Zhao Xiaoguang from Tianfeng Securities: The essence of OpenAI's power struggle lies in the lack of data. In the future, the most important aspect of AI will be its application.

Zhitong App learned that on November 18th, Zhao Xiaoguang, Vice President and Director of the Research Institute of TF Securities, made judgments on the recent power struggle within OpenAI and the future development of AI. Zhao Xiaoguang believes that Chat GPT will soon encounter bottlenecks because this round of Chat GPT is driven by technology. In the past, due to the use of open-source data on the Internet, whether it is Chat GPT 3 or Chat GPT 4, a pulse was quickly generated. However, the problem is that once he obtains the open-source data, there is no more data. The recent changes in the internal management of OpenAI are due to their encounter with bottlenecks and stagnation. The fundamental reason for the stagnation is the lack of data.

Zhao Xiaoguang pointed out that data comes from scenarios, that is, applications. Whether it is in the consumer (C) or business (B) sectors, AI will stop here if there are no applications.

This is why Nvidia (NVDA.US) has such heavy restrictions on China. Zhao Xiaoguang believes that half of it is due to the United States and half is due to their own factors. They are very anxious because on the one hand, they face future competition from companies such as Qualcomm, AMD, and even Intel. On the other hand, international giants are also making their own chips, such as Microsoft, Amazon, and Apple. At this time, Nvidia must continue to develop applications, but what they are most worried about is that there will be a large number of competitors before the applications are developed.

Meanwhile, China has purchased a large number of chips and computing power for servers. At the same time, China is the largest application scenario for the business-to-business (B2B) industry in the world, which Nvidia attaches great importance to. If China can develop applications and large models first, Nvidia may face competition from all aspects, just like Cisco did in the past when it was purely a technology or hardware company and faced competition when it reached a certain level of development.

Therefore, from the perspective of Nvidia and OpenAI, applications have gradually become the most important part of the next round of AI.

Zhao Xiaoguang said that on the one hand, we see the development of these applications in the B2B field, and on the other hand, we also see that in both the B2B and consumer (C2C) fields, without scenarios and data, it is difficult for AI to obtain the core productivity, and it is difficult to optimize the EM (Efficiency Management).

In the C2C field, it is necessary to create scenarios in the future, whether it is VR or the metaverse, these are very important. In the B2B field, the Internet of Things (IoT) is undoubtedly very important. Without data, AI cannot be realized. Even if the AI model is very powerful, it will definitely be wrong.

Therefore, it is believed that the next step for AI is applications, and the next step for applications is scenarios. In the B2B field, it should be the Internet of Things, and in the C2C field, it should be various AI devices, mainly video-related devices represented by VR and MR. From a technical perspective, it is easier to achieve in the B2B field compared to the C2C field.