Crazy speculation on AI PCs in the US stock market! Intel, AMD, and Qualcomm join forces to challenge NVIDIA's dominant position
Intel, AMD, and Qualcomm are all embedding their own NPU (Neural Processing Unit) into their chips, posing a threat to NVIDIA's GPU dominance in AI PC computing power. However, even without an NPU, NVIDIA's GPU can still deliver outstanding performance on PCs.
As the AI PC market heats up, chip giants see their stock prices soar.
Last Friday, AMD rose more than 7%, while NVIDIA and Qualcomm rose more than 4%, and Intel rose nearly 3% in the US stock market. NVIDIA and AMD have seen cumulative gains of 23.5% and 25.7% respectively this year, making them the two biggest winners in the backdrop of sustained high demand for chips.
At the recently concluded global consumer electronics "Spring Festival" CES exhibition, chip giants such as Intel, AMD, and Qualcomm gathered together to showcase their latest achievements in the AI PC field. Intel, AMD, and Qualcomm Snapdragon X Elite all embed NPUs (computing processors with integrated neural processing units) specifically designed for AI computing, forming a siege against NVIDIA's GeForce RTX 40 Super.
2024 is widely regarded as the year of AI PC, and NPU is seen as the computational hub of AI PC. Compared to GPUs that focus on large-scale parallel computing, NPUs are mainly designed for deep learning and neural network computations, enabling them to handle AI workloads with simpler control flow, higher efficiency, and lower power consumption.
NVIDIA under Siege
At the CES exhibition, Intel heavily promoted its newly launched Core Ultra processors, which incorporate NPUs for accelerating AI computing. The company claims that more than 230 models will be equipped with Core Ultra in 2024.
AMD showcased more details of its new Ryzen 8040 series chips, which were launched at the end of last year. The company claims that by embedding an NPU, the chip's performance in large language models has improved by 40%.
Qualcomm showcased its new generation Snapdragon X Elite PC chip, which was released in October last year. X Elite is Qualcomm's first processor targeting the PC market, with each chip equipped with an NPU that can provide 45TOPS (trillions of operations per second, one of the performance indicators for AI chips) and can run models with 13 billion parameters, claiming to "set a new benchmark for the industry".
In this round of global competition in the AI PC market, NVIDIA seems to be a step behind, only launching the GeForce RTX 40 Super series graphics cards for the PC market at this year's CES exhibition in January.
However, comparing the latest AI PC chips from the four companies, NVIDIA can still deliver excellent performance on PCs even without using an NPU. The NPU of Intel Core Ultra supports 10 TOPS, while AMD's NPU performance is 16 TOPS, both far behind the computing power of GeForce GPU, which exceeds 800 TOPS.
NVIDIA claims that the company is in a leading position in the AI PC competition, and its GPUs are the best engines driving new laptops and desktops.
Will NVIDIA Lead the Future of AI PCs?
In the past year, NVIDIA has accumulated a trillion-dollar valuation in the AI data center market. Analysts believe that if the company invests more effort in the AI PC field, it can not only expand the sales of its consumer-grade GPUs but also further consolidate NVIDIA's dominant position in the AI hardware industry.
On one hand, considering that ordinary computers, apart from gaming laptops and desktops, are usually not equipped with dedicated GPUs due to cost considerations, if NVIDIA successfully markets "all AI PCs should be equipped with GeForce dedicated graphics cards," it can drive the sales growth of GPUs at various price points.
GPUs have always been the foundation for AI application development, even the recent popular generative AI technologies require NVIDIA GPUs to achieve optimal performance. For example, the Stable Diffusion image generation tool defaults to requiring NVIDIA GPUs and requires a lot of adjustments to run properly on Intel or AMD chips.
Analysts believe that if NVIDIA can benefit from the AI PC market, it will further consolidate its market share in AI hardware and enhance its influence in the entire AI industry chain. Even with the emergence of various AI-specific chips like NPUs, they still cannot shake the position of NVIDIA GPUs in the AI field.