Microsoft made some unusually major moves ahead of its Build developer conference: It announced a new Copilot+ initiative for powerful AI PCs, which will be led by the new Surface Pro and Surface Laptop. These machines are powered by Qualcomm's new Snapdragon X Plus and Elite chips, and they come with a special version of Windows 11 optimized for Arm mobile chips and AI. Basically, Microsoft is doing for PCs what Apple did with its M-series Macs four years ago.

In this bonus episode, Devindra chats with Pavan Davuluri, Microsoft's head of Windows and Devices, about the new Surface devices and the Copilot+ PC initiative. We still don't know how well these new machines will perform, but it sounds like Microsoft has certainly heard our complaints about Arm-based Windows devices.

Listen below or subscribe on your podcast app of choice. If you've got suggestions or topics you'd like covered on the show, be sure to email us or drop a note in the comments! And be sure to check out our other podcast, Engadget News!

Subscribe!

• iTunes

• Spotify

• Pocket Casts

• Stitcher

• Google Podcasts

Transcript

Devindra: Hey everyone, this is Devindra here. I had a chance to chat with Pavan Davuluri, the head of Microsoft Windows and Devices, basically the team in charge of Surface and Windows. And we talked about the new Copilot Plus Surface PCs, the Surface Pro and the Surface Laptop, and the whole new Copilot Plus initiative in general.

It's kind of a big move for Microsoft. We've reviewed quite a few of the ARM based Windows PCs and you know, they have not worked out so well. So I think this could be different, at least from the benchmarks we've seen. We still need to test these things, but I think Pavan is also aware of Microsoft's own issues around this kind of hardware and they're aware that this seems like a big push for them and a good opportunity to kind of shift to a mobile platform, just like Apple did.

So anyway, here's my chat with Pavan.

Devindra: Pavan, thank you so much for speaking with us, how do you feel about these new devices and Copilot Plus as an initiative?

Pavan: I'm excited. It's been a multi-year journey for our team. So I'm excited to share. I think the work that we've been on for some time now. And I'm also excited because I think it's the start of a journey for us.

So we had an opportunity to tell our story, bring a bunch of product and value out there and then I'm excited to see what people will do with it.

Devindra: Awesome. I remember when there were rumblings about this I think for the past year we were hearing that Microsoft was doing something maybe with ARM chips again and with Qualcomm and we saw the new Snapdragon benchmarks I think last fall.

And we started getting a little excited and also a little worried, because personally, I have like a love or hate relationship with Windows on ARM, I reviewed the first Surface with Windows RT, it was, it was a worthy attempt, and then most recently we've seen the Pro X devices and all those things, and recently I did the Pro 9 with 5G, and even then I was like, I don't, I just had a lot of issues with it.

So, what did you guys learn from your past experiences with Windows on ARM that you brought into this one?

Pavan: That's a good question. I do remember your thoughts on them at the time, and we actually did pay attention to them yeah, as a team, collectively. There's a couple of things I think that we learned that were meaningful.

One that I think we've addressed in the conversation. Current generation of the Snapdragon X parts that Qualcomm team has delivered. Performance matters. And so one of the foundational investments we've made in much more performance CPUs, both in terms of the CPU energy efficiency, but also peak performance of these CPUs.

We love multithreaded performance because it matters to the operating system, but we wanted to make sure we had good sustained and peak performance. I think that's a significant improvement from our previous generations that we, you know, we learned about. The second big thing was the work that we did on what we're calling Prism, the new ARM emulator.

We, we certainly learned through that journey that emulation performance matters both in terms of the efficiency of the emulation itself and the breadth of catalog compatibility with emulation. So that was a big lesson learned. I think we've made good progress. The third thing really and hence it's been a multi year journey, is making sure we have great performance.

Native arm app experiences for us. And so we have worked across the breadth of the windows ecosystem to go deliver on great app experiences, and we are very excited that our collective ecosystem partners across a range of top app categories. Now, those apps are native. And so whether you look at you know, Disney and Spotify had one in the spectrum, it'll be photoshopped or, you know, your most used browsers.

I think at this point in time, Mhm. Certainly they're all native. Our ecosystem is certainly committed. And then the big thing for me was the biggest developer on Windows is Microsoft. Microsoft bringing native apps has been a huge thing for us. And so I'm super proud of the Edge team having a native version now for us.

The M365 app estate really is native. A bunch of our security products are native. Your inbox apps, your calculator is native. And so, yeah. I think those are big lessons learned and be a problem to bear.

Devindra: Gotcha. I will let you know, like, I'm personally, maybe I'm weird about this, I'm more interested in what's going on with Windows on ARM than the AI stuff.

I think the AI stuff is cool, has a lot of potential, but we've lived through Windows on ARM for a long time. Like, oh, this is finally here, you've got good battery life, you've got a lot of stuff. Can you tell me, like, these improvements, will they trickle down to the existing devices? Will the Surface Pro 9 5G, the Pro X, will they get benefits from these changes to Windows?

