Today is a research report on Qualcomm. The reason for paying attention to Qualcomm is that I have been looking at edge AI recently, and Qualcomm will play an important role in it. Qualcomm will release a new financial report after the US stock market closes tonight, which is definitely worth关注的.

Edge AI is not popular at all right now, but what about five years from now? A major player in five years, a valuation currently underestimated, is worth taking a closer look at.

So-called edge AI refers to: cars, smartphones, robots, AI glasses, etc., the market still has imagination.

Qualcomm's current valuation is extremely low, mainly because the mobile phone business will continue to deteriorate, coupled with the split from Apple. However, on the other hand, the turning point of edge AI has appeared (I will detail this in the following article), and things become interesting.

One point to pay attention to in tonight's financial report is that, based on the situation after the first quarter report was released, the market is very sensitive to the decline in the mobile business, and the stock price still dropped 9% after hours. It is also expected that the mobile business will not see significant growth this quarter, and it is very likely that it will be a decline in data + further downward guidance.

Currently, the market's view of Qualcomm has split into two sides—one side feels that the valuation is still too high due to the shrinking mobile business, while the other side believes it is an undervalued edge AI target.

As of late April, the target prices from 23 brokerages range from $100 to $216, showing a wide span.

But looking at it over a longer timeline is where the value lies.

I began to pay attention to Qualcomm from the research of inference, as inference will become the mainstream in the future AI world, and much of the inference will occur at the edge.

But what problems do edge models actually solve? The applications we use on our phones, such as Doubao, ChatGPT, and Claude, seem not to require edge models—they are all cloud-based.

This research report revolves around three questions:

First, what exactly is edge AI, how large is the market, and where is technological evolution headed?

Second, where are the real application scenarios for edge AI, and are they a substitute or supplementary relationship with cloud LLMs?

Third, what is Qualcomm's position in edge AI, and is the current 12-14x Forward PE an opportunity or a trap?

1. What Exactly is Edge AI

1.1 Conversations Do Not Require Edge Models

Any AI like ChatGPT and Claude, almost all computations are done in the cloud. The smartphone is just a terminal for sending and receiving text; cloud AI is stronger than any model that can be stuffed into a phone. Therefore, if "edge AI" is defined as "running a LLM on your phone to converse with you," the cloud capabilities crush it.

But edge AI definitely does not mean this. It addresses problems that cloud cannot solve, cannot do well, or do not make financial sense to tackle.

For the full text, click here, 18,000 words, Twitter can't handle it.

https://mp.weixin.qq.com/s/M_qU__JSLUSw7Y1BoNy8-w

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。