Ask Paul: July 17 (Premium)

Happy Friday, and welcome to another set of surprisingly diverse questions, albeit with a curious focus on thin clients.

Which company should buy Arm?

OldITPro2000 asks:

On Windows Weekly everyone briefly discussed the rumor from Bloomberg about Softbank selling all or part of Arm Holdings. I agree that it makes no sense for Microsoft to buy them, but I’d like to know your thoughts on who might. A sale could potentially shake up the market if the new owner changed (or eliminated) future Arm licensing.

For those unfamiliar, Arm designs its chipsets/SoCs internally and then licenses them to third parties, but it doesn’t manufacture or sell them itself (in sharp contrast to Intel). So when we talk about “Arm processors,” for lack of a better term, we’re really talking about a variety of offerings from firms like Apple, Huawei (HiSilicon), Qualcomm, Samsung, and many others that license the technology and then create their own chips.

Anyway, Arm is owned solely by a Japanese telecommunications giant named Softbank, which purchased the firm for its intellectual property in 2016 for about $32 billion. Earlier this week, the Wall Street Journal published a story claiming that Softbank was considering selling Arm, though it’s unclear whether it will actually pursue a sale or simply decide to hold on to the firm.

Given how important Arm is right now, I have to think it’s worth a lot more than $32 billion here in 2020. And that cost, combined with suitability, really limits the number of companies that could/would purchase Arm.

And I see only one obvious possibility: Intel.

Intel has been down this road before, and at least a few times. In the early days of the original Tablet PC wave, its processor designs weren’t mobile-friendly enough, so it purchased an Israeli firm and used its technology and designs to create the Core processor family it’s still selling today. Intel also failed to make Itanium the standard for 64-bit computing and so it had to license the AMD-64 architecture, which we now know as x64 (or x86-64).

Today, Intel is likewise failing at mobile, and it seems to have a few side projects aimed at creating chipsets that mimic the Arm big-core/little core (performance core/efficiency core) design. But I don’t see how Intel survives without actually owning this architecture. Over half of Intel’s revenues come from cloud/datacenter, and that market will be all Arm within a few years, thanks to its heat/efficiency/power benefits. And most of the rest of its revenues come from PCs, a dying legacy business. Intel will be fine on PCs for the foreseeable future, but there’s no growth there. And Intel has zero market share in mobile, which is the volume part of the personal computing market.

So, we’ll see what happens. But the big picture here is that I see Intel failing unless it makes major changes immediately.

Microsoft layoffs

SeattleMike asks:

On Twitter you mentioned layoff rumors? Have you heard any more detail?

No, and because of the sensitivity of this issue, I don’t have much to say beyond the fact that I heard it from a long-time friend at Microsoft and that involved program managers in the Azure group.

Privacy Shield

ggolcher asks:

Any thoughts on the US-EU Privacy Shield getting struck down? I own a company with a mobile app for expats that serves people worldwide. Our lawyers are suggesting we move all of our data to European servers or block access from European IPs! This is devastating for us and I suspect a ton of companies.

I had a bunch of tabs set aside to write about this yesterday but I realized it would just generate the typical xenophobic and close-minded nonsense comments from the same idiots as every article I write about the EU, Huawei, or whatever, and because this isn’t central to the mission of the site, I could safely avoid it.

But since you ask, I’ll just say this. As an American, it’s uncomfortable to know that we are the problem in this situation, that the issue here is that Privacy Shield agreement would not have protected EU citizens from privacy invasions by the U.S. government. In other words, what has to change is us, or U.S. law. And no offense to half my readers, but that is not happening under the current administration.

As always, I’m hopeful for the future but open-eyed about the possible realities.

Thin clients

Three (!) questions about thin clients this week, which is interesting…

kevinbouwman asks:

I noticed a commenter on one of your OneDrive stories this week requested that you do and article on setting Windows up as thin client using OneDrive as much as possible and local storage as little as possible. I would like to second that. What gotyas might there be? I remember trying to copy my primary data folder to OneDrive and being told that the file path for many of my files exceeded a limit. I gave up at the time because it was clear I was going to have to do A LOT of tedious folder maintenance to change that. I would love to have a comprehensive check list for prep and for the transition.

I’ve been living this dream for years, though I’ll point out that a modern PC—desktop or laptop—in no way resembles the thin clients that Oracle and Sun championed a bit too early back in the 1990s. But I noted this change years ago: For a long time, our storage needs on PCs grew and grew, and I have very clear memories of the things I’d do to prep a laptop for a work trip: I would make sure that it had all my latest work data, my entire music and podcast collections, and some selection of videos so that I’d have everything I’d need for work or entertainment while away. This kind of thing required a lot of storage.

Over time, this changed though. The rise of the iPod and other MP3 players meant that I could just use a device for music and podcasts, and the addition of an iPad/Kindle meant I had a device for reading and watch videos. And then smartphones took over all of that. What that meant to the PC was that while the other specs kept improving and improving, my/our storage needs kept declining.

Today, I could easily get away with a 128 GB SSD on my laptops and/or desktop PCs because I no longer need to store so much content. And thanks to OneDrive and Files on Demand—coupled with some good organizational skills, which I feel like I must have documented somewhere—my day-to-day storage needs are tiny.

I have three document-related folders that I sync to all of my PCs: To-do, which contains my current articles and article series; 2020-07, which contains the articles (and related graphics etc.) for the articles I’ve already published this month (and this changes month to month, of course); and Book, which contains all the files related to the Windows 10 Field Guide. That’s it.