Pavan: Some benefits will trickle down to those. Some are going to be platform dependent, and so a lot of the work that we did was really are learning to make the OS and the silicon platform a deeper, more synergistic vehicle for delivery. And so, in those instances, the OS improvements are tied to the platform themselves, but not all, some of them.

And we continue to look at ways of finding ways Find a way to, you know, bring them back to our broader ARM install base but I think we're going to prioritize the Snapdragon X series going forward. Gotcha.

Devindra: Is there a new name for what you're doing with this Windows? Microsoft has been very, very big on different Windows names, like RT and we, we've gone through a couple.

Are you guys calling this anything differently? Referring what this Windows on ARM is versus the old stuff?

Pavan: It's Windows 11. Okay. And in, you know, in my mind's eye, there isn't, there's only one flair of Windows, it's Windows 11. And that's true for consumers Devindra, it's true for our commercial customers, it's true for developers, in my mind, you know, importantly.

So we don't want people to have to think about all of that, we want them to be able to build apps for us to be predictable in certain set of ways that matter to them, and then unlock the ability to build devices and, you know, people have choice on the kinds of products and platforms they want to build.

Devindra: Gotcha. And these first batch of devices, the Surface Pro, interesting, you guys aren't using the numbers anymore. I see Surface Pro 11th generation is how you're referring to it in documents, and the new Surface laptop, these are Qualcomm powered machines right now, but we're also, you talked with Intel, you talked with AMD, they were here, or they at least said that they have hardware coming out.

Can you tell us how that will fit into the Surface lineup, or are we just thinking about Qualcomm Surface devices for now?

Pavan: Our partnerships with AMD and Intel have been longstanding indeed. Certainly for the Windows ecosystem at large, the operating system itself, but also on Surface. And I think we will make decisions on what silicon we use in Surface products based on how do we best build a system that serves end customers and certainly represents innovation in the Windows ecosystem.

And we look forward to working with all of them.

Devindra: Gotcha. Yeah. And you guys also have a lot of partner hardware coming out, which I was surprised to see. Normally, when this happens, you have your devices and you're like, Hey, everybody, follow along. But you have them now. You have all the major manufacturers.

You have new machines coming. Can you talk about the process of basically coordinating that? Because I feel like that makes Copilot Plus feel like a true initiative and not just Microsoft putting a flag in the ground saying, Hey, everybody follow us, right?

Pavan: You nailed it. I think that is exactly the intent.

And that is the, I think first I will say I am Deeply grateful for the partnerships across the windows ecosystem. So you saw the major partners out there today. They have been with us again on a multi year basis on this journey. For sure. I think for me, having been on the windows team and in the past in the surface team for some time, I think the level of partnership we have seen from them and from us is the thing that I have not seen before.

I guess it's probably the simplest way to describe it. They have been great in terms of Making sure we were building product with. the value prop that actually mattered and delivered to them. That was, you know, a great place up front where they were part of the product making journey with us. We've had a deep co engineering cycle, because building these great systems at the end of the day does require both the hardware platform, the silicon, certainly the operating system, app layers to all be equipped.

This is sort of the idea of the system configurations that they talked about. And so we went through, you know, a whole generation of co engineering with them together. And then now, when it comes to going to market, and telling our story, and landing our customers through it, I find that we are deeply aligned together.

Because I think, to your point, whether it's performance, and, you know, fundamentals that we can go deliver on, or unlocking new AI things, it's a thing that, you know, it captures all of our imagination, and all of our mind share. It's a thing I think they're as excited about as we are. And you're right. It doesn't, I don't think it's a meaningful exercise across the industry if our partners are not there with us.

I'm actually just grateful that our collective team, our marketing team specifically our field team, we were able to just, and engineering, of course, you know, pull them all together. It was great.

Devindra: I think it's kind of interesting because Apple kind of did this four years ago, too. Like, we all covered the, the M series chips and everything, but they have the advantage you guys don't, right?

Like, they own the hardware, they own the software. They could be like This is a big ship, but we're just turning. We can just do it. You guys can't. You have to work with your partners to kind of coordinate all that. How long have you guys been working on, like, making this transition?

Pavan: To your point earlier ARM itself is not a new construct for us.

This particular, and we've been at it for many generations, as you know, this particular iteration has been a few years for us. I'm, you know, I'm almost, I'm drawing a blank and where do I draw the line or the starting line for the exercise? Because some of these things start as a sort of an organic, you know, thought process.

Devindra: I feel like the Pro X was this, was the point where you guys were like, Hey, here's a premium surface that's thinner, lighter, you Here's a vision of what the surface can be and then kind of the design follow that. So were you thinking about it by that point? That was three or four years ago.

