Lei Jun: XIAOMI-W achieves breakthrough in high-end implementation stage.

"XIAOMI-W's high-end has achieved a breakthrough," this is Lei Jun's declaration.

From Lei Jun's analysis of the XIAOMI-W high-end strategy, the breakthrough in the high-end market is due to Lei Jun's cognitive breakthrough.

Marcel Proust, the great French novelist and master of stream of consciousness, in the 20th century, said, "The real journey of discovery does not consist in seeking new landscapes, but in having new eyes."

"Having new eyes" is in line with Lei Jun's concept of "changing cognition to bring new growth."

Based on such changes, XIAOMI-W has also launched a corresponding new technological strategy. The core of this strategy is to adhere to the principle of "technology-oriented" and always focus on how to realize human value.

Under the guidance of XIAOMI-W's new technological strategy, what kind of chemical changes will XIAOMI-W undergo or will undergo?

How was the breakthrough in the high-end market achieved?

"XIAOMI-W's exploration of the high-end market, after three years, billions of dollars of investment, and the efforts of tens of thousands of people, has finally achieved a breakthrough," said Lei Jun, the founder, chairman, and CEO of XIAOMI-W.

Breaking through the high-end market is the consensus and collective behavior of the Chinese smartphone industry in the past three years. XIAOMI-W is no exception. Because any technology company or cultural phenomenon is influenced by the macro industry environment.

XIAOMI-W 10 is the first-generation high-end product of XIAOMI-W, followed by the lackluster XIAOMI-W 11 series, and then the XIAOMI-W 12 series, which fell short of expectations and suffered heavy losses. When the XIAOMI-W 12S Ultra, equipped with Leica dual cameras, was released and received positive market feedback, Lei Jun said, "(Finally) the turning point of the entire high-end strategy has arrived."

From XIAOMI-W 12 to the turning point of the high-end market, 12S Ultra, to the highly successful XIAOMI-W 13 series and XIAOMI-W 13 Ultra, what happened in between?

Here are some noteworthy statistics.

According to a report released by Canalys on May 30, in the first quarter of 2023, XIAOMI-W became the highest-ranking Chinese smartphone brand in the $500+ price range, and the XIAOMI-W 13 entered the top 15 in terms of shipments in this price range, ranking 13th.

Another statistical report on the Chinese market also shows that in the second quarter of 2023, in the Chinese smartphone market priced at 3,500 yuan and above, the XIAOMI-W 13 ranked sixth in terms of shipments and second among domestic brands (source: Canalys). Obviously, the market response to the XIAOMI-W 13 series is indeed excellent. So, going back to the original question, what happened from the XIAOMI-W 12 series to the 13 series?

Although in terms of absolute sales/shipments, the high-end achievements of the XIAOMI-W 13 series are not dazzling. Lei Jun also said, "The success of XIAOMI-W 13 today may not be significant," but "it has great significance within our company because it is our first representative work that truly achieves leading parameters and prioritizes user experience."

Lei Jun believes that the success of the XIAOMI-W 13 series is due to the "user experience first" consensus formed within XIAOMI-W.

This consensus is based on three principles known as the "XIAOMI-W high-end design principles": to create a high-end product, there must be no shortcomings. This means that a high-end flagship should not sacrifice performance for a certain "selling point" and thus affect the overall user experience.

For example, the latest generation of XIAOMI-W's foldable screen, the XIAOMI-W MIX Fold 3, follows the guiding principle of "flagship first, foldable second," resulting in a combination of "thin and foldable" with a comprehensive flagship experience.

Although Lei Jun recognized the importance of "thinness" as the only core choice for the maturity of the foldable screen category before the release of XIAOMI-W MIX Fold 2, when launching the XIAOMI-W MIX Fold 3, he did not sacrifice features such as imaging capabilities, which were the breakthroughs chosen by XIAOMI-W for the high-end market.

