Recently, AMD held a grand AI PC innovation summit in China, providing a comprehensive overview of AMD's technological advancements, plans, and advantages in the AI era, showcasing the widespread applications of AMD AI across various industries.
Dr. Lisa Su, Chairperson and CEO of AMD, graced the summit with her presence, delivering a keynote speech on AMD's AI strategy and global trends. Following the summit, she also sat down for an exclusive interview.
Reflecting on her visit to China, Dr. Lisa Su remarked that AI is a rapidly evolving field filled with new trends and developments. Hence, she expressed her desire to come to China and organize an event specifically for the Chinese ecosystem.
Trends in AI PC Development: Hardware and Ecosystem Are Key
For the current hot topic of AI PC, some voices claim it's just a marketing concept, suggesting AI as a topic has become overly hyped. However, Dr. Lisa Su bluntly disagrees, stating she doesn't see AI PC as merely a marketing concept.
She points out that in reality, we're all seeking "killer apps" that can drive PC upgrades and usher in new cycles. She genuinely believes that AI will be the catalyst for growth in the PC market.
From the interview with Dr. Lisa Su, it's clear that the concept of AI is appropriately hot right now. This is because everyone recognizes the immense power of AI technology and wants to get their hands on it as soon as possible. People are experimenting, learning, and within AMD, AI is being used internally to help manufacture better chips for the future and enhance various aspects of company operations.
Regarding what exactly constitutes an AI PC, Dr. Lisa Su proposes that from AMD's perspective, an AI PC, or what they call Ryzen AI, should ideally combine CPU, GPU, and NPU. Particularly considering that certain models can run on any of these processors.
For AI PC, AMD aims to optimize AI compute power in terms of TOPS (tera-operations per second). Hence, they typically provide a TOPS figure, such as the Hawk Point processor offering a total compute performance of up to 39 TOPS, encompassing all engines.
As time progresses, each generation of NPU will become increasingly powerful. The significance of NPU lies in AMD's desire to extensively employ AI in PCs while also achieving good battery life. Therefore, the efficient execution of AI operations per watt TOPS by NPU is a crucial metric.
Speaking of the importance of ecosystems, it's crucial because the development and widespread adoption of any new product or technology depend heavily on the rapid formation and refinement of ecosystems, especially in the case of AI PCs.
Dr. Lisa Su noted that while the traditional PC ecosystem has been running for many years and is quite mature, the ecosystem for AI PCs is just beginning to take shape. Every month, the number of AI PC applications and users is rapidly increasing. She believes that there will be many different types of applications emerging and the same will be seen across various industries, whether it's in finance, productivity applications, content creation, or gaming. Many new and different use cases will emerge within the ecosystem.
In my view, currently, AI PCs have already made some significant applications in improving collaboration experiences and enhancing local productivity experiences. Most local training and inference can be done on the PC, which, compared to constantly calling upon the cloud, can undoubtedly provide users with a better experience.
AMD has set a target to have over 150 ISV software partners using Ryzen AI technology by the end of this year. Dr. Lisa Su added, "I believe we will actually exceed this target because there are so many new companies emerging both in China and abroad with new applications."
Dr. Lisa Su believes that PC demand was very high during the pandemic, but it has slightly declined post-pandemic. However, she is confident that the emergence of AI PCs will drive growth in the PC market, with some growth expected in 2024.
For the AI PC category specifically, sustained growth is anticipated in the coming years because the fundamental capabilities of AI PCs will prompt consumers to upgrade their PCs. Despite being in the early stages, 2024 is deemed a crucial year.
While most AI PCs are positioned in the higher end, Dr. Lisa Su expects to see AI PCs appearing in every price segment as time progresses. Over the next few years, there will be significant developments in hardware. On the software front, numerous applications are already being implemented, with more expected to emerge in the coming months.
AMD AI Advantage: The True Trifecta
When it comes to the development of AI and AI-powered PCs, having robust hardware and ample computational power are fundamental prerequisites. This provides optimal support for various AI models and applications.
In this regard, AMD undoubtedly holds a significant advantage. Compared to its competitors, AMD possesses a balanced and robust trio of CPU, GPU, and NPU, catering to both data center and consumer-level markets.
Dr. Lisa Su has pointed out that depending on specific use cases, some may heavily rely on CPU, others on GPU, and yet some on NPU. Consequently, AMD's product lineup includes these three types of processors, representing a significant advantage.
For instance, concerning laptops, the importance of NPU is expected to become more prominent over time. Each generation will likely become more powerful, considering the power and battery constraints inherent in laptops. Conversely, desktops, workstations, and the cloud have more power headroom, ultimately achieving different optimization points, which is the core competitiveness of AMD's product lineup.
The Ryzen 8000G series marks the first comprehensive AI engine for desktops, and when combined with laptops, AMD sees it as a continuum of various systems, enabling AI capabilities to permeate throughout AMD's product portfolio.
Regarding AMD's core competitiveness, Dr. Lisa Su believes that it's crucial to use the right engine for the right application. While AI primarily exists in the cloud, which is undoubtedly important, it also extends to the edge and client-side. Therefore, concerning AMD's product portfolio, having engines suitable for all these domains is the fundamental advantage of AMD.
Looking ahead, AMD expects to see more interaction between the cloud and the client-side. By providing engines on all types of devices, AMD aims to deliver a more optimized experience to users.
Beyond hardware, Dr. Lisa Su remains confident in AMD's ecosystem because of the significant investment and progress made in the ROCm ecosystem capabilities.
In my opinion, AI developers typically don't often code at a lower software level; they prefer using higher-level frameworks like PyTorch and TensorFlow. Hence, AMD has invested heavily in these areas to ensure that developers using AMD's platform achieve excellent performance.
In the author's view, the curtain has only just been raised on the era of AI-powered PCs, with fierce competition and intense developments underway.
AMD undoubtedly stands at the forefront, especially with its comprehensive coverage across all fields and scenarios, leveraging the three major computing engines of CPU, GPU, and NPU. With strong support from numerous industry partners in both hardware and software, AMD has firmly grasped its core competitiveness.
Through this innovative summit and the dialogue with Dr. Lisa Su, we have witnessed AMD's wholehearted dedication to AI, particularly in the AI PC domain. This is not only inspiring but also highly anticipated.
There will be challenges along the way, but with perseverance, we can navigate through rough seas. Here's to looking forward to AMD and the entire AI PC industry bringing forth even more excitement and innovation!
