Ask Paul: November 24 (Premium)

Happy Friday! I’m finally done moving and traveling for a bit, so here’s a great set of reader questions to kick off the long weekend.

Margins

Zeppelyn56 asks:

Paul, you have reported lately on the financial results from the likes of Lenovo, NVidia ana HP. Do you you ever think about the difference in the low margins these companies actually make as opposed to the sometimes near 40% margins the big tech companies (Apple, Microsoft and Google) enjoy? They sometimes seem obscene to me.

Generally speaking, hardware margins are much lower than those of software/services. And it’s common for companies that offer both—hardware and related software/services—to lose money on the former and make money on the latter. This has been the reality of the Xbox, though Microsoft had always hoped to eventually make money on hardware through in-generation cost reductions, a strategy that never actually panned out. Which explains the Phil Spencer-era strategy pivots to acquiring more in-house game studios (to mimic Sony’s more successful console model) and to subscription services.

This was also the model for the biggest PC makers until fairly recently (historically): IBM (now Lenovo) and HP are very different companies than they were in the heyday of the PC market, while Dell, oddly, remains largely the same. By which I mean, these are/were large, diversified companies that offer/offered a range of hardware, software, and services to both businesses and consumers, the idea being that a slowdown in any one area could be temporarily subsidized by other parts of the business. But the PC market experienced what was likely a natural byproduct of commoditization and then consolidation. And things are different now: PCs are very low-margin products, probably negative margin overall, and the only way these companies can be profitable now is with higher-margin premium PCs, which include business-class PCs, gaming PCs, and what we now call hybrid PCs (convertible/tablet/folding/etc. PCs).

IBM sold its PC business to Lenovo a long time ago, and if you look at that latter company today, you see what I’d call an electronics giant with footholds in PCs, phones, and related devices, and enterprise services and infrastructure (storage, etc.), and a solid presence with both consumers and businesses. HP split off from the enterprise software/services part of the original company (now called HP Enterprise, or HPE) and while we see it largely as a pure-play PC (and printer) company, and thus a hardware maker, it also offers businesses and consumers a range of complimentary subscription services related to security, remote access, maintenance/service, and more. And Dell remains what it was, a diversified company that targets both businesses and consumers with a wide range of options across hardware, software, and services.

When I write about those companies, I focus on what matters to us, meaning PCs, of course, but also whatever hardware, software, and services target individuals, meaning consumers, prosumers, and those individuals who use business-class hardware either through choice or because they work for a large company. But there is more going on with each of them. Take HP as an example: The company has a growing collection of subscription services that target both consumers and businesses, and while I don’t write about this all that much (and maybe should), there are some interesting details about each in its earnings report (most recent here). For example, HP keeps its earnings simple by dividing its business into two segments, PCs and printers, but the PC segment is further divided by consumers and businesses, and then by hardware and software/services. Businesses supply about two-thirds of HP’s revenues overall, and consumers the remaining third. Its key growth areas are hybrid systems,  gaming, workforce (business services), consumer subscriptions, industrial graphics (labels and packaging), and 3D personalization (3D printers, workstations, and services aimed at graphics professionals). And then there are little micro details: HP’s Instant Paper consumer subscription now has over 500,000 paying customers, driving its entire consumer subscription business to year-over-year (YOY) revenue growth.

Apple is, as always, unique: Unlike any other PC maker—and, I think, any other hardware maker in personal technology—Apple achieves high double-digit margins on its hardware sales, often north of 40 percent. (This is getting difficult to measure accurately. Apple, like other Big Tech companies, has gotten less and less transparent in recent years with no repercussions from the SEC, somehow, and it does this to make it harder for outsiders, including investors, to fully understand the relative strength of its various offerings. The idea here is simple enough: The strong offering will prop up the weak ones as needed and as long as the company overall is doing great, no one seems to care.)

The obvious observation here is that Apple only makes premium products, and so that is what enables the higher margins. There is truth to that, but there are other reasons, and we should recognize that not everything Apple sells is “premium” per se or, in some cases, any more expensive than comparable offerings elsewhere. And the company successfully engages in the same customer bait and switch by getting them in the store (virtually or physically in its case, another unique differentiator) with lower-end offerings with the express aim of upselling them at purchase time. And that its virtuous cycle extends to its ecosystem and customers, and that those who have a positive experience with, say, the iPhone, inarguably the most successful consumer electronics product of all time, will be more likely to not just keep upgrading to new iPhones over time, but will also graduate into other Apple hardware and services offerings too. Apple is sort of an “affordable luxury” brand that has done a unique job of capitalizing on its success while maintaining the level of quality customers expect. Customers feel rewarded by Apple and Apple is likewise rewarded by its customers.

