Source: Waking of Insects Research Institute, Author: Xiaoman

Image source: Generated by Unbounded AI tool

In anticipation, Lei Jun's annual speech came as scheduled. This more than 3-hour speech with the theme of "growth" continued from the long aftertaste of "the past of Wuhan University" to the sadness of Xiaomi's impact on high-end, but unfortunately, among the full chicken soup, the industry is most concerned about the progress of Xiaomi's car manufacturing But nothing was mentioned.

Surprisingly, Lei Jun, who did not disclose the progress of car-making, passionately announced in his speech that Xiaomi is actively embracing large models. You must know that among the technology companies that have announced their entry into the big model, Xiaomi has nothing unique, and the outside world is more concerned about how Xiaomi can achieve new growth after the mobile phone market is cold.

So can the large model become a new growth point outside of Xiaomi's car manufacturing? Can Xiaomi use the large model + car building to create a new G-spot that was so popular a few years ago? Lei Jun obviously knows the answers to these two questions.

Xiaomi’s new antidote to growth, after the car is built, it will be a large model

The current mobile phone market is hotly contested. The top players include Apple, which cannot be defeated by iron, and Huawei Hongmeng, which will soon be incompatible with Android applications. Afterwards, there are players in the middle of the waist from Honor, OPPO, and vivo, plus OnePlus and Nubia, which have taken away precise users with gaming phones. Surrounded by a pack of wolves, Xiaomi's high-end dream is extremely difficult, and Lei Jun couldn't help but express his uncontrollable anxiety during his annual speech.

Compared with Lei Jun, he has known for a long time that although there is a certain base of rice fans sticking to the camp, it is not easy to win the real high-end market. **The previous heap parameter route is obviously no longer feasible, and the high-priced folding screen mobile phone may not be able to take all high-end products. **What is placed in front of Xiaomi is that the mobile phone market has let it eat up its dividends prematurely. Although it has released a new product equipped with a self-developed surging chip, compared with Huawei Hongmeng Ecology, which claims to not support Android applications, Xiaomi Based on the self-developed chip ALL IN, the confidence of a brand new Xiaomi new ecology is still slightly insufficient.

As for the stalemate between the second-tier market and Honor, OPPO, vivo, and OnePlus, Xiaomi is also in a dilemma. According to Counterpoint data, global smartphone sales in the second quarter of 2023 will fall by 8% year-on-year and 5% quarter-on-quarter. Some analysts predict, "The recovery node of the entire industry has been delayed. This year, the global smartphone market may remain flat and slightly decline, and recovery may have to wait until the end of the year or next year." **Obviously, mobile phones, whether high-end or mid- The low-end can no longer become a new growth point for Xiaomi. **

Lei Jun naturally understands this truth. The future he plans for Xiaomi has become clearer with the annual speech: a large model with his left hand and a car with his right hand, plus a robot like Tie Daner. Smarter, ubiquitous, multi-scenario attack, so a brand new, intelligent Xiaomi ecology seems to be coming soon.

**However, Lei Jun announced that the Xiaomi car-making plan to "achieve mass production in 2024" is about to enter the countdown, and the Xiaomi car has not yet revealed its true face. ** Not long ago, it was rumored that even the employees of the Xiaomi Yizhuang factory had never seen the whole picture of Xiaomi cars, which added a layer of illusory atmosphere to the matter of Xiaomi making cars.

Judging from the past situation, the story of Lei Jun’s official announcement is the same as the official announcement of self-developed chips. Although it will eventually appear, it is always long overdue. If Xiaomi Auto continues to skip tickets like Jobs' car-making plan, how will Xiaomi describe a differentiated future vision of smart travel in the competition with "Wei Xiaoli" and other car-making forces? **Surprisingly, Lei Jun did not disclose one or two of these issues that people are extremely concerned about in his annual speech in the sensational circle.

However, after Lei Jun's annual speech filled everyone with chicken soup, there was an episode of actively "embracing" the large model - plunged into the ocean of "Hundred Models Battle". Looking at Lei Jun's confidence, it seems that he has found a new growth point for Xiaomi outside of car manufacturing.

However, like building a car, a large model is a track that needs to "burn a lot of money". Before the official launch of the Xiaomi car, Xiaomi joined the "Hundred Models War". If ** is not highly confident in its financial situation, then the official announcement of the large model is probably just to catch the heat of the new trend. ** After all, the famous saying "Standing on the wind, pigs can fly" comes from Lei Jun. So, will Xiaomi's large model bring new growth to Xiaomi?

**How much is the difference between Xiaomi's large model and AGI's OpenAI? **

Xiaomi, which is good at digging out product highlights, is also different in the development strategy of large models. In Lei Jun's words, Xiaomi's large model follows the path of "light weight and local deployment". However, in the past 9 months when ChatGPT set off a large-scale model craze, when many technology companies launched a large-scale model card battle, there was never such a statement as "lightweight and local deployment". Therefore, although Lei Jun showed a lot of confidence in his own large-scale model in his speech, ** but the fact is that the simple path of "lightweight and local deployment" is not the mainstream of the current mainstream large-scale model. **

The "Three Musketeers" composed of OpenAI, Microsoft, and Nvidia are intensively deploying the "general-purpose large model" on the east coast of the Pacific Ocean, instead of taking the lighter Xiaomi model. In China, Baidu, Ali, SenseTime, Huawei, Tencent, JD.com and other leaders eager to occupy the large-scale model market are also firmly cultivating "universal large-scale models", and none of them follow Xiaomi's "lightweight, local deployment" route.

