AI Continues to Advance Toward Edge Devices, Multiple Chip Giants Eye AI PC Market
-
With ongoing industry discussions and refinements in the development and commercialization of large AI models, the transition of these models to edge devices is becoming increasingly evident. Recently, multiple leading manufacturers worldwide have announced their initiatives and strategic plans for the AI PC industry.
The PC industry has shown signs of recovery in the third quarter, with analysts predicting that generative AI-equipped personal computers will mark a turning point for the industry, potentially driving higher average selling prices in the market next year. Additionally, industry insiders believe that PC operating systems with AI capabilities will introduce new interaction models, potentially stimulating fresh market demand. Generative AI also opens up innovative possibilities for software and operating system applications.
Multiple Chip Giants Focus on AI Applications for Edge-Side PCs
The increasing enhancement of software functionalities through AI and machine learning algorithms has positioned AI PCs as a pivotal moment for the PC industry. The Science and Technology Innovation Board Daily notes that several major computer chip manufacturers have recently announced their support for the development of the AI PC ecosystem.
Intel recently launched its "AI PC Acceleration Program," aiming to bring AI features to over 1 million PCs by 2025, starting with the Intel Core Ultra processors set for release on December 14. According to Intel, the program seeks to collaborate with independent hardware and software vendors, leveraging Intel's AI toolchain, co-creation resources, hardware, design expertise, and market opportunities to maximize the performance of AI and machine learning applications. This initiative aims to attract broader industry participation in the AI PC ecosystem.
Similarly, Qualcomm unveiled its X Elite chip for PCs and laptops during the Snapdragon Summit. Qualcomm executives stated that laptops equipped with the X Elite chip will hit the market next year. The chip has been redesigned to better handle AI tasks such as summarizing emails, drafting text, and generating images. The Snapdragon X Elite is specifically optimized for AI, supporting on-device execution of generative AI models with over 13 billion parameters. It boasts 4.5 times the AI processing speed of competitors, outperforming Apple's M2 Max chip in certain tasks while being more energy-efficient than both Apple and Intel's PC chips.
AMD has also introduced its Ryzen AI engine, based on Xilinx IP, for Windows platforms within the Ryzen 7040 series PC processors. This engine accelerates machine learning frameworks like PyTorch and TensorFlow, handling up to four concurrent AI streams and processing INT8 and bfloat16 instructions. AMD claims the engine is faster than Apple's M2 neural engine.
PC manufacturers are also accelerating the launch of their AI PC products. Lenovo Chairman and CEO Yang Yuanqing showcased the company's AI PC for the first time at the Lenovo Tech World Innovation Conference, declaring that "the personal computer is entering a new dawn." Lenovo plans to release its AI PC by September next year.
According to Lenovo, AI PCs can create personalized local knowledge bases by leveraging model compression techniques to run individual large models, enabling natural AI interactions. These devices are tailored to serve as intelligent productivity tools, enhancing efficiency, streamlining workflows, and safeguarding personal data privacy.
IDC has noted that while use cases are not yet fully defined, market interest in this category is strong. "AI PCs can personalize user experiences at a deeper level while ensuring data privacy. With a wave of AI PC launches next year, overall selling prices are expected to rise significantly."
AI Unlocks New Innovation Potential for PC Software and Operating Systems
While computing power underpins generative AI and large models, Intel emphasized during its AI PC Acceleration Program announcement that software leadership may be the key to delivering the AI PC experience.
Canalys highlights that, in addition to chip manufacturers' efforts, enhancements in AI functionalities within new operating systems will also drive the rollout and adoption of AI PCs, accelerating growth starting next year.
Industry insiders told The Science and Technology Innovation Board Daily that operating systems, as the foundation of the digital era, provide the platform for all computer software. "Large models, as a form of software, also run on operating systems. In the future, AI will become a fundamental capability of operating systems, with general-purpose large models exploring more application scenarios through OS integration."
Microsoft introduced Windows Copilot, making Windows 11 the first PC platform to offer centralized AI assistance. Copilot, an AI assistant based on GPT-4, helps users with tasks such as document drafting, coding, and image design, offering suggestions and error corrections based on user input.
Canalys predicts that as x86 architecture continues to enhance PC AI capabilities, a new wave of AI-enabled models will emerge in the first half of 2024. By Q4 2024, shipments are expected to reach approximately 20 million units, accounting for over 25% of global PC shipments.
With Microsoft set to release a new generation of Windows in late 2024, featuring AI upgrades and widespread integration of AI tools in business and productivity software, the AI-compatible PC market is poised for explosive growth in 2025 and 2026.
Notably, domestic PC operating systems are also aligning with industry trends by integrating large model capabilities and exploring productivity transformations.
Deepin, operated by UnionTech, recently announced that it has become the first open-source OS to integrate large models, launching UOS AI. UnionTech UOS is not only empowering AI through underlying XPU drivers, runtime optimizations, and AI framework support but has also collaborated with major model partners to embed large models into the OS. UnionTech plans to further explore AI-native applications, natural language interaction compatibility, data security, and other technical aspects to build the next-generation OS and innovation ecosystem.
"Under its current leadership, Microsoft has shifted its strategy toward the cloud, aiming to integrate the OS, cloud, and AI into a unified product suite," said Zhang Lei, Senior Vice President and CTO of UnionTech, in an interview with The Science and Technology Innovation Board Daily. However, cloud-based AI deployment currently faces challenges such as cost and security. UnionTech UOS AI aims to explore both cloud and edge deployments.
Edge-side large models face challenges such as chip computing power and ecosystem deployment. "Chip manufacturers like Intel and NVIDIA are promoting new AIGC capabilities in their products, driven by market demand. Meanwhile, edge-side large models and software require more powerful computing devices," Zhang noted. He described the relationship between chip and software vendors as a "two-way street" in terms of commercialization and technical collaboration. Overall, "the explosion of generative AI has unlocked boundless innovation potential for software and operating systems in the next phase."