And no, PC makers cannot try to copy Apple and succeed. These are just very different businesses. And the iPhone is why.

For example, in the most recent quarter, I noted that the iPhone directly contributed 49 percent of Apple’s revenues but that you might calculate that its total contribution (including indirect via services and peripherals) was probably about 84 percent. Meanwhile, Apple’s Mac business is relatively small, with “just” $7.6 billion in revenues, or about 8.5 percent of all revenues in that quarter.

Over at HP, the PC is directly responsible for over two-thirds of total revenues, (or close to 100 percent including indirectly), and that business delivered $9.4 billion in revenues in the most recent quarter. So HP’s PC business is currently bigger than that of Apple (and HP is the second-biggest maker of PCs in the world). But HP overall is tiny compared to Apple overall. HP not only doesn’t have an iPhone, but it also relies almost solely on the PC for revenues (printers are overly reliant on PCs, of course, though you can print from mobile now too). Different businesses.

Microsoft, meanwhile, has long enjoyed the high margins we’ve come to expect from a software (now services) company, and its big financial shift started with figuring out how to even out earnings so that it no longer has such big financial peaks (new software version released) and valleys (mid-way between new software version releases). This shift started with software assurance and volume licensing (subscriptions for business customers) and extends to today with subscriptions of all kinds and the absence of big annual paid updates. (This is why Windows is “free” and there is no attempt to sell upgrades, and why it’s pushing Office in this direction.) The goal is to have an overall growth line on the chart, not something up and down. But this is also what makes its hardware investments so confusing: These businesses (phones, Surface, and Xbox consoles) will never be profitable because of a “non-virtuous” cycle in which Microsoft never sells enough to achieve the volume it needs to get better components pricing. Xbox tried that “razor/blades” model but never succeeded. And here we are: Xbox is quite different now than it was in 2001-2005.

Google has a similar problem to Microsoft in hardware, but it’s a different company otherwise because it was born in the web era and all of the software and services it releases are mobile/web/services oriented. This is good and bad, but if it weren’t for advertising (essentially a services business), Google and its most popular mobile software/services couldn’t exist: Its enterprise and paid subscription businesses are relatively tiny, with Google Cloud (including Workspace) delivering $8.4 billion in revenues in the most recent quarter (just under 11 percent of the total).

I guess the way to look at this is that these are all very different companies. They compete with each other in some ways but not others, and they all have different strategies, structures, and approaches, and each is very successful in specific ways, but most are less successful in many other ways. But comparing them at a high level is tough and perhaps unfair, even though we all of course do that. It’s hard not to.

Microsoft Offline Sync

wright_is asks:

Did Microsoft Offline Sync (SMB shares, not OneDrive) ever work properly?

Seeing this term was like using a muscle you forgot you had, meaning in this case that it made my mind hurt. This is the old client-side caching thing, right? (C:\Windows\CSC, God help me it’s still there.) And if so, it’s something that the rest of the world moved away from a long time ago because of the shift to cloud computing and now remote/hybrid work. I’d have to look it up to be sure. But I assume if you have an employee using it you must have some kind of server-based in-house infrastructure.

I assume it did work in traditional, pre-cloud infrastructures, and that this was (is?) about using folder redirection to sync a local folder with a remote folder on a storage server. But this is similar to the (individual/consumer-oriented) discussions (here, for example) about how backup has evolved in recent years to become sync in most cases, with cloud storage services on the backend and PCs and other devices on the edge.

I don’t know what your infrastructure looks like, but surely there is a more modern SharePoint/OneDrive for Business sync capability (on-prem, cloud, or hybrid) that is automatic and just works. And that this person should use that.

Linux + PC makers

helix2301 asks:

Lenovo announced a new line of machines with fedora. Dell has line with Ubuntu and HP has Pop! OS. While I’m a big fan of Linux why are pc makers doing this are they really going to sell that many units? How do the manufacturers determine what distro to put on there pcs?