Pavan: Yeah. Yeah, we were definitely thinking about it in that window of time.

Devindra: Yeah.

Pavan: Yeah. I would also say the big thing in addition, I mean, the thing that we really learned, I would say 2020, 2021, is thinking of it from a system standpoint and then adding AI as a first order construct from Silicon through, you know, platform level components, the OS to the shell layer to app experiences.

Yeah.

Devindra: I think something I say often, and don't take this too harshly, but I think timing has never been Microsoft's strong suit. I think Microsoft tends to deploy things, maybe a little too early. I saw, I was, I was growing up when web TV was a thing. I remember pocket PC devices. I remember Windows Mobile.

I had a Zune. I had multiple Zunes. So, I The timing, it seems interesting how quickly you guys jumped on AI last year with Bing search and everything and with the launch of co pilots, it does seem like you guys are kind of taking a leadership, I guess, role in timing for AI because everybody kind of had to respond to what you were doing.

Can you talk about just how that feels? Does it feel different to you for how Microsoft is approaching new technology?

Pavan: It does. It does feel different for sure. I personally find it Pretty exciting and energizing at the same time. It's a little humbling as well I think to your point there is we certainly want to be in a place where we have enough composition and you know meaningful value and completed execution, you know quality of the product itself. And so it certainly puts pressure on that to make sure we are showing up in a leadership time frame with a great product But it is a different feeling for sure and I'm kind of excited about it because I do think we have a bunch of things we have learned in the last 18 months with the starting with the Bing work that happened last year.

Through the M365 co pilot work that has happened and now we're calling the Windows co pilot runtime on device models and Windows 11. Where we have a ton of great lessons learned across the company that is a kind of a flywheel and accelerator for us. Responsible AI is one good example. That team has learned a lot.

They are now an integral part of how we think about it. Same thing applies for app teams. You know, they're collectively in an AI first world. And so it's a lot easier for us to go orchestrate work across the company.

Devindra: Can you talk more about the idea of responsible AI and how you guys are thinking about it?

Because the Recall feature seems very cool. Seems like something a lot of people could use. And then I think, like, oh man, this is, the way people are worried about people snooping in their browser history, right? There's the meme, if I die, please delete my browser history. Stuff like that. And recall is just like, oh, you've created this thing that will see everything we're doing online.

How are you guys thinking about the usability aspect of something like that, and also the data privacy aspects of it?

Pavan: Yeah, it's a, it's a really good question, and I think it's important. For us, thinking about earning permission and trust through, you know, security and privacy, I think, was front and center in our mind as product makers as we were building that feature, Devindra.

If you, you know, when you start playing with these Copilot Plus PCs, you'll see that as you go set up the PC for the first time, we are very deliberate in taking customers through a user experience, you know, onboarding journey that makes it very clear for them on what that experience is like. And And making sure we're educating them and then giving them control is fundamental.

So in the onboarding experience. The second big thing is in the ongoing use of the product, we make it simple and intuitive so you are in control and you understand what is happening. And so I think as you get comfortable over time you know, you either give the experience more or less license. And so you're always in control.

And we do that with the task bar and keeping the feature front and center with you for recall. And the third thing, I think we want to take the stress out of making it feel like you made a decision that you can't undo. And so we give you enough control in the feature in the arc of time. And so you can, you know, delete, you know, instances from the past for instance, and so on.

So we're using a lot of constructs that people are familiar with, with, with data that it's already, you know, You know, the content rather that they're sharing. Finally kind of an important decision we've made. For those PCs is the data, the semantic index is stored locally on your PC. And that is a thing that we feel confidence in being able to stand by.

And I think at the end of the day that is a foundational component of saying we're not moving that data set to the cloud, we're not training on it, for instance. There's no other framework that has programmatic access to it. And so so I think those are, you know, first step in the journey for sure.

But those are some core components in making sure we meet those expectations.

Devindra: How are you guys thinking about recall when it comes to like multi user systems or shared systems? Because, hey, if everybody's using a different account, then problem solved. That's pretty easy. Everybody logs in. But for a lot of family computers, it's just kind of sitting there and people run up to it and do their work.

How does recall work in a shared user environment like that?

Pavan: It follows a user account. And so to the extent there's multiple users, you have multiple indices. If there's one account that's shared by multiple people, that's a shared index. Hmm.

Devindra: Gotcha. I guess I think that makes sense. Can we talk about the devices specifically?

So first of all, interesting that you guys kind of did the step back from the numbers just Surface Pro and Surface Laptop. It always feels interesting when a company does that, like you're kind of. Stake and claim. This is different. Yeah, point in time there. The actual hardware doesn't seem that different.

You know, the Pro I've always liked is a, like, very sleek, light device. Thinner bezels on both. Are there, are there hardware changes that you want to bring up? I know OLED is now an option on the Pro. Yeah. Any in the quad cameras, the quad HD cameras, very nice. Anything else you want to highlight in the new models?

