AI PCs Are Going Mainstream Says AMD’s Jason Banta

Generative AI is quickly entering the PC market. Recent Omdia research found AI PCs to be a leading growth driver for AI edge hardware.

In a Q&A with AI Business, Jason Banta, vice president of product management at AMD, discussed the future of AI in PCs, highlighting its potential to enhance productivity and collaboration.

We’re talking about AI PCs. Explain to those unaware of how AI fits into the PC space.

Jason Banta: AI PCs are evolving; I would say that AI experiences are evolving as well. In the broadest sense, an AI PC is a PC that is purpose-built to handle and run AI models in an optimized, efficient way. AI is a technological capability that produces experiences that make people more productive, enhance their collaboration and make their overall PC experiences better on the back of large language models, generative AI models and AI software. The AI PC is built from the ground up to run generative AI models efficiently from a performance and battery life perspective. If you look at non-AI PCs compared to an AI PC, the experience you're going to get on those AI workloads and experiences is going to be much better. And that is where we at AMD are going and that is where the PC industry is going as a whole.

Microsoft created a dedicated Copilot key that brings up AI features – the first major change to keyboards in three decades. How important was this move in terms of bringing AI to general consumers?

Related:Edge AI Chip Market to Hit $60B by 2028 as Small Models, PCs Boost Demand

We think it’s important because it indicates the level of AI integration the entire ecosystem is working toward. We have a deep partnership with Microsoft, and they are predominantly an independent software vendor (ISV) in the market that is leading a lot of what’s happening in the AI PC space. We are continuously working with the folks at Microsoft to bring new features and capabilities to the market for consumers and business professionals leveraging AI locally. And then we’re also doing something very unique, which is opening up our software such that everybody can get access to it. One of the most important aspects of bringing AI to consumers is ensuring developers can experiment with the software and bring their ideas to the market as well.

What market are AI PCs likely to penetrate first? Is it consumers? Businesses?

AI PCs are simultaneously being adopted in every major customer segment.

For users in the content creator market who run Adobe, AutoCAD or other content creation applications, AI is enhancing their efforts by embracing generative content creation and being able to predict and optimize workflows.

Related:Microsoft Adds AI to PC Keyboard – First Change in 30 Years

For business PCs or the PC that you are issued at work, what is being optimized is around collaboration, so from transcriptions, understanding key points from meetings and contextual search to finding pieces of data to help get things done.

On the consumer side, we’re seeing a lot of workloads and AI applications being applied for things like gaming.

Every segment has been unique in the way it leverages large language models but ultimately it is about making those experiences better. The place where I think the impact is a combination of the size of the impact and how close it is to us is probably going to be the commercial market. Seeing that an AI PC can cut four hours of drudgery from my day so I can apply myself to solve bigger problems is going to drive consumers to migrate to AI PCs faster.

What about a dedicated developer unit? Could we soon see devices tailored specifically for the growing developer space?

It’s happening. We're seeing that both from the perspective of our CPU offerings as well as our GPU offerings that people want to build upon our Ryzen AI neural processors. We released a developer suite for that and started developer contests so people could experiment with building different types of applications

Related:AMD, Nvidia to Bring AI to Consumer PCs, CES 2024

I think of the developer community as our first audience when it comes to Ryzen AI. Our original equipment manufacturer (OEM) partners are very inspired by this as well. They see developers as their first audience and many mobile workstations and AI-based gaming PCs are being picked up by developers.

What about the gaming market? PC gamers spend thousands of dollars on high-end graphics cards – what tangible benefits can AI bring to this section of the market?

The gaming market is a very adept space – people in that market generally understand the technology well. It's a market where you have to prove what you are offering. There are marketing buzzwords in that space, but in general, gamers tend to eschew pure buzzwords and see right through the hype.

From the perspective of AI, the gamer market is looking at AI and what AI can do and a lot of things are starting to emerge in that space. For example, in-game content creation and real-time content creation are some of the things video game developers are looking at. Whether that’s level generation or character generation, leveraging AI for that makes a lot of sense.

