For more than a decade, many advanced computing tasks have depended on remote servers. Personal computers increasingly served as access points to cloud platforms where much of the processing took place.
A new generation of AI PCs is beginning to challenge that model. By integrating neural processing units (NPUs) into computers, devices can perform AI tasks locally rather than relying entirely on cloud services. The change could prove particularly significant in regions like Africa.
“First of all, you’re getting power and compute on device, and therefore not relying on connectivity wholly,” Craig Booth, Acer GM for the UK, SA and Nordics, told Gadget last week on the sidelines of the Acer EMEA press conference in Milan, Italy.
“Connectivity and access to the cloud are ideal, but with NPU processing power on the device, it’s not essential. You can still do a lot of productivity and have a lot of capacity on device to do things remotely without access to the cloud. In Africa particularly, that’s an important item and something that will really be an enabler for people.”
Booth said the processing power will allow organisations to run analytical workloads directly on the device rather than relying on cloud services. The same capability could support language translation locally, which he said could be particularly valuable in Africa, where tools that convert or transcreate content between languages can improve communication and productivity.
Creative work is following the same shift as devices gain more AI capability.
“Content creation will be possible locally rather than in the cloud. There will be lots of productivity outputs that, until now, the computing power hasn’t been there to do on the client. It’s getting there, and it will be. As we’re showing with Acer devices, there’s already a lot of on-device capability emerging, including things like image generation.”
For businesses, running AI tasks locally could change how digital tools are accessed.
“I think potentially there are things you’re going to be able to do on device where you won’t have to go down the software-as-a-service (SaaS) route to access those services. You’ll be able to do them directly on the device.”
Booth told Gadget that the architecture of AI PCs may influence energy use and the lifespan of devices.
“Power output and potentially longevity of device will help reduce total cost of ownership and power consumption. With an NPU, it’s not that it needs more power to provide the trillions of operations per second, it’s just using power in a slightly different way. And actually is, in the end, using net less power.”
For now, Booth sees AI primarily as a productivity tool rather than a replacement for human work.
“Some people talk about AI as an alternative way of getting things done, where people don’t necessarily have to do the work anymore. particularly I think of AI at the moment as assistive, as improving productivity. You may take a task today that takes an hour, and with AI capabilities you can turn that task into five minutes. It just means you can get more done. We’re all looking to get more done. We’ve all got more to do than we’ve got hours in the day.”
Small and medium enterprises could find practical value in those capabilities, particularly where enterprise software platforms are expensive to deploy.
“Let’s think about inventory control, which, at the moment, is difficult. You might have to outsource to a SaaS to have some kind of enterprise resource planning (ERP) system for inventory control. But you probably could have an AI assistant on your computer, locally, doing that work for you, and not necessarily needing access to the cloud.”
The rapid development of AI hardware is already reshaping the PC market.
“We are already seeing a democratisation of the technology. We’re already seeing a democratisation of the price that makes AI computing more accessible to more people. If you talk to Microsoft, you know they’ll already be talking about 40% market penetration in certain economies where the devices are AI capable or ready devices.”
Affordability remains central to how quickly the technology spreads in Africa.
“The average selling points of devices in Africa tends to be low relative to the global average or to the regional average.”
As AI capability moves into more devices, Booth expects it to become standard across the market.
“We’re going from a world where 10% of devices on offer are either AI capable or AI ready. To now, in our own estate, for devices for this year, 50% of everything we sell will be AI capable already. Computers without AI capability or AI readiness are going to be unusual and exceptional in the market very quickly.”
* Jason Bannier is a data analyst at World Wide Worx and deputy editor of Gadget.co.za. Follow him on Bluesky at @jas2bann.