Transforming the PC Landscape: Microsoft’s Vision for AI Integration with Copilot+ PCs

Microsoft continues to focus on innovation across platforms to unlock everyone’s full potential, with AI giving time back in the day to be more creative, more strategic and more innovative. On May 20, the technology giant announced how they are revolutionizing the PC for the AI era with Copilot+ PCs. By unleashing hybrid AI with Copilot+ PCs, Microsoft is extending their vision for AI across the Windows ecosystem. This session with Windows + Devices CVP Pavan Davuluri provides a closer look at:

  • The integration of AI in Windows,
  • How Copilot+ PCs can transform user experiences, and
  • Why NPU is needed for Copilot+ PCs launching this month and for future innovation across the ecosystem.

Transcript

Daniel Newman:
For this next session, we had the great honor of filming it while we were with the Microsoft Surface team in Taipei. Signal65 President Ryan Shrout and I had the pleasure of sitting down with Microsoft’s Pavan Davuluri. He’s the Corporate VP of Windows and Devices, and we talked to him about how the company is integrating AI into its products with a focus on its new and exciting Copilot+, and of course, the overall PC portfolio. There’s no doubt that AI is transforming the PC landscape. We saw it on display at Computex. We saw it recently at Microsoft Build, and in this conversation, Pavan is going to talk to us all about that. Tune in, watch it now.

Pavan Davuluri:
Thank you both for having me, and we are grateful to be a part of this Copilot+ journey, for sure, and we are absolutely grateful for the partnerships across our ecosystem to bring them to life.

Ryan Shrout:
It’s really been an impressive last month of things moving forward with this. The first question I want to ask you is really about your ambitions for AI on a larger scale with Microsoft. Microsoft’s been integrating AI into data center, into client for a while. We’ve seen the original version of Copilot come out there. With Copilot+ you’re taking that to the next step, but I’m curious if you can give a high-level view with as much detail as you want. What’s the next wave? How do you see user experiences fundamentally changing because of this AI adoption?

Pavan Davuluri:
It’s a great question, Ryan, and you’re right. When we talked about it at Build a couple of weeks ago at this point, Satya referenced this as really year two for us in the AI journey. In many ways, I think we’re at the start of what is probably going to be a multiyear, decade-long perhaps evolution, transition in the space. For us, I think what we are amazed and excited about is it’s happening at a pace that we had not seen, or quite frankly, perhaps not expected.

To answer your question, Ryan, I think there’s a couple of places where Copilot+ PCs and AI and Windows and devices take themselves. I think one thing we are learning and seeing is a desire for having Copilot and agent experiences show up more pervasively in the operating system. It was literally in January that we started with the idea of a Copilot key, and now we’re in a place where we’re talking about a more fully supercharged OS that is powered by models running real-time, all the time.

I think the first thing we’re going to see is, at the shell and user experience level, we’re going to continue to see Microsoft trying to meet customers where they’re at in the flow of activity and bring AI capabilities into task flows, and just have the operating system be an agent for taking them through those flows in a seamless manner, as seamless as we possibly can. I think you’ll see us continue to innovate on the Windows UX itself to take advantage of AI, and just have AI be a more pervasive part of the experience.

The second big piece, of course, is the Copilot+ PCs, all the excitement and energy here in the last couple of weeks. One of the powerful things that Copilot+ PC brings is the new platform capabilities with the hardware itself. It allows us to start running a class of models on device, and that in itself brings a new class of capabilities to the operating systems themselves. We’re very excited about the fact that the OS will have new superpowers, as will apps that are going to show up especially from Microsoft in the near term.

The third big thing is the thing that we announced at Build called the Windows Copilot Runtime. The Windows Copilot Runtime is important for us because it really gives developers a namespace to target to start building AI capabilities into their apps and experiences, and for us from a Microsoft standpoint, to provide them a platform where it’s a spectrum, from they can build their own models and capabilities to just take advantage of inbox models and capabilities. I think the three of them together are going to fuel, I think for the next year or two, a bunch of new experiences on these devices going forward.

Daniel Newman:
Yeah, the developer is so important, but I gotta ask you, Pavan. I need you to be the arbiter a little bit here. Microsoft has pretty much set the standard of what is a Copilot+ PC, and there’s also talk about AI PC, and these things are not exactly the same. Can you, for the audience out there, just talk a little bit about what it entails to specifically be branded as a Copilot+ PC?

Pavan Davuluri:
Yeah. Sure, Daniel. I think maybe we should start with AI PCs, because I think we are seeing a strong signal with AI PCs right now in the market. With what we started with in January, AI PCs in my mind, certainly they have the Copilot key. The Copilot key is a gateway to a set of OS capabilities that are going to show up that are agent-driven, that are AI-driven, that are Copilot-driven for us. That’s an important attribute for AI PCs for us, for sure.

