Recently, Micron lowered its revenue forecast due to lower-than-expected sales of AI computers and smartphones, triggering market concerns about the decline of the AI industry. However, this concern may stem from a misunderstanding of the current state of the AI market. This article will analyze the current AI market, especially the current application status of AI on personal computers and mobile phones, explore its development prospects, and clarify some market misunderstandings.
In recent years, there has been much discussion about whether the computer and mobile phone markets are in recession. Recently, graphics memory chip manufacturer Micron has lowered its revenue forecast for the next few quarters due to lower than expected sales of AI computers and smartphones. This has caused many people to worry that "AI is dying." However, in fact, AI shows no signs of declining, especially as evident from Nvidia's performance.
Many laptops and mobile phones currently on the market that claim to have AI capabilities do not have enough processing power. Even high-performance gaming PCs find it difficult to run complex AI programs like ChatGPT locally because these applications require huge amounts of data and computing power and cannot simply be completed on a PC. While there are some alternative applications available, they are nowhere near the performance and responsiveness of most server-run AI programs.
In the AI ecosystem, most of the outstanding companies and tools have become established. For example, users with Nvidia RTX graphics cards can often outperform many modern NPU-equipped CPUs in AI performance. Comparison shows that the performance difference between notebooks equipped with RTX4080 and Intel Core Ultra9185H under AI workloads can reach 700% to 800%. It can be seen that the server plays a key role in providing AI performance.
Google has extended its AI model Gemini to most Android devices and plans to bring it to Nest speakers. Even though these devices are four years old, they still demonstrate the broad applicability of AI technology. Looking back in the past, graphics card performance was once considered to need to reach tens of billions of calculations (PFLOPs) to achieve a true virtual reality experience, and current graphics cards have not yet reached this standard, which also reflects the challenges still faced by the development of local AI.
In the development process of GPU manufacturers, AI programming often relies on parallel computing, and GPUs are superior in this regard. Therefore, future GPU design still takes time, and it may not be until the launch of the RTX60 series that significant AI performance improvements are seen. This generation of graphics cards may make it possible to run local large models (LLMs).
Highlights:
AI technology is not dead and market performance is affected by misconceptions.
Many devices that advertise AI rely on servers for performance, making it difficult to implement complex operations locally.
Future technological advancements in GPUs may drive the development of local AI models.
All in all, AI technology has huge potential for future development, and its current market performance does not reflect its true strength. With the continuous advancement of hardware technology, especially the improvement of GPU performance, local AI applications will usher in new development opportunities and ultimately change people's lifestyles.