Pavan: First of all, your point, I think your observation is right on the money for us. We think co pilot plus PCs are the start of a new generation. And so we did want to put a marker for these devices represent that new class for us. And, and like you said earlier, it's not just Surface. There's a whole ecosystem that believes in that.

One of my favorite personal features that I love on the Pro is the detachable keyboard. I do love that. That It was, you know, when we talk to customers, the number of times people ask us for that feature without quite knowing that they're asking us for it yeah, yeah, the design team in my mind, sort of, you know, going through intuitive understanding of how people work that flexibility and freedom is probably the Kind of the iconic thing that we've introduced this time around that has been in the work for some time.

It's not easy because you want to deliver on great battery life on that little keyboard. It's gotten thinner, it's gotten more stiff and so you still have to have a battery and have great battery life on it. It has to be reliable, you know, because it's your keyboard experience. On the laptop there are a bunch of things that I love.

You, you observation on the bezels getting thinner and lighter. I love bezels, yeah. Thin bezels, yeah. Thin bezels. I continue to love the 3x2. I feel like in the form factor it gives me the most bang for buck in terms of just workable real estate on the machine. I personally love the the user physical design of the product with ports and accessibility.

One of the things that is awesome about Windows is, I think, what people expect by way of I. O. and port interop, and so, on my laptop, at least, I have multiple 4K monitors on the screen. That is a thing I can kind of go after. So there's lots.

Devindra: I do want to talk about the Flex keyboard, because it seems like, again, Microsoft does this thing where it's like, ah, so close.

So close. It's a great idea. It is $350. And that is just... you're killing me. And then you have the bundle at $450. Okay, fine. I can accept that, although I think that's a little too expensive. But I also know you guys are keeping the existing Pro covers around. At their existing prices. This felt like a great opportunity for Microsoft to at least put a keyboard with the Surface.

And I think, Okay. Ever since the Surface RT, I feel like I've been asking this question every year for the past 10 plus years. I'm going to keep asking it. What is the thinking about just putting a keyboard, like including a keyboard with the Surface? Because I think about these things differently than the iPad.

An iPad is a tablet. You're going to use a tablet first, maybe occasionally with a keyboard. You guys keep talking about the Surface as a PC, as a Windows machine. Yes. Nobody wants a Windows machine without a keyboard. So, I don't understand the justification of not at least bundling something with it.

Pavan: You know, so first, your dataset comment is absolutely right.

We, we have, you know, Most customers attach a keyboard to to the device itself. One thing we do have is flexibility of choice in the keyboards themselves, and so bundles and attach wise is freedom and flexibility across different configurations of keyboards and such. But I hear you on the desire to have them attached.

You are not alone, Devindra. And I'm sure the team will find a way, as best as we can, to make that happen when we can. The opening price point comment you made, I think that's a great point as well. And we are looking to learn through this iteration. We'll look at where customers are, where we are with demand, how the Flex Keyboard performs across countries, and We'll certainly account for it as we go forward.

Devindra: I would, I definitely can't wait to see that drop in price. It just feels like whenever you guys talk about surface pricing in particular, it's like, well, yes, that's the tablet. So, yes, it's $999, but you have to add another $140 or $180 or $350 for the Flex to actually make it a useful computer. So, I feel like that just fudges with the way we talk about pricing around these things.

So, I'm hoping Microsoft is just aware. And I feel like, I don't know, I feel like it would be easier to talk about these things with consumers and to sell them on it if they didn't have to think about like, oh, I got to add this to the price. I have to buy a Surface Pen. I have to do all this other stuff.

It seems like the Surface experience has always been a little too complicated rather than You buy a MacBook, you know,

Pavan: The $999 comment versus a MacBook. Specifically, the surface laptops are $999 is a full laptop experience. And it is a, you know, I think, in fact, there's more in that laptop feature set wise, performance wise, touch screens and so on.

That is an easy AP compare. You're right. Surface Pro is a unique two in one device for sure. And you do need the attach for the peripherals. Yep.

Devindra: Okay. So we're here. We're also talking about Copilot Plus and all the AI stuff. I'm wondering, you know, we seek recall. We see Windows Studio FX and you guys are showing off some partner stuff.

How useful is that within the next year? How useful are the AI features for people within the next year on these existing machines?

Pavan: The Copilot+ machines. Yes. Okay, great. So you're gonna, if you're at Build tomorrow, you're going to hear a lot more about us talking about the breadth of the ecosystem there.

Let me maybe think about it in a couple of slices. So obviously Copilot Plus PCs now The operating system itself has a bunch of new AI superpowers, for lack of a better word. And those are built into the operating system, and they're powered by these models. There are a set of Microsoft Inbox apps that will take advantage of these capabilities.