AI PCs in themselves have NPUs of varying degrees in size. One great example where AI PCs are actually celebrating NPU capabilities right now are things like StudioFX, for example, Teams, Zoom, WhatsApp. All use camera and audio and microphone stacks for AI in real time today with these AI PCs. AI PCs are more performant, and I think of AI PCs as the start of the journey, and Copilot+ PCs take us to another level of capability.

Copilot+ PCs, Daniel, certainly have a bunch of silicon attributes that define a performance and capability standard, for sure, but Copilot+ PCs is really a platform experience where, yes, the silicon is a huge part of it. We, Microsoft, had to do work around the operating system itself to unlock these capabilities in the silicon.

It was a lot of work with our OEM partners for making sure. In addition to memory and storage, we were doing a bunch of work at the platform level because at the end of the day, customers want to see a set of experiences. You want to deliver these in a way that is a compelling, complete set of capabilities. Copilot+ PCs really embody all three elements of silicon capabilities. They are well-defined and tailored for this quantum leap in compute, in my mind, a set of OS work that celebrates those OS capabilities, and then our OEM partners doing a bunch of system and platform work that completes the device experience and the system solution for us.

Ryan Shrout:
You brought up some of your OEM partners earlier. I talked with one today in this very room where we were looking at how does the PC ecosystem itself fundamentally change, and how much of it is driven by AI versus how much of it is just coincidental with AI. When I think through the idea of maybe a PC doesn’t look like a PC anymore in three years or five years, I don’t need a keyboard, I don’t need a touchpad, I don’t need a mouse …

Pavan Davuluri:
Great question.

Ryan Shrout:
… how do you view that shifting? Right now, everything’s kind of a laptop that has some added capabilities, but I think long-term, that could change a lot.

Pavan Davuluri:
Yeah. I think this is a space with a ton of opportunity, and I can see a variety of different things playing themselves out. What I would tell you, there are two things that have been helpful for me. First is I actually think we have a long run with actual capabilities with the current form factors in themselves. I think people are going to be more surprised by the way of what is possible than what they see today with their form factors.

The example that I use for our own journey within Surface, for example, is the Flex keyboard on the Surface Pro. AI is increasingly going to become multimodal. Yes, it looks like a chat, a prompt-input-response thing out of the gates, but really as models get multimodal, touch and pen and ink and voice are all options. Just the idea of detaching the keyboard from the screen, allowing you to use both pen and touch and ink at the same time, opens a set of app experiences that we had not conceived of even two years ago.

I would say they are, relatively speaking, in the form factors that we know and love today, for instance. That’s the first thing, is I think you’re going to see a lot more value in the current form factors. It’s actually exciting because like, Daniel, we talked about, those developers I think are going to target them out of the gates and push the envelope for what’s possible.

The second point that you’re making, which I think we are in the infancy of really appreciating what this is going to look like, is I think the ability for us to have these models do things that are not possible otherwise for us. We oftentimes think about new capabilities, and a platform shift in compute is really what drives innovation in form factors. It’s this virtuous cycle between what’s possible in compute, what’s possible in form factor, and what’s possible in interaction and software design. I think that cycle is in its early cycles when it comes to AI-driven devices or AI-first devices in themselves. I’m very excited about the latter as well, I think. I think a lot of what’s happening on PC and phone will in fact spur what those new devices can look like.

Daniel Newman:
I want to talk about something that’s a little bit more baseline in this whole Copilot+. It’s power. We’ve got a handful of apps, very powerful, exciting apps, that are going to make these new devices super cool to use. Having said that, I think a lot of people are still, “How do the developers build? What are they going to build?” Part of the whole Copilot+ is about efficient. It’s about bringing that mobile experience that says we can go 16 hours, and you can watch videos and be social and use your apps and your thing’s going to go. How do you balance that, stay super performant, but at the same time meet those efficiencies?

Pavan Davuluri:
That’s a great question, Daniel. I think, going back to your earlier question, I think we start with a customer orientation on this. I think people, they have an expectation for the devices to be just great devices, so they want them to be instant-on. They want them to be able to have all-day battery life. They want them to be responsive, at the end of the day. I think all of those attributes require a platform that is indexed on energy efficiency. I think that is an attribute for us. When we think about modernizing the platform over time, it is going to continue to be important.

One of the things that we value with the Snapdragon X series really is the focus on energy efficiency for the NPUs, because in turn, what makes those app experiences delightful is the fact that the AI is available on a high-performance basis and an always-on basis when you need it. The step function improvement that you get with always-on, background-running capabilities is really fueled by the ability to have a tremendous amount of energy efficiency. We talked about our launch event having like 100X the efficiency for running AI workloads. Those step functions are hard without entirely new neural architectures and NPUs, and the ability to be efficient with matrix math fuels that.