This is very much related to the modern PC business model noted in the first question above. PCs, overall, are a low-margin business. But in addition to relying on higher-margin PCs to float the entire business, PC makers have tried different strategies over the years to improve profitability. Bundling what we now call crapware is one such strategy, but eliminating costs is another. And one of the bigger costs PC makers have is the per-PC Windows license. The trick is that they can’t just replace PCs with Chromebooks or replace Windows with Linux broadly: Businesses do not want that for most of their customers, and few consumers would either.

But there is a market for Linux on PCs, primarily with developers, and on workstations with customers who have specific needs related to science, engineering, data visualization, and so on. And as Linux has become more mature and PC makers can certify complex combinations of hardware components to work reliably, it’s easier to support Linux formally and address those customers’ needs. But this makes sense specifically because these PCs are so lucrative (expensive and high margin): You won’t see HP offering Pop! OS on its entry-level Pavilion laptops anytime soon.

As for which distributions each has aligned with, that I don’t know. But I assume these are mutually beneficial partnerships of whatever kind and that each company is at whatever stage of experimentation on the way to seeing whether this is a sustainable and necessary alternative to Windows that enough customers want. The Chromebook market is probably similar but further along, in that these tend to be lower-end devices where the margins are much thinner. So where Chromebooks become a viable alternative to traditional Windows PCs at the low end, perhaps premium Linux PCs and workstations can become a viable alternative (from the PC makers’ perspective) in specific lucrative markets too.

AI: Disruptor or flash in the pan?

Markld asks:

I happen to use Open AI and the MS Chatbot frequently to help me with python code as I like checking code before I run it. It saves me so much time. It catches those syntax errors and I actually am so much better at coding than before AI. This is a practical application for it.

To me the jury is out on whether it going to be beneficial or not beneficial. Maybe it depends on a person’s work. I do believe its going to be a disruptor in many ways and in the end we will adapt.

I know you have written a lot about it but what’s your ‘spider sense’ telling you where it’s heading and will be like in the next 5 to 10 years? Technology changes so fast I think I maybe asking you the impossible. If you said I don’t know, I would respect that!

I mean, I don’t know, no one does. But after going through a “7 stages of guilt”-type thing with AI this year (thanks, Microsoft!), I’ve settled into a definitive stance on it. And I believe that AI will come to be seen as the single biggest advance in technology in history ever. But we’re also in the nascent stages of this transformation. And it’s going to be messy.

I will be writing a lot more about this soon, it’s inescapable. And like my coming “ecosystem” series (for lack of a better term), it’s going to consume us in the sense that it’s an infinite topic that can’t be explained or concluded in any way in a single article let alone a single answer to a question here. But I do think there are a couple of key points that impact us all in the context of this site, technology enthusiasm, and day-to-day life. So, it’s worth discussing at a very high level. With the understanding that this discussion will never end, nor will the debates about the good vs. evil/ethical/safety aspects of AI, at least not in our lifetimes.

Practically speaking, we will be wading through a lot of AI hype in the coming weeks, months, and years. Some of this will be warranted, a reality that AI naysayers seem to have trouble admitting, and some will not. It’s our collective responsibility to sort through the nonsense and find the ways in which AI is truly transformative for the better. And given our inability to distinguish between fact and fantasy right now, it’s reasonable to worry about that. There will be educated takes on this, uneducated takes, and scammers taking advantage of the latter. Again, it’s messy.

In my tiny corner of this tiny corner of the broader world, I am fascinated by how AI will transform workflows broadly, but also devices, software, and services specifically. There’s no reason to get high-minded about this: AI is the next logical step in an evolution that included such things as the first electronic spreadsheet, word processors, graphical user interfaces, and every other advance that transformed personal technology: In each case, advanced capabilities were made available to the masses while eliminating the need to white lab coat-wearing overseers with narrow skillsets, and while those in power always resisted these changes, they were inevitable. Microsoft, for whatever it’s worth, historically played an outsized role in this democratization of technology (which is really about capabilities, not tech for tech’s sake). So it’s interesting how prominent they are in this shift.