Therefore, the XIAOMI-W MIX Fold 3 aims to achieve a "thin and flagship imaging" foldable device with the performance of "the only full focal length quad-camera in a foldable screen to date."

The market has provided feedback on this goal with a set of data. According to XIAOMI-W's Senior Vice President and Head of the Mobile Phone Division, Zeng Xuezong, "Within 5 minutes of going on sale, the sales volume increased 2.25 times compared to the previous generation, which should be a new record for our foldable screen series."

From the concept of "thin and foldable" combined with a comprehensive flagship experience, it is evident that XIAOMI-W places unparalleled emphasis on "user experience." How should we understand this? Which is more important, specifications or user experience?

The second principle of XIAOMI-W's high-end design is to firmly choose user experience when there is a conflict between specifications and experience.

XIAOMI-W used to be a staunch advocate of "specifications." Whenever a product emphasized outstanding performance, XIAOMI-W always liked to show off its specifications. This preference for specifications led the industry to question: Does excellent specifications equal high-end?

Clearly, XIAOMI-W has now chosen the latter between specifications and user experience. So the question becomes, what constitutes a good user experience? Or rather, how can we achieve a good user experience starting from the system? This leads to the third principle of XIAOMI-W high-end design: deep integration of software and hardware, aiming for "complete connectivity". As Lei Jun said, "Only in this way can we ensure a seamless experience throughout the entire process."

How to achieve cognitive change and breakthroughs

The three principles of XIAOMI-W high-end design are actually based on cognitive change and are the result of breakthroughs in cognition. In the journey towards high-end, as Lei Jun puts it, "Every tiny progress is hard-won, requiring changes and breakthroughs in cognition."

Marcel Proust, a 20th-century French stream of consciousness novelist, once said, "The true journey of discovery lies not in finding new continents, but in having new perspectives."

Having a new perspective can be seen as a cognitive breakthrough. So what is a cognitive breakthrough?

For example, let's take the "Professional Photography Kit" accessory for the XIAOMI-W 13 Ultra. According to Lei Jun, this accessory was rejected twice when proposed, but after continuous and in-depth communication with photographers during a photography workshop held by Leica within XIAOMI-W, Lei Jun fully supported the proposal on the team's third attempt without any hesitation. And that's how we have the highly popular "Professional Photography Kit" today.

How can we tell that this accessory is well-received in the market?

Lei Jun said, "The initial discounted price was set at 799 yuan, but at one point, it was being scalped online for over 1,400 yuan." Wall Street News checked the price of this accessory on the XIAOMI-W online store, and it is currently priced at 999 yuan. As of August 20th, there are 13,656 reviews, with over 100,000 views and a 99.6% positive rating.

Looking back at the process of launching the "Professional Photography Kit" for the XIAOMI-W 13 Ultra (two rejections and one approval), Lei Jun commented, "With the same idea and the same product, the decision is completely different after changing the cognition."

This is the result of cognitive change. First, we need to change our cognition, and then the results will follow suit.

It should be said that cognitive change does not happen out of thin air. In other words, what factors support/cause cognitive change and breakthroughs?

For example, the high-end transformation of XIAOMI-W: When launching the XIAOMI-W 11 Ultra, XIAOMI-W fell into a "self-indulgence" trap. According to Lei Jun, the XIAOMI-W team "self-defined" the goal and was heavily influenced by parameter thinking, believing that as long as the imaging capability of the XIAOMI-W 11S Ultra reached the top score in DXO tests and the hardware configuration was maximized, it would be on par with successful high-end products.

But how did the market react?

"On the first day of the XIAOMI-W 11 Ultra's release, users were extremely critical, and we were completely caught off guard." Lei Jun and his team felt frustrated and confused, "With such amazing imaging, being ranked first globally by DXO, and having all the parameters maxed out, why are users still not satisfied?"

After a long period of reflection, Lei Jun and his team acknowledged that "having top parameters and being ranked first by DXO does not necessarily mean good photography, let alone user satisfaction."

