With the upgrade of AI PC's hardware and software, the semiconductor industry, including CPU and DRAM, is experiencing a significant boost.

The personal computer (PC) industry is gradually heating up, reaching a new peak with the integration of artificial intelligence (AI) technology. In this technological revolution, Intel and Microsoft, two pioneers of the PC revolution, are actively promoting PCs equipped with AI CPUs and intelligent software assistants, successfully extending the application of AI from the cloud to the personal computer domain.

In other words, AI PCs with embedded dedicated chips can run AI models locally without relying on the cloud. Intel CEO Pat Gelsinger stated that this would make AI services cheaper, faster, and more private compared to using cloud-based data center services. He said at CES 2024 in Las Vegas: "In the future, you will unleash this power for everyone, every use case, and every location."

What kind of CPU does an AI PC need?

Due to the development of AI, traditional CPUs can no longer meet user demands, and the cost of separately configuring GPUs is a significant challenge for ordinary users. Currently, Intel, AMD, and Qualcomm are actively positioning themselves in the AI PC chip market, constructing a new generation of AI processors.

Advertisement

So, what kind of CPU does an AI PC require?

AI PCs demand high-performance CPUs because they need to process large amounts of data and complex computational tasks. Therefore, the CPUs in AI PCs must be capable of high performance to quickly handle AI algorithms and large datasets. AI computing typically requires parallel processing of multiple tasks, so the CPU needs to support multi-core and multi-threading technologies. This allows for the simultaneous processing of multiple computational tasks, enhancing overall computing power. While maintaining high performance, the CPUs in AI PCs also need to have low power consumption characteristics to extend battery life and maintain system stability.Additionally, some modern CPUs have integrated AI acceleration units. For instance, Intel's Meteor Lake laptop CPU incorporates an NPU to support third-party AI software functionalities. These acceleration units are specifically designed to handle AI-related computational tasks, enhancing the speed of AI inference and training. The current trend among manufacturers betting on AI PCs to adopt a CPU+GPU+NPU architectural approach is not coincidental but rather a well-considered choice based on the current technological developments, market demands, and application characteristics.

This architectural approach fully integrates the advantages of various processors. CPUs excel at sequential control and low-latency tasks, capable of handling smaller traditional models; GPUs are better suited for parallel tasks with high precision formats, such as video and games with extreme quality requirements; while NPUs are specifically optimized for neural network computations, efficiently executing inference and training tasks for neural networks. By combining these three types of processors, AI PCs can simultaneously meet a variety of complex computational needs, enhancing overall performance.

The CPU+GPU+NPU architectural approach also helps to address the challenges of generative AI applications on the edge. These applications are constantly evolving and diversifying, demanding higher computational requirements. Single hardware deployments often struggle to meet these stringent and diverse computational demands. Therefore, adopting heterogeneous computing has become key for hardware manufacturers to tackle these challenges. The synergistic work of CPUs, GPUs, and NPUs can fully leverage their respective strengths, providing more powerful computing capabilities and more efficient energy utilization.

As AI technology rapidly develops and becomes more widespread, the market demand for computing devices with AI capabilities is also growing. AI PCs with a CPU+GPU+NPU architecture can meet this demand, offering higher performance and a richer array of application scenarios. This also brings more market opportunities and competitive advantages for manufacturers.

Intel states that starting with Meteor Lake, it will widely introduce AI into PCs, leading hundreds of millions of PCs into the AI era, with the vast x86 ecosystem providing a broad range of software models and tools.

AMD, Intel's competitor in the PC processor field, is also launching processors that support AI. NVIDIA recently showcased new GPUs suitable for AI laptops during a virtual event, indicating that the rise of AI PCs has garnered attention across the entire PC hardware industry.

AI PCs: Starting with 16GB of RAM, computational power exceeding 40 TOPS

Beyond processors, memory chip manufacturers such as Micron, Samsung, and SK Hynix are also actively investing in the development of AI PCs. Their goal is to enable AI accelerators to run more powerful assistants on personal computers. Currently, new laptop models are equipped with up to 8MB of RAM, and in Windows-based AI PCs, this memory capacity may double. With the advent of more powerful AI accelerators and processors, the memory requirements for AI PCs are expected to increase further. Taking Meta's Llama 2 series AI model as an example, its standard version requires nearly 30GB of RAM, and the hardware requirements for AI PCs may gradually surpass the current standard memory capacities.

Windows will once again play a key role in promoting the acceptable minimum memory capacity in PCs, which should be good news for memory manufacturers, creating a new significant demand for memory.Additionally, the expectation for a PC to run the Microsoft Copilot AI assistant smoothly also depends on sufficient local acceleration. Microsoft believes that at least 40 TOPS of computational power is required, which can be provided by a dedicated GPU. Nowadays, almost all so-called AI PC processors come with efficient neural processing units built-in, capable of meeting this computational performance target.

Intel emphasizes that AI PCs will become a turning point for the rejuvenation of the PC industry and will play a key role in the industry highlights of 2024.

Currently, consumer enthusiasm for AI personal computers is still in its infancy and will take some time to wait for the emergence of more native applications. However, as time goes on, both the hardware and software of AI PCs will undergo more powerful upgrades, which undoubtedly brings exciting news for semiconductor devices such as CPUs, GPUs, and DRAM.

*Disclaimer: This article is the original creation of the author. The content of the article represents his personal views. Our reposting is solely for sharing and discussion and does not represent our approval or agreement. If you have any objections, please contact the backend.