it. You know, you give the device to someone else. What's cached on it? What do you have to think about? Then also, like if you put it out in a school, you have to go turn off the store. You can't let all the students go and download apps. It's sort of, whereas before you

could just block things at a firewall or use ratings or a whole bunch of other stuff that before it even gets the device. And so, and then on top of all that, the Android apps aren't really built for a device with a clamshell keyboard. In fact, one of the things I'm been struggling with quite a bit is that the keyboard, the soft keyboard pops up all the time. And that's exactly earlier on in building Surface. Like, even on Arm, that was one of the big things. And we realized, whoa, if there's ever a keyboard attached, you never ever want the soft keyboard. And the iPad with keyboards handles this super, super well. Like, you never get the keyboard. But so I find that somewhat puzzling. Yeah. But this is kind of the transition is, on the one hand,

you can look at this and say, okay, we've moved all of our stuff. from on-prem or mainframe into the cloud, and we moved the front end from a Windows application that we had to think about all the time and manage an update to a browser. And so the end-user device, now, for all very, very large proportion of the people who have a PC, actually only needs a browser.

Yeah. And so what you want, therefore, is a device that only has a browser in some way because you don't, you know, the trade-off of being able to run third-party apps is a whole other layer of management problems and cost and so on. So if you don't have the apps, then you get rid of all of it. that. And so I think maybe the genesis of this conversation regionally was I sort of thought, well,

you know, imagine the office floor for a health insurance company in a city somewhere in the Midwest and there's 500 people in the building. And presume for the sake of argument, the whole thing hasn't been disappeared by AI and they're all actually still doing their job. But they're all they need is a browser and a keyboard and maybe a mouse. And so what is the CP, what is the box that's powering the browser? Is that a $200 PC from Dell, which is certainly one option? Yeah. Is it a $200 or is it a Chrome box, or is it a tablet? And it seems like in that use case, where it's never going to get picked up and carried around,

then a tablet may be not the right solution. Well, you know, the screen's not going to be big enough. It's touch only, which is not the best interface for a person sitting doing a call center. So it becomes something that looks like a PC, except that what's in that little box, which actually is probably going to look like a hockey puck. Or an HDMI stick.

Yeah, exactly. That's what I think Google should be doing with Chrome. They should be making HTML sticks. But it should be a, it's going to be something that basically has no native IPIs at all and no applicant native. Well, and also it's not going to necessarily have to focus on battery life. It certainly doesn't need to make everything more complicated by thinking about offline or wireless WAN connectivity. Like those are things where the tablet and tablet apps are going to be way, way better. And that actually introduces one other dimension to this, which is while IT and enterprise is very focused on building browser-based apps for the past, say, five or seven.

years and migrating. The other thing is they're under a lot of pressure now to produce like mobile apps, like for iOS and Android. Yeah, so there's a leapfogging thing right, right, right, isn't there? Because you took your mainframe app and you turned it into an on-prem PC app with a Windows app. And then you have the Windows app now not talking to on-prem, but talking to the cloud, but it's still a Windows app. And then the cloud is now talking to a browser app. But then that browser app then turns into a native app on the PC, a native app on a smartphone. And so you've got these kind of parallel development tracks of where is the stuff being stored and what does the client look like and what is

the development process for those kind of leapfrogging each other. Right. But there's also, I also think that there's two other dynamics at play if you're the IT developers building the software to run your company. One of them is that your marketing team is saying we need a direct-to-consumer app. And so they're saying like we're a big retailer, we need a retail shopping app. And so there's all this pressure on that. And those companies have been making software for 50 years. They just never made it for consumers. Exactly. It was the stuff you saw if you kind of leaned over the desk in the stool. And so you've got, you've got that dynamic. The other dynamic, which is very

real, is the IT people want people to want to use their software. And so, like, they're kind of motivated to like, well, we'd rather have it not look like back off as software. And so they're actually drawn to thinking about maybe if we just use the mobile app and the sales associate and the customer both used it at the same time. And in many ways, that's what's like the convergence of electronic medical records. Like, now, like, Epic has a consumer app that you look at to look at your own chart and your test results from your physician and things like that. And like the physicians are looking at it going, wow, the browser one is kind of clunky compared to the consumer. Plus, I have to use the mobile one anyway when