Compared with the general-purpose large model, what are the advantages and disadvantages of the Xiaomi model? What is the future prospect? Will it combine with its big dream of smart travel to create a new "ecological millet"?

The prospect of "lightweight" large models is unknown, but the strategic path choice of "local deployment" has obvious advantages and disadvantages. Indeed, as Lei Jun said in his speech, **"local deployment" can effectively avoid privacy risks, and it can also be more easily implemented on the mobile phone side. However, compared with "cloud deployment", large models based on "local deployment" may have limited computing power, and computing power is the cornerstone of large model operation. **

In addition, the "universal large model" is the common choice of almost all players in the field of large models, except Xiaomi. Perhaps, the "universal large model" is a development paradigm recognized by leading companies in the industry that can fill large models with sufficient imagination and application scenarios. **Xiaomi's new "lightweight, local deployment" route is more like an additional function for "empowering" Xiaomi phones. **

It is worth mentioning that the evaluation test results of Xiaomi's large model MiLM-6B were announced at the press conference, and Xiaomi, which is good at running points, naturally achieved good results. In the C-evaluation, the average score of MiLM-6B is 60.2, ranking 10th in the overall list and ranking 1st in the same parameters. However, according to the information given on the GitHub project page, the parameter size of Xiaomi's large model MiLM-6B is only 6.4 billion, and the training parameters are the lowest among the top ten large models. I don't know if this is also one of Xiaomi's differentiation strategies.

**Frankly speaking, the difference between the Xiaomi model and the general model has been widened in terms of future growth opportunities and imagination. **Although "light weight" can be quickly built, "local deployment" can protect privacy according to Lei Jun's words. Perhaps this is the differentiated competition strategy that Lei Jun found for Xiaomi's large model, but the "self-contained" Xiaomi large model model requires If you want to be accepted by the market, I am afraid that you will need to pay a lot of effort and trial and error costs in the future.

** When will "Dreamer" Lei Jun realize his dream? **

In an annual speech that took a long time, Lei Jun didn't mention the progress of car manufacturing, but threw out the "big model" a little slowly. It must have been a helpless choice after some time. **But Xiaomi does have both long-term and near-term worries. **

The difficulties encountered by Xiaomi mobile phones in the pursuit of high-end dreams made Lei Jun unabashedly express his anxiety in the past year in his speech. In fact, on the whole, as the basic business of Xiaomi, mobile phones are indeed a core business that cannot be easily given up. But how does Xiaomi, which has grown up with its cost-effective advantage, realize its "high-end dream"? This is a problem engraved in the genes, and it has also determined the foundation of Xiaomi's product, brand and market establishment from the very beginning.

The problems encountered in the high-end dream of Xiaomi mobile phones, on the one hand, come from the "gap" that is difficult to bridge between Mi Fan's own consumption capacity and high-end pricing. **If Xiaomi wants to seize the high-end market, it will inevitably give up most of its rice noodles, and turn to obtain more "new rice noodles" with high-end consumption capabilities, but will the market wait for Xiaomi in this process? The answer is self-evident. **

On the other hand, under the intensive bombardment of high-end models from friends, it is really difficult for Xiaomi's handful of high-end models to form an effective combat force to meet and challenge. At a time when competition in the high-end market has become extremely fierce, it is difficult for Xiaomi's chances of winning to be clear.

In fact, Lei Jun's own interpretation of Xiaomi's high-end experience is more "vivid". He said bluntly, **"Xiaomi's high-end exploration in the past three years has been the most painful and rewarding growth for me in the past 10 years." **Since childhood From the smooth progress of Mi 10 to the major setbacks encountered by Mi 11, and then to the failure of Mi 12, Xiaomi did not expect that it would be so difficult to make high-end, and the investment was so large. There were even great differences within the team, so that Lei Jun’s The whole Spring Festival holiday has not been good. But Xiaomi, who is stuck at the threshold of transformation, has no choice, and the "dreamer" Lei Jun can only continue to dream and chase dreams.

As the smartphone market welcomes the "sunset", although Lei Jun has already achieved financial freedom, as the soul of Xiaomi's growth, he must grit his teeth and continue to turn the pressure into motivation for the realization of Xiaomi's "high-end dream" of mobile phones. Xiaomi also continues to look for new growth points, and Lei Jun can't stop dreaming. Xiaomi is attracting the market with new stories about building cars and large models, and Lei Jun is also continuing to write his own new stories.

For the sake of enterprise development, Xiaomi really needs new growth points, not only to tell stories to the market and rice noodles, but also to show real skills, either to promote the launch of Xiaomi cars, or to be visible to the naked eye like Wang Xiaochuan The speed allows your own large model to really run.

People love hearing stories, but are even more disgusted by listening to stories without seeing the actual development of the plot. If both car manufacturing and large-scale model layout cannot be recognized by the market, the so-called "new growth point" will not become a new G-spot that drives the carnival.