How can we satisfy users? "My experience from university helps me, find someone who understands and ask them." Lei Jun said, "Can we talk to Leica?"

At the beginning of the collaboration with Leica, it was not a direct joint release, but rather a deep immersion in the field, engaging in conversations with photographers and organizing photography training sessions with Leica as the main speaker, in order to understand the image process and consumer psychology. The latter even influenced the choice and breakthrough in Leica's image tone.

Due to concerns that the classic Leica image tone, which had never been achieved on a smartphone before, would be difficult for the general public to accept, Lei Jun and his team had multiple discussions with Leica and decided to introduce the "Leica Vivid" tone alongside the Leica classic tone to balance the preferences of the general public. However, this required a long period of tuning (in fact, it took two years).

Thus, the dual-camera system with Leica technology was born. It was the result of joint research between Leica and XIAOMI-W, and it represented a new breakthrough in smartphone imaging for XIAOMI-W and Leica. The first smartphone equipped with the dual-camera system was the XIAOMI-W 12S Ultra, which Lei Jun referred to as a "turning point in the entire high-end strategy." He said, "The effect after its release was particularly good, not only loved by the general users, but also highly praised by photographers."

What are the upgrades in the new technology strategy?

From the 12S Ultra to the consecutive successes of the XIAOMI-W 13 series and XIAOMI-W 13 Ultra, Lei Jun believes that this is because "there has been a change in perception and breakthrough in perception, and only with a breakthrough in perception can there be real growth."

We are currently in an era of tremendous change, and with the industrial transformation brought about by AI large models, the continuous evolution trend in the field of information technology and the new opportunities it brings, combined with XIAOMI-W's own breakthroughs in perception and growth needs, XIAOMI-W has reached a significant turning point for growth in this era.

Based on such a macro-industry background and micro-changes, XIAOMI-W announced the advancement of its technology strategy and unveiled XIAOMI-W's technology philosophy: "Two Long-Terms" - choosing technology fields that have "long-term value" for human civilization and adhering to "long-term and continuous" investment.

Based on long-term investment, XIAOMI-W has established four key paths and principles: deep cultivation of underlying technology, long-term and continuous investment, deep integration of software and hardware, and comprehensive AI empowerment. Among them, the deep integration of software and hardware is the fundamental guarantee for XIAOMI-W to provide unique experiences to users.

Long-term and continuous investment is not just a slogan: In 2017, XIAOMI-W invested 3.2 billion yuan in research and development expenses; in 2022, this figure increased to 16 billion yuan, a year-on-year growth of 21%; Lei Jun predicts that in 2023, XIAOMI-W's research and development expenses will exceed 20 billion yuan; in the next five years (2022-2026), XIAOMI-W's research and development investment will exceed 100 billion yuan. This R&D investment of hundreds of billions will continue to focus on 12 technical fields, including 5G mobile communication technology, big data, cloud computing, and artificial intelligence technology. It will also encompass intelligent manufacturing, such as robots, unmanned factories, and smart electric vehicles, with a total of 99 specific areas.

As of March 31, 2023, XIAOMI-W has obtained over 32,000 global authorized patents. Among them, in the ranking of global patent families with effective 5G standard essential patent declarations released by the China Academy of Information and Communications Technology, XIAOMI-W's global patent family accounts for 4.1%, entering the top ten for the first time.

Among the four key paths, the deep integration of software and hardware and the comprehensive empowerment of AI are the core. XIAOMI-W summarizes the relationship between various elements with the formula "(Software × Hardware)ᴬᴵ".

In July 2016, XIAOMI-W established the AI Vision team, and in April 2022, the large-scale model team was formed. Over the past seven years, after six expansions, XIAOMI-W's AI team has grown to over 3,000 people.

XIAOMI-W AI adheres to the self-developed full-stack technology route, currently covering various directions such as vision, acoustics, speech, NLP, knowledge graph, machine learning, large-scale models, and multimodal.