AI’s explosive expansion of capabilities across such a broad (perhaps limitless) spectrum of tasks makes it hard to discuss on many levels. But we will also be dealing with a Luddite/Chicken Little contingent that has a hard time seeing past the world as it is arbitrarily today, or their own little kingdoms which they’re trying to protect. Here, I’m reminded of the story I’ve told many times about the TechEd pre-conference session I hosted in which an Exchange admin rhetorically asked whether his last official act in that role would be to migrate his company’s email to the cloud. The answer was yes, and that, yes, he would lose that job. But this was what was best for his company—which was not created to build a huge in-house infrastructure just to manage email but to sell widgets or whatever—and that he was in a unique position to not make that decision because he would act in self-preservation at the expense of the company. The shift to AI will make this type of complaint much more common. But it’s the same basic problem. (Again, I will be writing more about this.)

The debate over whether AI is “good” or “bad” falls into this type of debate, but this one is easy. It’s not either. It can be used for both. It’s up to us to use our newfound superpowers for good or evil. Both will happen. It is happening already.

But to bring it back down to me, this site, you, and anyone who cares about our little world, I see AI mostly through the lens of what I’ll call productivity, though I mean that broadly. It will make hard things easier for more people, doing what that Exchange admin feared by eliminating the need for narrow experts and middlemen, and expanding the audience of “jack of all trades” types in the masses. It will let us focus on the task at hand and not get sidetracked by researching, learning, or adopting new and unfamiliar things that are tangential to our personal or professional lives. This is important, and ultimately it’s good. The sticky part is the short-term: For example, we all understand the need to move off fossil fuels just as clearly as we don’t know what to say to those whose jobs in the oil or gas industry are lost as a result of that shift. It’s the right thing to do. It’s hard on some people of the generation that’s alive when it happens. These things are not mutually exclusive.

Predicting the future is always hard, and has never been a personal strength, and the rapid explosion of AI capabilities makes it exponentially harder. But anyone who has longed wistfully for the humanist future imagined by Gene Roddenberry in Star Trek surely understands that this is the first step in realizing that perhaps impossible dream. Every era, every shift and transformation, will have its rough patches. And they will be lost to history.

But first things first. I just want spelling and grammar checking to work properly in Word.

Kids and technology

ken_loewen asks:

Paul, we’re giving our 12-year-old a Windows laptop for Christmas. I haven’t seen you write anything recently about Windows Family Safety or other approaches to helping insulate her from the toxicity of the open internet. Are you planning on a chapter in your Field Guide on this topic – or have you seen any really thorough references in this vein?

Yes, I do have a Microsoft Family chapter planned for the Windows 11 Field Guide, but my experience this ever-evolving functionality is limited because my kids are no longer kids: My son is 25, my daughter is 22, and neither lives with us anymore. Also, the world has changed a lot: Most kids will get their own smartphones before they get their own PCs, and regardless will spend a lot more time with those devices, which have their own management platforms for kids.

You could get lost in the weeds here, and the overly technical may still feel the need to build a locked-down infrastructure at home similar to what they created or experienced at work. But I feel like a loosely managed approach is best. And while I do not use it today, Microsoft Family as exposed in Windows 11 by the Family app—it’s really an online service—should meet that need on the PC. (It gets more complex on mobile, though I guess it’s at least worth looking at Microsoft Family Safety. I feel like the platform makers probably have the best tools there.)

I’m not aware of any Microsoft Family resources, sorry. But it’s straightforward enough: Assign yourself and your spouse/partner as family organizers, create/add a child MSA for your daughter, and then examine and configure the content filters, screen time, and other features to your needs. There are activity reports and notifications for exceptions, and family features like shared email, calendars, OneNote, and more that you may find useful.

Newsletters

christianwilson asks:

This may be better as a Behind Thurrott.com article but I thought I would ask it anyway. When you took control of the site, you wrote about how running a newsletter is costly which ultimately ended up with you partnering on Windows Intelligence. I am curious what makes the newsletter component so costly for a website. Is it the cost of running the back end services, the manpower involved to maintain the newsletter and the site, some other reason, or a combination of a few?

I do have a coming Behind Thurrott.com post about newsletters coming now that our newsletter partners at The Intelligence have gone public with the final branding and their optional paid subscriptions. But there’s not much to say about the cost of us doing this ourselves. It’s just too expensive.