That can, that'll show up in a variety of different ways. We talked about painted photos, essentially. Those are using on device image gen models, essentially diffusion class models. You saw live translation. That is a way to go think about, you know, communication. Certainly, we've made Windows Studio effects available in the Service Pro X, I think, generation of time to start with.

We've built on that. You'll, you'll see us doubling down on that. You'll get a lot more by way of real time camera and audio stacks getting enhanced and supercharged by AI. And then we introduced this notion of a Windows Copilot runtime, because we now do have a targeted environment for developers to go use to build AI apps on top of them.

In fact, I would tell you in my iteration now, I see more excitement from Windows developers, both web apps and, you know, native Windows apps, if you will. Excitement about adding AI capabilities, because we are now building the tools for them to get them access to the breadth of the ecosystem. And to do it in a way where you can either take advantage of it.

Like, you can get AI powers without having to have a bunch of, you know, AI scientists and model engineers and so on and so forth because we're building, you know, APIs and DDI interfaces that, you know, services built on top of the models that we have in box. So you don't have to be an AI person to take advantage of AI in your app.

We're also giving the other end of the spectrum. People can bring their own models. They can, you know, we have stores, infrastructure for deploying them, managing them. And so somebody wants to, you know, go build a, you know, rag vector index in Windows in the future. Absolutely impossible.

Devindra: Gotcha. I just want to take this audio file and dump it in Windows, have it edit the audio for me, get it podcast ready and transcribe it.

I want that. I hopefully, I'm hoping that comes soon. Feature noted.

Pavan: Okay. Love it.

Devindra: Thank you so much, Pavan.

Pavan: Devindra, a pleasure as always. I'm grateful for the time. Thank you.

It’s criminal that there’s been no way to play Paper Mario: The Thousand-Year Door for over a decade. The only way to experience the original 2004 Gamecube title was on that console or the Wii, which thankfully supported Gamecube discs (something that feels like a genuine miracle now). There was no Gamecube hardware support on the Wii U, unfortunately, and The Thousand-Year Door never popped up on its online store. So here we are, 20 years later, with a complete remake for the Switch. It’s fantastic, befitting a game that was already a masterpiece — it’s just a shame that Nintendo took so long to revisit the game.

Here’s some sobering perspective: I first played through The Thousand-Year Door as a senior in college, where my roommates and I made it a communal adventure. Now I’m married with two kids but I still lament the loss of Gamecube titles almost daily. Where’s Eternal Darkness, one of the best horror games ever made? Where’s Skies of Arcadia, an RPG I adored on the Dreamcast and which was later re-released on the Gamecube? I realize re-releases take work, but surely there’s an audience for these beloved titles!

Anyway, back to the remake of The Thousand-Year Door: It’s great, you should play it. It’s an easily accessible RPG for newcomers with a cute setup: Princess Peach has been kidnapped (of course), but this time it's by aliens! It's up to Mario and a group of friends — including a treasure-hunting Goomba named Goombela, and Koops, a cowardly Koopa — to save her by solving the mystery of an ancient civilization.

Like Super Mario RPG and Paper Mario before it, Thousand-Year Door is sort of a hybrid action RPG. You get to explore worlds and level up characters like an RPG, but battles also involve some responsive button mashing to keep you on your toes. A well-timed button could let you jump on an enemy's head more than once, or counter incoming attacks. It's an innovative approach to RPG mechanics that I wish more games picked up – the excellent Sea of Stars was a rare exception.

Nintendo

Paper Mario: The Thousand Year Door is also well worth revisiting for the olds with nostalgia for the original. The graphics are richer and more detailed, with sharper sprites and lighting that makes the environments feel alive (the reflections, in particular, are often stunning). The game's score has also been revitalized to feel less MIDI-like – don't worry, there's also an in-game perk that can change everything back to the original Gamecube tunes.

It’s too bad Nintendo had to lower the frame rate down to 30fps from the Gamecube’s silky smooth 60fps, but it’s not the end of the world. If you can enjoy some of the greatest games ever made in 30fps, like Breath of the Wild and Tears of the Kingdom, The Thousand Year Door is no different. The remake also adds enough new graphical elements to make it look better than the original. I’m sure I nailed the game's timing-based moves more often in 60fps, but they're still fairly easy to pull off (except for those damn counters).

Nintendo

Had Nintendo released this remake earlier in the Switch’s lifespan, I’m sure fewer gamers would be complaining about the 30fps dip. But at this point, the Switch is on its last legs and we’re awaiting news about its successor. Both Sony and Microsoft have had “next-gen” consoles out for so long they’re considering mid-cycle upgrades. It’s simply odd to see a game running more slowly today than it did on the Gamecube 20 years ago, especially when Nintendo is charging $60 for a lesser experience.