There’s also AI frame generation, where you can see higher frame rates or a higher quality image in a video game via AI. Those capabilities are also emerging.

From a CPU and GPU perspective, we have great relationships with game developers and game engine developers so we have direct access to that space. As we pilot new AI capabilities, neural processors or AI on the GPU, we are sitting right beside those game engine developers and those game developers.

In terms of AMD, the new Ryzen 7 8000 series desktop processors have a neural processing unit to power AI workloads on programs like Adobe. Talk us through how AI is becoming embedded into your product line.

We have a few main tenants for our Accelerated Processing Unit (APU) enablement strategy. The first cuts across the PC space and that is our deep partnership with Microsoft. Our partnership with Microsoft is paramount to enabling great experiences and features using our NPUs.

The second main pillar of this strategy is working with large software developers like Adobe, Zoom, Topaz Labs and DaVinci, to create and enhance experiences. Applications like Adobe allow users to create content, well why not give that application itself the capability to generate content?

The last pillar is the independent developer. That's where the software development kits and sponsoring coding contests speak to that group. They don’t have a large following for their software suite, but they have an idea that could change the way you interact with a PC and we want to sponsor and encourage that.

One interesting use case for the new RX 7600 XTs is that they have AI-powered noise suppression to reduce ambient noise during streaming. How does your team come up with ideas for where AI can help?

We have a lot of smart people. Our engineers develop new ideas, not just to build a better processor or GPU, but to look for ways to enhance capabilities. We don’t just want to create hardware for hardware's sake. We want to bring experience with that.

Some of that is internal development and some of that is through partnership development. Our central engineering organization is in constant idea mode. For every idea we come out with, we throw away six or seven.

There was a time when a majority of AMD’s engineering team was hardware developers and a small minority were software developers. It's almost half and half now. We've staffed up software and user experience developers dramatically because we want to show how these experiences can be brought to life.

AMD isn’t the only big-name hardware developer eying the AI PC space – how do you keep competitive?

Competition is good and competition is fun – it keeps you in good shape and that is how we see it. We released Ryzen AI in 2023, we saw competitors come out with their own neural solutions. That validated the move we made in the market.

Our primary competitor came out with a CPU solution with a neural processor built-in and on a performance level, it matched what we did last year. And then this year, we came out with something that was 60% more performant. Getting ahead and making those moves quickly, we're able to keep our road map ahead of the competition.

We think some folks will, as their training models, look at what we've got, what our primary competitor has and what our emerging competitor has. They're going to look at those different solutions but by staying innovative, keeping that cadence of our road map and developing high-performance, low-power solutions, we're confident we are going to be able to deliver an AI experience. But our goal is not to just deliver an AI experience, we want to deliver the best experience and that means faster responses from AI assistant workloads, faster responses from generative AI requests and longer battery life. We know everybody's going to have an AI experience but we want to deliver the best experience.

What’s to come? Can we expect a maturing of AI PCs by the time the next CES comes around?

AI PCs are going to be all over the CES 2025 floor. I do not doubt that. When we did our CES announcement, we previewed some stuff that will be coming later this year. We did not open up on everything but we talked about our XDNA 2 technology. We have Zen for CPUs, RDNA for GPUs and our neural processing technology family is called XDNA. Our Ryzen AI 8040 series is based on XDNA 1. We previewed our XDNA 2 at CES technology which has over a three times performance improvement relative to what we just released.

We are in development and in partnership with our OEMs and software developers to get that solution ready for the market. Later this year and early next year, you are going to see the quantity and the capability of AI grow dramatically. It is a space that is not going to slow down. It felt like everybody was talking about AI PCs at CES this year. I think everybody is going to be talking about AI PCs next year and they're going to be talking more about some of the very cool experiences that are going to be coming around the corner because it's going to get much more tangible.