Imagine I wanted to simply continue with what we had done before and send out a daily newsletter five days a week that included all summaries of all or most of that day’s posts and then one premium newsletter each Monday. I don’t actually remember or know how many subscribers we had for either now, but let’s say it’s 20,000 people each. So that’s 180,000 emails a week and (roughly) 720,000 per month. Looking at the Mailchimp pricing structure, I could send 300,000 emails per month for $535 per month, which is less than half the number we were sending. Our monthly hosting costs for the website are less than $535, let alone whatever the cost would be of sending the real total number of emails. (I think it was somewhere between $700 and $1000 per month.)

We looked at every service imaginable and it just never made sense. I was surprised by this, as I assumed this would be relatively inexpensive. I don’t need the newsletter to make money per se, but I can’t spend that much on something that will literally never make money directly. The number of views coming in from email links just didn’t justify the expense.

Windows, Ignite, and Microsoft

will asks:

The past two weeks have been almost overload on tech news and information, all be it mostly from Microsoft, but I did have some thoughts from all the nuggets from Ignite last week and wanted your thoughts.

Almost?  😀

First, Satya mentioned that the future UI is all around AI and Copilot.  While some things were talked about on the future of Windows, more in a minute, since words are very selective I am guessing that Copilot is going to be THE thing in the next version of Windows.  What this looks like is open, but he mentioned UI and curious your thoughts?

I think Microsoft did a good job of course-correcting on the Copilot branding between its September event and Ignite, and that the current positioning makes sense: What we now call Copilot is the base level of functionality across Bing, Edge, Windows, and Microsoft 365, and that some of those offerings (Windows 11 a little bit, Microsoft 365 a lot) will expand on the base level of functionality with context-specific capabilities. But Copilot in Windows 11 today is not particularly useful, especially the stuff that is specific/unique to Windows. And that does make me wonder how or if that will expand in the coming year, and whether that happens in Windows 10, 11, 12, or all three. Should Windows 12 be unique too? Or is the current model what we get?

We didn’t get any clues about that at Ignite. But we did previous to that, and I keep referencing Stevie Batiche’s appearance at Build 2023 and how he described how Microsoft would evolve its use of AI across three “application structures”: Beside applications, inside applications, or outside applications. Copilots are, by their nature, beside applications. So perhaps the point of Windows 12, or some future Windows version, will be the shift to AI capabilities “inside” the OS, that this thing will be about integrating AI at a deeper level. If so, the current Copilot model has a lot of maturation ahead of it. Anyway, that’s on theory.

Second, the new AI chip that MS announced for the server side is interesting in that it is all ARM based. 

Yes, it is. We talk about Arm on the client for all the obvious reasons, but for Microsoft and the industry more broadly, shifting to Arm (or at least Arm-style) chipsets in cloud datacenters will have an even bigger impact on both efficiency and cost. I think Arm is the future of the datacenter.

While the C100 is a server platform, do you think there is a smaller version in the works for some future personal hardware say Surface?

Yes, at least sort of. Microsoft partnered with Qualcomm to create the minimally differentiated SQ series chipsets for Surface, and it has partnered with both Intel and AMD on similarly customized x64 chipsets as well. And it’s reasonable to think that these partnerships will result in further silicon customization and, in the future, even purely Microsoft-designed silicon on the client. And I think that’s true whether Surface survives as a business or not: With Apple unwilling to license its Apple Silicon chipsets to PC makers, a Microsoft silicon family (created in partnership or in-house) could supply that role. We’ll see, but to be fair, it’s more likely that its current partners will just do this themselves.

Third, Microsoft released a new remote desktop app for connecting to Windows 365, and it is called Windows.  Two of the sessions I was in talked about this and it was mentioned offline that it is not a coincidence the app was called Windows, not remote desktop or Windows 365, and there are some bigger things coming with Windows 365.  It was mentioned in the session they are not 100% there yet, but getting close.  So all of this to say, thoughts on this time next year we see a bigger push with hosted Windows?  Maybe a turn toward the idea of a thin client desktop, powered by some ARM processors, for users that just need Windows?

There is only one logical endgame here that makes any sense at volume, and it has (almost) nothing to do with delivering an entire Windows desktop via the cloud to whatever clients. And that endgame is cloud-based Windows app delivery. This is a bit simplistic, but in many cases, what keeps individual users on Windows isn’t the full breadth of the Windows app catalog, its one app (or one to three apps, or whatever).