Perhaps the Switch 2, or whatever Nintendo’s new console is called, will be able to run The Thousand Year Door at 60fps. But it really doesn’t matter. It’s still a masterpiece, even at half the frame rate.

The Surface Laptop has always been a bit of an anti-revolutionary device. After Microsoft struggled to make a splash with its original Surface tablets, it was created as a more mainstream option for less courageous consumers. It simply a Windows laptop, albeit a well-designed one. 

The same is true for the new Copilot+ Surface Laptop: It doesn't look very unique at first, but spend a bit of time with it and you'll notice the attention to detail around its case and keyboard, or the way its thin new bezels highlight its brighter screen. And together with Qualcomm's new Snapdragon X Elite and X Plus chips, it's also far more powerful than before.

Developing...

Catch up on all the news from Microsoft's Copilot AI and Surface event today!

What if you could run an entire Windows PC on a mobile Arm-based chip, bringing the power efficiency and thinner designs from smartphones and tablets to laptops? If you've been paying attention to Microsoft's PC strategy over the past two decades, this song probably sounds familiar. From the original Surface in 2012 (running Windows RT for Arm devices) to the recent Surface Pro 9 5G, Microsoft has chipped away at this dream, only to fail miserably every time. Now with its new Copilot+ PC initiative, which includes major upgrades in Windows for Arm systems and AI, Microsoft may finally have the answer to its mobile computing dreams.

Microsoft's portable PC ambitions didn't start with the Surface line: You can trace it back to Windows CE and Windows Mobile-based Pocket PCs. Then there was the short-lived era of netbooks: tiny, cheap and under-powered laptops meant mainly for browsing the web. I'll admit, I loved many a netbook, but they couldn't compete with the rise of the iPhone, Android and tablets.

Timing has never been Microsoft's strongest point. While Apple can just re-orient its platforms around its own homegrown hardware and software to pull off a monumental feat, like the move towards its Arm-based M-series chips, Microsoft has to wait on its many partners. In the case of Copilot+, the program wouldn't have been possible before Qualcomm's new Snapdragon X Elite chips, or before developers were ready to build apps to take advantage of neural processing units (NPUs) for AI work.

"We engineered this update of Windows with the focus on AI and specifically AI inference on those devices, and [with] making sure we were taking full advantage of the Arm 64 instruction set," according to Microsoft’s head of Windows and Surface Pavan Davuluri in a briefing with media earlier this month. "[In] this updated Windows, we built a new compiler in Windows for this exercise. We have a new kernel in the operating system that is built on top of this compiler. We have new schedulers in Windows that are built for taking advantage of these workloads."

Davuluri also noted that there's a new driver compute model that better integrates neural engines into Windows, just like CPUs and GPUs. Those core Windows updates will be a major boon for AI hardware, undoubtedly, but they will also make the OS function far better on Arm chips than we've seen before. Microsoft says that more native Arm apps will be coming to Windows, including Spotify and over 400 apps from other developers. But the key upgrade, a new emulator that's 20 percent faster than its previous solution, and is said to be faster than Apple's Rosetta 2 emulator for M-series Macs.

"We made gains on the breadth and the reach of the emulator," Davuluri said, referring to the amount of apps that Prism works on. "When you combine the new prism emulator with simply the raw performance and improvement in [the Snapdragon X Elite] CPUs themselves, we're in a place where we have great native apps and we're also in a place where the breadth of the app catalog also has tremendous performance, comparable to the rest of the Windows estate today."

While I haven't been able to benchmark Copilot+ PCs yet, I've seen a few compelling demos that point to raw performance and battery life that’s similar to Apple's M3 chip. I'm just hoping the company can finally deliver a Windows on Arm experience that doesn't stink. After reviewing the Surface Pro 9 5G, which was slow and incompatible with many apps, I had given up on the idea of a decent Arm-based Windows PC entirely. But with revamped Surface devices, as well as partners like Dell, ASUS and HP jumping on the Copilot+ bandwagon, maybe Microsoft has finally crafted a decent mobile PC platform.

Catch up on all the news from Microsoft's Copilot AI and Surface event today!

Microsoft's new Surface Pro, its first hybrid Copilot+ PC tablet, doesn't look much different than its predecessors. It's still a sleek and sturdy tablet with a kickstand. But the screen looks a bit more impressive, thanks to slimmer bezels, and it's potentially more useful on the go when paired with the $350 Surface Pro Flex keyboard, which lets you type wirelessly. As a Copilot+ PC though, its true value lies under the hood, thanks to a 45TOPS neural processing unit in Qualcomm's Snapdragon X Elite and X Plus chips. 

Developing... 

Catch up on all the news from Microsoft's Copilot AI and Surface event today!

Microsoft's Windows on Arm woes may finally be over. As part of the company's new Copilot+ AI PC initiative, which includes new Surfaces and partner systems running Qualcomm's Snapdragon X Elite chips, Microsoft says it has rebuilt core components of Windows 11 to better support Arm-based hardware and AI. That includes a new kernel, compiler, and most importantly, an emulator named "Prism" for running older x86 and x64 apps.