I'm not, when I'm at when I'm doing my family life. And like I'm at the soccer game and a patient calls and wants some information. And so there's this very interesting thing where it's not just it's a consumerization of the experience. Right. And which is a phrase that's been around for a decade. But this consumerization is also likely to create some tension between what do we do in the browser and what do we do as a mobile app? Because the good news is the cloud means that that's the definitive source of truth for data, which is another huge change that's happened, which is now there's only one budget. That's the one in the cloud, not the one that's on everybody's spreadsheet, XLS on a desktop floating around. But the question is, how do you get to it? And

like, what's the interesting thing? And I actually think people are underestimating the the desire for mobile, and because mobile is two big platforms, you're going to start to see companies invest much more in one of the two, because that'll be the only way to really deliver on this. And you can see Apple obviously trying to drive that with the deals that they're doing with IBM and Accenture and PWC and so on. Yeah. Just coming back to the mainframe PC point, and I'm sort of thinking out loud here, but that's what this is. Yeah, exactly. But what happened for IBM was that the platform became something no one really thought about, but they carried on selling mainframes and mainframe software and all the stuff, the business around

that for 20 years. The thing for Microsoft is obviously the server bit has gone away. Nothing gets run on Windows PCs on prem anymore. The Windows PC as client did great out of the internet, because what else were you going to buy? Now we're talking about, okay, the only thing in the enterprise that's Windows, that's Microsoft, or sorry, that's Windows, is these clients that are only there to run a browser and no one is writing anything for any Windows API. anymore or any Microsoft APIs there. And so at that point, this kind of Microsoft kind of becomes completely replaceable,

doesn't it? In a way that kind of, sort of in a way that IBM wasn't, if you see what I mean, because you were still running IBM mainframe applications. Well, of course, I don't like to, I see what you mean, and I understand,

but I'm going to push back on that because, A, I don't like to think about Microsoft as just totally replaceable, but I will look at it. I'm just thinking that the PC is replaceable because it's not running a PC-specific thing, whereas the mainframe was running mainframe specific thing.

Right. So the mainframe, the real business of the mainframe was keeping running like 20 years of software investment. You know, like, you really, you know, until you've really met with like a big bank or a big insurance company telling you about their cobal, like we met with a government agency this week. And they have one application for one branch for one scenario with seven million lines of cobal code. So it's not going anywhere. And they have to find people to help them to manage it and deal with it. Now there's a great, you know, web front end with APIs and stuff now that's all. You see it through the browser, but you know, you're very constrained in what you can do because its responsiveness is a giant cobal application going through an API. But the thing that the most interesting thing in this conversation is the vast richness of the Windows platform is rolled up into office. And that is the pretty much the active development on Win 32, Modulo, the count that you do of like the autocads and the solid works and the DeSos systems and the Photoshop and the Photoshop and like this.

So there's a core of perhaps, and I think you estimated 100 million worldwide. I thought it was like 100 million, maybe 150 million people worldwide that use all of those, what you would think of as like a professional application that's doing like hardcore, sophisticated, creative stuff on a PC. Yeah, like chip design is not like a browser-based thing. It's incredibly sophisticated. But there's, you know, only something, again, just to put like, you know, there's 2.6 million architects and engineers in the U.S. and many of them aren't actually the kind that sit in front of... Many of them are not actually wanting to catch.

Yeah, like, well, they're part of architecture firms, but they're not necessarily the engineers. And so the interesting thing is, like, you know, how much... And this is really where people really push back, because what's going to happen is people, as you've pointed out many times, like people's jobs are going to change, and they're not going to be, like, formatting a word document

for 12 hours at a time, or as I used to think about it as, like, debugging the word document. And so, like, on my Chromebook, I've used, I only use the browser apps in Office 365. And so they work great. And in fact, they're in a race to add back all the features, even though I don't need them. Like, I'm never going to use Kearning or page layout because I don't really print anything and things like that. And so what's super interesting is that's the split. And that's really what happened with mainframes.