This is exactly the type of thing that would make sense from the cloud. With a local solution—Parallels Desktop on the Mac or Chromebook, for example—you need to have that entire virtualized Windows install on disk. But with the cloud, delivering just an app becomes possible, and Microsoft could scale its pricing based on usage (number of apps/time). Plus it could architect Windows/Windows 365 in such a way that this becomes feasible. It feels like the type of thing Dave Cutler would be good at.

I’ll check out what’s available for Windows 365 sessions from Ignite to see if there’s any hint of this. But that seems like the logical push to me.

OneDrive on mobile

ggore asks:

Hi Paul. I might be missing something here but why does Onedrive on Android require a Microsoft login to view a shared photo? This happens regardless of weather I set it to “Can Edit” or “Can View”. Like I said, I may be missing something but I don’t remember Onedrive asking for this in the past? Is this some sort of new thing Microsoft is doing to track you?

I don’t see that OneDrive requires a Microsoft account, though this perhaps depends on how you share it. To test, I found a photo, chose to share it, used the “Copy link” option, and pasted the link in a private tab in Brave. It worked fine. But you mentioned the “Can edit” and “Can view” options, so I tried that too and see that there is a “Require sign-in to access” option as well. And I believe that does what you’re looking for: I was able to open the emailed link (again, in a private tab) without issue.

But this is also sort of interesting in a way that would not be obvious to many but is related to my recent OneDrive complaints, though I didn’t write about it because, in the scope of the serious issues, it’s a relatively minor issue. But it goes like this: Windows 11 has a built-in Share feature that I use a lot—mostly to transfer book screenshots from a PC with a clean install over to the PC I’m writing on—and it works completely differently for files that are in a OneDrive folder than it does for those are not. More specifically, I prefer to use Nearby Share to transfer files from PC to PC but this option is not available in OneDrive folders. And so when OneDrive Folder Backup is silently enabled without my consent, I’ll try to use that feature and it’s not there. It’s one of the lesser offenses of OneDrive on Windows.

I’m not even sure that makes any sense to anyone, sorry. But the general point is that there is OneDrive sharing and then there is platform sharing. And in general, I prefer what’s available on the platform, whether it’s Windows, Android, or iOS. And so when I read this question, what went through my mind was, what does this even look like on mobile? That is, I know OneDrive sharing includes the ability to send or copy a link to the thing being shared. But on mobile, there’s an internal Share pane on both platforms that links to email, messaging, and whatever other mobile apps, plus platform capabilities like Nearby Share on Android and AirDrop on iOS. And I tend to stick to those features. In other words, I had to look.

On iOS, OneDrive puts the platform sharing features up top, and the OneDrive-specific choices (Copy Link, Invite People, etc.) are deemphasized below that in a text list. On Android, where Microsoft has more control over the experience, the OneDrive-specific sharing options are up top and resemble what you see in Windows. In both cases, however, sharing is better than Windows, because you can choose between the OneDrive and platform sharing options. I wish Windows worked that way.

Manifest v3 and Chromium browsers

AnOldAmigaUser asks:

Apparently, Google is going to be going to Manifest v3 for Chrome in June, 2024, effectively crippling ad blocking. Not that this is an immediate issue, but will this affect just Chrome, or will all the Chromium based browsers be affected? I assume, at a minimum, that any browser using the chrome dot google dot com webstore will be affected.

We’ll see: During its original attempt to implement this change, Mozilla and several Chromium-based web browser makers said that they would not support or implement Manifest v3, the only notable exception being Microsoft, which of course embraced this technology because it, too, tracks its users online activities. But there is still some debate about whether ad and tracker blockers can work effectively using this specification: AdGuard, for example, says that the changes Google made in the new implementation addressed its original concerns and that its ad blocker will continue to work as well as always. I’m curious to see how browser makers respond to this now.

And as an aside, why am I being informed that the question above might violate community standards. Perchance, the settings are a wee bit too tight?

This is controlled by OpenWeb, and while I can (and do) adjust the setting, I don’t have any way of seeing why an individual comment was moderated. However, based on my experience with it, I’m guessing it was the word “crippling” and that the worry was that it was a commentary on the author and not on ad blocking.