PC is currently at a turning point with the advent of AI PC. With a combination of CPU, GPU and Neural Processing Unit (NPU) – productivity, creativity, gaming and more can now be enhanced with AI locally with incredible efficiency.
So imagine how a few lines of instructions in PowerPoint can help create a visually stunning presentation in just a few seconds.
Some might say that they are already doing this through a browser on a laptop that is three years old. That might be possible, but older PCs take longer to process, consume more energy, cost more to send data back and forth between the cloud and the PC, and it gets complicated when you’re dealing with sensitive data. who cannot leave your location or country.
These problems are magnified in enterprise environments. More and more employees are using AI applications in their daily work; more and more companies must train or fine-tune their AI models with proprietary data; And note that many enterprise software such as database management applications have licensing models that charge companies based on the number of CPU cores in the cloud used to run the application.
With AI PC, these devices can optimize the running of AI workloads, resulting in better utilization of hardware resources.
Imagine how much faster and more cost-effective it would be if companies could run many of these AI applications directly on employees’ PCs, without having to continually pay fees for cloud computing. The potential for reduced operational costs, increased efficiency and productivity can result in significant business benefits over time.
Having the ‘Edge’ in the AI Era
In addition to data centers and PC AI, more and more AI will move to the “edge.” The edge includes applications for the Internet of Things (IoT), autonomous vehicles, and devices for smart cities, which will complement everyday AI experiences.
Computing for the edge requires processing data ‘around’ or at the edge of a network, closer to the location where the data is generated, rather than relying on centralized data centers.
The need for edge computing is critical in the AI era. First, it enables real-time processing of critical matters where decisions in a few seconds can have an impact on safety, such as automation in industry.
For another, processing data locally will reduce the volume of data sent to the cloud, which will reduce ‘congestion’ in the network, cut data transfer costs, and increase security by minimizing exposure of sensitive data during transmission.
Finally, when internet connections fail, edge computing ensures mission-critical applications continue to function – vital for the healthcare industry.
AI use cases that utilize trained machine learning models to make predictions or decisions based on new input data are called inferencing.
In contrast to training, which often requires a more demanding computing infrastructure to support it, inferencing can be more easily performed at the edge via a common computing server, with familiar hardware, lower power consumption, and the flexibility to survive in challenging environments. different.
In fact, IDC estimates that by 2025, as much as 75% of the data generated by enterprises in the world will not be generated and processed in traditional data centers or the cloud, but at the edge.
It’s important to note that not only will there be more AI and computing at the edge, core workloads will also experience inferencing.
Think about how many people “build” weather models versus how many people “use” weather models. It’s training versus inferencing, and the latter will take over a large portion of AI workloads in the future. Knowing this will help companies prepare the right computing infrastructure for the future.