Like the platform became existing software. So the analogy there would be like there were specific kind of applications even now that we've won better in. principle on a mainframe than they do on a PC server farm because they're sort of there's character of that kind of application like running a national tax system for example is very different from doing Google search. It's a different kind of computation. And so like there was a core bit of the mainframe market that actually needed to be a mainframe. And there's a whole much larger portion of it that could happily move over to PC. And it just takes much longer than anybody thought. But then and then what we're kind of getting at is there's sort of a similar

piece here of there's a sort of a core part of the PC market that actually needs to be on a PC because you're running Excel or you're running Photoshop or AutoCAD. And the other 90% of it actually doesn't need to be on a PC. It can be on a Chromebook or it can be on something that has a browser or it can be on a smartphone or a tablet. And one of the things that I think really capitalized on what IBM did was they also got smart about how to manage it. And this is something that I think is super interesting, which is they said, well, if it's really important to be valuable on this scenario running this way, let's not radically change this way. Like they didn't listen to like mainframe enthusiasts, they listened to the business leaders. And so what's interesting

is how those systems evolve so that they capitalize on those kind of applications without churning them. Because those people are also not like the biggest about change. I mean, Adobe Lightroom on the PC just got an update. And the forums are just exploding with like the keyboard shortcuts changed or this change or that. And I kept thinking, why did they do that? like why bother to have that big a change when you fundamentally the scenario doesn't change at all yeah well there is always that there's like the one last reset yeah yeah well it's super interesting so quick i'm going to wrap up with like my my desire or what i think if i were running enterprise IT my view of it is that what's going to happen is the call centers of the world

the vast majority of administrative assistants of office workers of mid managers and things are all going to gravitate towards basically sitting in a browser all day, but mostly using their phone. So it'll be like 80% of the communication will happen in their phone, mostly using communication apps and SaaS apps and things like that. And then when they're on their PC doing sort of the creation step of their job, they're connecting to SaaS apps, they're integrating things, and it's going to be in a browser. And then I believe that IT is going to really push for a device that doesn't have an execution

engine on it. And I think it's going to be really interesting to see how Chromebook evolves because I kind of feel like the Android step was a step backwards. When they really had like the exact right device that everybody wants, it just, they wanted like one more thing or something. And I do think that the tablets are on the uptick because that's really what's going to replace the core marketing executive traveling, sales executive travel. It replaces all the executives traveling with laptops. Yeah. And even executive, we say in a very broad People traveling with laptops will travel with a tablet instead. Yeah, and I think that's right.

And I said, you know, to come back to the way I opened this, it's interesting. I remember like sort of 18 months ago or something, I was trying to find stats on mainframes. And I got this number for the install base of IBM mainframes, measured in Mitz, of course, millions of instructions per second. It's not like how many mainframes, it's how many instructions per second are installed. And the number went up, I think, something like 10 times between 2000 and 2008. Nobody in the tech industry was thinking about mainframes in that period. And yet the IBM business was actually doing really well.

And I think we're now at this period where people are still arguing about whether the PC is going away at all. But actually, as we think about what happened over the next 10, 20 years or so, you're going to see a sort of a similar process where the business is going to be there in some form, but just kind of changing. Yeah. Well, it'll drop. What's happened is it's half the run rate that people thought it was going to be. And what happened with the mainframes was super interesting,

which is that the need for information technology exploded, and they couldn't move to that more scalable platforms. So IBM had them locked in. The PC is different because it's an individual empowerment tool. So everyone wants to do their taxes online, so suddenly you need more mainframes. Right. And you don't need more devices running America Online connected

that have the processing power to download chips off to do your taxes. And I think that's the difference in the evolution with the PC is that the number of people that are going to, to build Excel models isn't going to increase or run Photoshop or Autocad. But the number of people communicating with them and coordinating and vendors and things like that that are going to be cloud-based is going to go up astronomically, which is why we see the mobile phones. So that's what we have for today. That was our hallway conversation. Thank you.