Artificial intelligence (AI) is coming to the personal computer (PC) market. AI companies, computer manufacturers and chipmakers need to find profitable applications for generative AI technology. These organisations have been scrambling of late to find a way to make their technology profitable. Now, they may have struck upon a way to push the technology from controversial curiosity to mainstream commodity.
Increasingly, a lot of the returns from the (eye-wateringly) big bets on AI made by companies like Microsoft and Intel look like they might come from AI-enabled PCs.
What is an AI PC?
Essentially, an AI PC is a computer with the necessary hardware to support running powerful AI applications locally. Chipmakers achieve this by means of a neural processing unit (NPU). This part of a chip contains architecture that simulates a human brain’s neural network. NPUs allow semiconductors to processes huge amounts of data in parallel, performing trillions of operations per second (TOPS). Interestingly, they use less power and are more efficient at AI tasks than a CPU or GPU. This also frees up the computer’s CPU and GPU up for other tasks while the NPU powers AI applicaiton.
An NPU-powered computer is a departure from how you use an application like Chat-GPT or Midjourney, which is hosted in a cloud server. Large language models AI art, video, and music tools all run this way and place very little strain on the hardware used to access it. AI is functionally just a website. However, there are drawbacks to hosting powerful applications in the cloud. Just ask cloud gaming companies. These problems range from latency issues to security risks. Particularly for enterprises, the prospect of doing more on-premises is an attractive one.
Creating an AI PC brings those AI processes out of the cloud and into the device being used locally. Running AI processes locally supposedly means faster performance, and more efficient power usage.
The AI PC “revolution”
AMD was indeed the first company to put dedicated AI hardware into its personal computer chips. AMD’s Ryzen 7040 will be the first of several new chipsets. These chips have been built to accomodate AI application and are expected to hit the market next year. Currently, Apple and Qualcomm have made the most noise about the potential of their upcoming chips to run AI applications.
Recently, Microsoft announced a new line of AI PCs with “powerful new silicon” that can perform 40+ TOPS. Some of the Copilot+ features Microsoft is touting include an enhanced version of browsing history with Recall, local image generation and manipulation, and live captioning in English from over 40 languages.
These Copilot+ PCs will reportedly enable users to do things they can’t on any other consumer hardware—including the first generation of Microsoft’s AI PCs, which are already feeling the pain of early adopter obsolescence. Supposedly, all AI-enabled computers sold by manufacturers for the first half of the year are now effectively out of date as AI applications become more demanding and both hardware and software experience growing pains. Windows’ first generation AI PCs, specifically, won’t be able to run Windows Recall, the Windows Copilot Runtime, or all the other AI features Microsoft showed off for its new Copilot+ PCs.
“This is the biggest infrastructure update of the last 40 years,” David Feng, Intel’s Vice President told TechRadar Pro at MWC 2024. “It’s a paradigm shift for compute.”
AI computers will dominate the enterprise space
The potential for AI computers to enhance efficiency and deliver fast, reliable AI-enhanced productivity tools is already driving serious interest, particularly from enterprises. AI PCs will supposedly have longer battery life, better performance, and run AI tasks continually in the background. According to Gartner VP Analyst Alan Priestley, “Developers of applications that run on PCs are already exploring ways to use GenAI techniques to improve functionality and experiences, leveraging access to the local data maintained on PCs and the devices attached to PCs — such as cameras and microphones.”
According to Gartner, AI PC shipments will reach 22% of the total PC shipments in 2024. By the end of 2026, 100% of enterprise PC purchases will be an AI PC.
- Data & AI
- Digital Strategy
