Why AMD believes AI will sell business laptops in 2024

Despite the rise of generative AI, cynicism remains about AI on laptops and PCs. What, people ask, will it actually do? Here at CES 2024, though, “AI PC” has become a rallying cry for the industry, now that NPUs (neural processing units) are coming to computers powered by the latest AMD, Intel and Qualcomm chips.

All three companies are betting big on AI PCs, so when we got the chance to interview Matt Unangst, who’s in charge of AMD’s Commercial Client PC and Workstation business unit, we were keen to find out his strategy. In short, how would he convince businesses – from SMBs to enterprises – that they needed local AI compute on their laptops and PCs?

We asked Matt to summarise in one minute why he thought 2024 would be a great year for the business

AI on business PCs gives you scale

His first argument is a question of scale. “When we think about generative AI and the types of models and the processing that’s required, up to this point, it has largely been relegated just in the cloud,” he said. “But that’s not going to scale. It’s already scaling so quickly, that it’s almost untenable to continue to deliver those experiences to a large number of people.”

AMD’s strategy, then, is to be present everywhere that AI may take place. In the cloud with its Instinct accelerators and on computers.

AI PCs improve your battery life

With a relentless demand for more power-efficient computing, from phones to laptops to servers, Matt Unangst was also keen to focus on how NPUs could aid efficiency.

“One of our core tenets, especially on the mobile side but even on the desktop side, is our power efficiency and the fact that we deliver this uncompromising capability of leadership performance and battery life,” he said.

“When you actually look at some of our collaborations, think about things like Microsoft Teams and Zoom, when these take advantage of our NPU you actually get better battery life on those applications.”

It may not be a radical breakthrough, Matt admitted, but it answers one practical problem. “You go talk to a lot of IT decision makers today and battery life is a real concern for them.”

Summarisation of business information

Unangst describes efficiency and scale as the “foundational part” of the AI PC story. “Part two of the story is the fact that you have an increasingly large number of still kind of immature, but rapidly maturing, use cases around some of these AI capabilities,” he said.

“One of the areas that we think is gonna be kind of an early use case for a lot of businesses, whether it’s SMB or enterprise, is what we call summarization. It’s the ability to feed a large document into an application that can very quickly summarise it.”

The idea is that employees will ask questions, ChatGPT-style, to help them find information. Perhaps at a company that has huge pools of information via documents about their products.

“At AMD [for example], we design complex microprocessors, we have design specs and large sets of documentation. All of a sudden, you have a tool that can dramatically increase the productivity of your engineers because they don’t have to sort through multiple hundreds of pages. They can literally have a conversation with their computer and quickly find the information they need.”

And, unlike when employees upload documents to a cloud-based service, organisations can keep control of their information as it is only shared locally.

Staying productive for years to come

The next argument for buying AI PCs in 2024 rather than waiting? To be prepared for the future. “Everybody understands we are in the very early stages of [AI],” said Unangst.

“Think about where we were in terms of applications and use cases six months ago versus where we’re at today versus where we’re going to be at three years from now,” he added.

“When IT decision makers or SMB leaders make purchase decisions on PCs they’re looking to buy, a PC that their employees can use and be productive with for three to four years, you want to be prepared for the future. We know [AI] is coming.”

Unangst concedes that “the use cases are still limited” but argues that it’s growing at “an exponential rate right now. You want to make sure you have a PC that is built for that future need as these applications increasingly take advantage of it.”