To-do takes up about 5 GB, but should be trimmed a lot (I have a lot of Programming Windows and .NETpad stuff in there). 2020-07 is 328 MB. And Book is about 1.5 GB. So maybe 7 GB of data all together. That’s nothing, even with just a 128 GB SSD.

I will think about writing this up more formally. But the key to keeping data in OneDrive, and avoiding those long file names/locations, is organization. And with regards to thin clients, the short version is that PCs are obviously rich clients in the sense that they run heavy, locally-installed applications. But they are like thin clients when it comes to storage.

ErichK also asks:

Paul, Leo on Windows Weekly has been mentioning thin clients a lot lately. Do you remember the interview with Larry Ellison in the documentary Triumph of the Nerds where he says he “Hates the PC” and was advocating an “Internet Appliance?” Sounds like the same thing. In 1996. So is this going to happen or not?

I think so. The real push here is towards simplicity and convenience. And what needs to happen for true thin clients is something that was impossible in the 1990s and is still not possible here in 2020 depending on where you live: Truly pervasive and truly fast wireless Internet access. Remember that the promise of 5G, as originally outlined by Qualcomm, was that it would be faster to run an app from the Internet than to load it off a disk locally. And the fanciful story that you’d be standing in line to board an airplane, suddenly remember that you forgot to download a movie for the flight and could do so then, in just seconds, thanks to the magic of 5G.

I told this story on Windows Weekly, but for those who don’t listen to the podcast, I’ll just repeat it here. I’m reviewing a Lenovo Flex 5G which, as its name suggests, comes with a 5G SIM card, in this case from Verizon. The problem is, I live in Pennsylvania, and Verizon doesn’t have any 5G coverage in this state, let alone where I live. So I see 4G speeds here, which is fine, but certainly not life-altering.

My friend Rich Woods writes for Neowin and he is also reviewing the Flex. He messaged me last week to ask about whether I was able to get 5G speeds, and mentioned that he was going to take the train into Manhattan (from Long Island) so he could see what 5G was really like. And I saw later that he posted a shot of the speed test: He got 711 Mbps (down), I believe. That’s amazing, but the 5G promise is even faster.

To put that in perspective, my home wired (cable-based) Internet connection is 330 Mbps down. So Rich saw speeds of more than double that, wirelessly and out in the world. With that kind of speed, a movie download in 1-2 minutes (or whatever) is a reality. If that were pervasive—i.e. literally everywhere—it would change everything. Even that 7 GB of local data sync that I mention above wouldn’t be necessary: I could use a truly thin client and not worry about local data storage at all.

And that’s just the beginning: Samsung this week started talking up 6G networking, which could hit speeds of 1000 Gbps. That sounds fantastical and it is. But it’s not even necessary: if 5G just happens and happens basically everywhere, we’re set.

So Sun and Oracle definitely on to something in 1996. But they were way ahead of their time. (And not to give too little credit to Mr. Ellison, but of course he hated PCs: He made server products and wanted the focus to be there and not on less expensive client computers.)

And AnOldAmigaUser also asks:

Joining the ranks of those asking about thin clients, do you think there might be a version of 10X that would fit that bill? Perhaps if the Win32 bits are only loaded if necessary?

Some of the conversations above relate to this question, but yes, a simpler UI like the one planned for Windows 10X would certainly make sense on a thin client. In fact, one might view Surface Neo and other presumed devices like it as a step in that direction and away from more full-featured PCs. Windows 10X without Win32 support would certainly work today, and with remote desktop/apps and 5G+ speeds, there’s another compatibility solution to consider too.

This kind of reminds me of the type of conversation one might have about programming languages, and how some of them are referred to as “high level” (meaning more human-readable) and some are “low-level” (meaning machine-readable). Back in the day, machine language was low-level and assembly was high-level. But over time, assembly was considered low-level and C was high-level. Etc. With thin clients, type of hardware we’d think of as thin today would be impossibly powerful for a user in the mid-1990s. (And a thin client from then would be laughably unusable today.) I mean, a Surface Go with 4 GB of RAM and eMMC storage is sort of a thin client. Not exactly but sort of.

UK’s Huawei ban

Usman asks:

Wanted to get your thoughts on the recent Huawei 5G ban in the UK, Huawei is still allowed to be used in 4G equipment and this ban is specifically targeted at 5G so to me it just doesn’t add up. Is this still motivated by the tariff agreement that Huawei didn’t sign and USA using it’s muscle in the five eyes alliance?

Yes. It’s important to understand that the UK didn’t suddenly find evidence that backs the US claims that Huawei is a threat to national security. That evidence does not exist. Instead, the UK determined that the US ban on Huawei and its inability to do business with US component firms means that Huawei will not be able to fulfill the demands of the UK’s 5G infrastructure. It’s kind of a neat trick: It can show its allegiance to a major ally without having to publicly support its unfounded claims against Huawei.

I remember you previously mentioning ZTE, that once ZTE paid the tariffs, they were able to sell their Android phones. ZTE is another 5G vendor but they’re not under any scrutiny considering they’re also a Chinese company. I’ve looked at news articles, I always see the framing of China being an intelligence risk, but no mention of ZTE.

Our unilateral attack on Huawei has always confused me. There are all kinds of Chinese companies that the US government could be going after. If it really believes Huawei is evil or whatever, then they all must be. But yeah, they’ve only gone after Huawei. I suspect it has to do with its market power: Huawei is/was number one in networking infrastructure (and is/was number two in smartphones, though that’s just collateral damage and not a primary target).