I think the next step really will be growth on both those vectors. I think we’ll continue to push on be more energy-efficient, so we can then reuse that currency in a variety of ways for devices and experiences. We will push the envelope for what is possible compute-wise, because I think hopefully we will see people finding value in leading in it and generating momentum for it.

Ryan Shrout:
I think the technology is critical for that. It’s an interesting outlook of the roadmap and how all this goes forward. I want to ask a little bit on the commercial side of this.

Pavan Davuluri:
Sure.

Ryan Shrout:
We talked at the beginning how Microsoft is playing in AI in all of these spaces, client devices, data center, services, everything. How do you see Copilot PC’s, AI PCs in general, come into the Microsoft commercial strategy? How do you monetize this? How do you make it interesting for partners and ISVs to come join the group?

Pavan Davuluri:
Yeah, it’s a really good question. A clear line, I think the future is going to look like distributed compute for us. You’re seeing huge investments with Azure, for sure. You’re seeing large language models get more performant. It’s a breathtaking pace of innovation in that space, and I think we truly believe in scaling laws, bringing new capabilities through the cloud. At the same time, I think we will move into a world of an efficient frontier. You see small language models showing up. You see capabilities on devices in themselves. That in turn creates things like Copilot+ PCs.

At the end of the day, I think what will drive us is a world that looks more of a hybrid pattern, where we will build apps and experiences that are the most efficient use of AI and its capabilities both from a performance standpoint, from a cost standpoint, from a privacy standpoint, a security standpoint, and that will in turn dictate what does that hybrid pattern look like across both the cloud and the client. I think the business models will look like a version of what we have today, an involved version based on what subscriptions and consumption and device capabilities look like.

Daniel Newman:
Yeah, I definitely give a lot of credit to Microsoft for its abilities to go left to right, at least that’s how I explain it, like really from the edge to the cloud. There’s a lot of what I would call marketecture about that, but you’ve really accomplished that and you’ve done it with things you own, things you build, things you share, democratizing, and of course, this is a great example.

I’m going to thread a needle here, because you do it every day, Pavan. Threading the needle is here you are, celebrating with all of your OEM partners that they’re building these wonderful devices. Microsoft also has worked very hard, and I think it’s underneath your tutelage, building Surface, and Surface builds some wonderful devices. I use them every day, personally.

Pavan Davuluri:
Great to hear.

Daniel Newman:
Big fan. You’ve launched some exciting Copilot+ PC experiences in the Surface portfolio. I’m going to give you a time-out from trying to be the ecosystem person. I want you to talk to your book a little bit. Talk a little bit about the Surface and how you see that evolving, because you gotta be excited. You’re glowing.

Pavan Davuluri:
Thank you.

Daniel Newman:
You’re glowing.

Pavan Davuluri:
Yes, we are. We’re super excited about them. We are. I think our collective team is very jazzed. It’s been a multiyear project with the Qualcomm team, for sure. We announced two new products recently. Both devices are on preorder right now. Customers can buy them today. They’ll be available generally on June 18th, and one of them is the Surface Pro. The Surface Pro, it is a modern definition of our view of what two-in-ones can look like.

It is by far the most powerful two-in-one we have built. It has the best battery life of the products we have created so far. The Flex keyboard we talked about earlier is delightful, in my mind, because it brings a set of new capabilities. It celebrates Copilot+ experiences in a way that’s not possible. It has a new, gorgeous OLED HDR display that I love. I use that device, in fact. Thank you for your usage of those Surface devices. I certainly use a Pro all the time.

Daniel Newman:
I only talk my book when it’s true.

Pavan Davuluri:
That’s lovely. We also have a new laptop, a brand-new laptop, which is a completely redesigned Surface laptop, which also pushes the envelope for us by way of performance and battery life in itself. I love using both those devices, for sure. The one thing with that laptop that I find particularly compelling is, as my work machine, I’m in calls and meetings and connecting with team and folks a bunch. A lot of the features that we’re building in the product, we get to use them, whether it’s live captioning type stuff, or camera stacks and audio stacks and studio effects. Our Teams team is building on them. Zoom builds on them. WhatsApp is using those features right now. We’re super excited about both those devices.

Ryan Shrout:
I’ve been using them for a little bit as well. We did some testing on them with Signal65. It was a great experience. Pavan, I want to thank you for taking some time and coming out to talk with us today. I know you’re super busy, you’ve got to run to a bunch of different spots, so thanks for coming on.

Pavan Davuluri:
Absolutely. It’s been a pleasure. Thank you.

Other Categories