You'd be forgiven for being skeptical, though. Since the launch of Windows RT-powered Surface in 2012, Microsoft has proven it can't be trusted to deliver a decent OS experience on Arm. That device couldn't run legacy x86 apps (who would want to do that in Windows, right?), and it was far slower than PCs with Intel and AMD CPUs. Windows on Arm slowly improved over the years, to the point where it had serviceable emulation on the Surface Pro 9 5G. But that slate still couldn't keep up with its Intel-equipped sibling, especially when it struggled to emulate popular apps like Chrome.

Microsoft says it reworked Windows 11 schedulers to take advantage of Arm capabilities and AI workloads. There's also a new driver compute model that recognizes neural engines more like how Windows sees CPUs and GPUs, as well as AI APIs built directly into the OS. Basically, Arm hardware should no longer feel like an afterthought and developers should be able to tap into AI capabilities more easily.

While Microsoft is pushing the availability of more native Arm apps for Windows 11, it's hard to deny the importance of supporting older software. That's where the Prism emulator comes in. Microsoft claims it's around 20 percent faster than its previous emulator, and it also worked to improve the number of apps that it supports.

Developing...

Catch up on all the news from Microsoft's Copilot AI and Surface event today!

I knew what to expect from the Surface Pro 10 for Business the minute Microsoft announced it: A faster processor with a neural processing unit (NPU) inside the same case as the Surface Pro 9. As the first "AI PC" Surface devices, the Pro 10 for Business tablet and its more traditional sibling, the Laptop 6, seemed disappointing. And the fact that you could only buy it on Microsoft's website, or through enterprise resellers, made it seem as if the company was trying to hide its shame.

But after testing the Surface Pro 10 tablet (I'll omit the "For Business" part of its name for the rest of the review, you get it), I realized Microsoft just wanted to make a PC upgrade to appease IT workers. It’s boring; there aren't any major new features to learn, and your company's existing Windows software will all work just fine. No alarms, no surprises.

I can accept this Surface Pro 10 for what it is because we know Microsoft has some intriguing consumer-focused Surface hardware on the horizon. The company has invited media to an event in Seattle on May 20th (a day before its Build conference begins), where we expect to see some consumer-focused Surface devices powered by Qualcomm's new Snapdragon X Elite chips. It makes sense for Microsoft to take a bigger risk with its consumer hardware, instead of the machines IT workers need to manage.

What's new in the Surface Pro 10 for Business?

Both the Surface Pro 10 and Laptop 6 for Business are powered by Intel's new Core Ultra chips. These are the company's first processors to include an NPU for handling AI tasks. Admittedly, there aren't many AI-driven features in Windows 11 yet, but we're expecting to see Microsoft sell some new AI capabilities at Build.

For now, having an NPU means you can use Windows Studio Effects, which can glam up your video chats with blurred backgrounds and other features. By tapping into your NPU, instead of your CPU or GPU, Windows 11 can also handle those tasks more efficiently, and without slowing down the rest of your system. Who wouldn't want that?

Design-wise, the Surface Pro 10 is almost exactly the same as the Pro 9 — the 13-inch 120Hz PixelSense screen returns, and Microsoft is still relying on its clunky kickstand to prop up the tablet. But at least the company shoved in a new 1440p webcam with a wider field of view, which should lead to clearer video chats.

What's good about the Surface Pro 10?

In spite of its lack of updates, the Surface Pro 10 is still a reliable and capable hybrid tablet. It delivered fast performance, a great display and excellent battery life. Here are my favorite things about it:

• Intel's Core Ultra 135U chip is slightly faster than the 12th-gen processors used in the Surface Pro 9: It reached 5,772 points in PCMark 10, compared to its predecessor's 4,010 points.

• That Core Ultra chip's NPU also makes the Pro 10 more future-proof. It'll be ready to tackle AI features as they appear (developers like Adobe and Audacity are already working on several).

• Microsoft's PixelSense display still looks fantastic, with bold colors and enough brightness to use outdoors in direct sunlight.

• While I would have liked to see a newer design, its case still feels shockingly slim and sturdy. It feels like an iPad, even though it's running Windows 11.

• The Surface Pro 10 lasted 12 hours and 19 minutes on the PCMark 10 Modern Office battery benchmark. That’s better than the vast majority of ultraportables we’ve seen this year, and it’s only an hour less than the latest Dell XPS 13.

Photo by Devindra Hardawar/Engadget

What's not good about the Surface Pro 10?

For a machine that starts at $1,200 though, the Surface Pro 10 is lacking in a few key ways:

• For a "Pro" device, it has too few ports for accessories.

