GenAI stands out as a key focus in the 2024 supply chain. As AI gradually transitions from cloud to edge computing, it also drives the emergence of terminal applications such as AI PCs and smartphones. Despite multiple brands launching PCs and smartphones with AI concepts, the software and hardware integration process, and consumer acceptance of these terminal devices, remain to be tested.
During the CES 2024 expo centered on AI-powered PCs, several tech giants such as Acer, Asus, and Samsung Electronics unveiled applications leveraging Microsoft's Copilot system. According to DIGITIMES analyst Joyce Chen, Microsoft is leading the peripheral supply chain in deploying AI applications.
The future of AI PCs
The introduction of AI-driven PCs is anticipated to breathe new life into the once-stagnant PC industry. However, there is still ambiguity surrounding the definition of AI PCs within the supply chain. Chen explains that AI PCs refer to CPU platforms equipped with NPU capabilities, broadly categorized as personal computing devices capable of offline AI processing. Currently, the optimal parameter range for offline execution of LLM models ranges from approximately 10 billion to 30 billion, with Microsoft's Copilot leading the coordination of hardware platforms.
The resurgence of AI PCs undoubtedly began with the emergence of ChatGPT, sparking the GenAI craze. Following this, the two major CPU camps, x86 and Arm, each launched new products to capitalize on the AI PC market, with PC brands such as Dell, Acer, Asus, and MSI also joining the fray.
Chen's analysis indicates that Microsoft leads the way in operating system advancements. Supply chain sources indicate PCs must possess 40 TOPS of computing power to run Copilot offline, which coincides with the market's anticipation for the new version of Windows to integrate Copilot.
In addition to Microsoft, Google and Apple have also been active recently. Apple's unexpected move on March 4, 2024, to release 14-inch and 16-inch MacBook Airs equipped with the M3 chip, underscores the company's intent to join the AI PC fray.
Analysis of Apple's release suggests that the M3 MacBook Air enables users to achieve real-time speech-to-text transcription, translation, text prediction, visual understanding, and accessibility features, with support for cloud-based solutions such as Microsoft Copilot, Canva, and Adobe Firefly.
Chen believes that although major PC brands have successively launched AI PC products, actual market acceptance will still determine the direction of the AI PC market. Many supply chain players express uncertainty about the precise definition of AI PCs and whether end consumers will embrace them.
Upgrades of Edge AI
Amidst the post-pandemic enterprise upgrade cycle, there is an opportunity for the PC industry to return to growth with a push from AI PCs in the second half of 2024. Additionally, as genAI gradually moves from the cloud to edge applications, it also tests the performance upgrade capabilities of HPC providers.
Chen also identifies Arm as a key focus for most HPC and SoC providers in upgrading edge AI specifications. She suggests that the developments related to dedicated NPUs or other XPUs are worth monitoring, particularly considering Arm's influential position in the ongoing wave of genAI device development.
Furthermore, with OpenAI, Microsoft, Google, and Meta being set as benchmarks, the critical aspect of AI development to monitor moving forward will be the balance between subsequent model compression and the accuracy in edge AI.