• Its trademark kickstand is just a pain to live with now. It's still awkward to hold on your lap, and it also makes the Pro 10 impossible to use on small surfaces (like my beloved standing desk laptop riser).

• The Surface Pro Keyboard is still sold separately, and it's still far too expensive at $140 (there are also pricier models if you really want to punish yourself). Nobody will ever buy a Surface tablet on its own! Microsoft's inability to recognize this basic truth has haunted these systems since their inception.

• It may be undercut by the upcoming consumer-focused Surface Pro 10. Qualcomm's Snapdragon X Elite chips are Arm-based, so they'll have to run an Arm version of Windows 11 and rely on emulation for older apps. But their NPUs also offer 45 TOPS (trillions of operations per second) for AI tasks, while the Core Ultra chips only hit 10 TOPS on their NPUs. These differences may not mean much for the next year or so, but they may make the Snapdragon Surface Pro 10 more future proof as we see more AI features roll out.

Photo by Devindra Hardawar/Engadget

Should you buy the Surface Pro 10 for Business?

At this point, you're better off waiting to see how Microsoft's new consumer-focused Surface hardware performs before investing in either the Pro 10 or Laptop 6. But, as I said at the start, these systems aren't really meant for consumers anyway. They exist to give IT workers a way to update their inventory without having to significantly shift their workflows.

We've spent some time with the iPad Pro M4 and new iPad Air... and the iPad Pro is still a bit too pricey for us. This week, Cherlynn and Devindra chat with Engadget Deputy Editor Nathan Ingraham about his reviews and why he still prefers the iPad Air. Also, we wrap up Google I/O 2024 with a Project Astra hands on, and we chat about Apple bringing eye tracking to iPhones and iPads as an accessibility feature.

Listen below or subscribe on your podcast app of choice. If you've got suggestions or topics you'd like covered on the show, be sure to email us or drop a note in the comments! And be sure to check out our other podcast, Engadget News!

Topics

• Nate Ingraham reviews the iPad Pro M4 and iPad Air: Powerful, but pricy – 1:11

• Google I/O news wrap up – 20:11

• OpenAI’s new GPT-4o can talk, sing, and sounds a little flirty – 28:36

• Intel brings back the days of FireWire data transfer with Thunderbolt Share – 31:43

• New accessibility features from Apple: eye tracking and expanded wake word options – 35:48

• Biden administration quadruples tariffs on Chinese EVs (along with solar and other hardware) – 43:20

• Listener Mailbag: The age old question of what to do with your old tech – 55:48

• Working on – 1:01:32

• Pop culture picks – 1:05:45

Subscribe!

• iTunes

• Spotify

• Pocket Casts

• Stitcher

• Google Podcasts

Livestream

Credits 

Hosts: Cherlynn Low and Devindra Hardawar
Guest: Nathan Ingraham
Producer: Ben Ellman
Music: Dale North and Terrence O'Brien

We just wrapped up coverage on Google's I/O 2024 keynote, and we're just so tired of hearing about AI. In this bonus episode, Cherlynn and Devindra dive into the biggest I/O news: Google's intriguing Project Astra AI assistant; new models for creating video and images; and some improvements to Gemini AI. While some of the announcements seem potentially useful, it's still tough to tell if the move towards AI will actually help consumers, or if Google is just fighting to stay ahead of OpenAI.

Listen below or subscribe on your podcast app of choice. If you've got suggestions or topics you'd like covered on the show, be sure to email us or drop a note in the comments! And be sure to check out our other podcast, Engadget News!

Subscribe!

• iTunes

• Spotify

• Pocket Casts

• Stitcher

• Google Podcasts

Livestream

Credits 

Hosts: Cherlynn Low and Devindra Hardawar
Music: Dale North

While Google and OpenAI have been racing to win the AI crown over the past year, we've seemingly reverted away from the idea of speaking to virtual assistants. Generative AI products have typically launched with text-only inputs, and only later add the ability to search images and basic voice commands. At Google I/O today, the company showed off Gemini Live, a new mobile experience for natural conversations with its AI. 

Google offered up a few potential use cases; You could have a conversation with Gemini Live to help prepare for a job interview, where it could potentially ask you relevant questions around the positions. It could also give you public speaking tips if you want to research a speech. What makes Gemini Live unique is that you'll be able to speak at your own pace, or even interrupt its responses if you'd like. Ideally, it should be more like having a conversation with a person, instead of just voicing smart assistant commands or generative AI queries.

At I/O, Google also showed off Project Astra, a next-generation virtual assistant that takes the concept of Gemini Live even further. Astra is able to view your camera feed and answer questions in real-time. It's unclear how long that'll take to arrive, but Google says some of Astra's live video features will come to Gemini Live later this year. Gemini Live will be available for Gemini Advanced subscribers in the next few months.

Developing...

Catch up on all the news from Google I/O 2024 right here!