Recently, news broke that XIAOMI-W's 6B parameter self-developed large-scale model ranked first in the C-EVAL authoritative list with the same parameter level, and CMMLU's Chinese large-scale model ranked first.

The Eval list, full name C-Eval Global Large-scale Model Comprehensive Examination and Testing List, is a comprehensive examination and evaluation suite for Chinese language models constructed by Tsinghua University, Shanghai Jiao Tong University, and the University of Edinburgh.

This suite covers four major directions: humanities, social sciences, science and engineering, and other disciplines, including 52 subjects, covering multiple knowledge fields such as calculus and linear algebra. There are a total of 13,948 Chinese knowledge and reasoning questions, which can comprehensively evaluate the language processing capabilities of models and provide valuable reference for the development of Chinese community language large-scale models.

The main breakthrough direction of XIAOMI-W's large-scale model technology is lightweight and local deployment.

XIAOMI-W is considering whether it can deploy large-scale models on everyone's mobile phones first, optimize on-device computation, and allow everyone to use on-device large-scale models. The advantage of doing this is that users can ensure data security while having the powerful productivity of large-scale models.

Currently, XIAOMI-W has achieved on-device computation of large-scale models on mobile phones and self-developed a 1.3B parameter on-device large-scale model, which can achieve effects comparable to 6B parameter cloud-based large-scale models in multiple scenarios.

When will on-device AI model capabilities be implemented?

In XIAOMI-W's technology layout for the future, the most attention-grabbing or currently popular technology and application direction is likely to be on-device AI large-scale models.

To be honest, XIAOMI-W has not fully implemented on-device AI capabilities yet. For XIAOMI-W, on-device AI large-scale models are an investment-worthy future. "AI big models have emerged, and when combined with smartphones, they become 'people, cars, and homes,' ultimately achieving interconnection between 'phones, cars, and IoT.' This will undoubtedly stimulate new innovations in the entire smartphone industry, leading to more practical applications."

Regarding the application of 5G, both the industry and the general public do not find it particularly appealing. It seems that apart from the difference in network speed between 5G and 4G, there are no significant advantages to 5G. Zeng Xuezong also believes that "so far, the investment and output of 5G have not met expectations."

"It is possible that AI big models are specifically designed for 5G, making them the most important killer application of the 5G era." Zeng Xuezong said, "If we consider the 2G era as the era of text messages, the 3G era as the era of mobile internet, and the 4G era as the era of videos/short videos, then the 5G era could be the era of AI big models. This is a tremendous opportunity."

Will the key variable for future smartphone innovations be the deployment of edge AI big models?

During the initial stage of large-scale 5G deployment, the ability of terminals to understand individuals was considered an iterative improvement compared to 4G. However, subsequent developments have shown that such improvements are not significant or particularly noticeable.

However, once edge AI big models are deployed on smartphones, achieving the "lightweight, local deployment" that XIAOMI-W aims for, it is highly likely that the goal of "understanding individual intentions at the terminal" can be achieved.

Zeng Xuezong said, "In the future, when terminals (that deploy edge AI models) can truly perceive what individuals think and want, help users make anticipatory judgments, and provide decisions, it will be a great leap forward."

"Big models are a technological revolution, and from a professional perspective, the most essential application of big models is a significant improvement in understanding users." This is the judgment of Wang Bin, Vice Chairman of the XIAOMI-W Technical Committee and Director of the AI Laboratory. "Things that were previously impossible or unimaginable can now be accomplished with (edge AI) big models. If these functions are combined with smartphones, they will undoubtedly create a tremendous synergy."

According to Wang Bin, XIAOMI-W will not develop super-large-scale big models. This aligns with Lei Jun's direction of "lightweight, local deployment." Wang Bin said, "We have our own pace, and our consideration is 'user experience first,' insisting on prioritizing the user experience."

So how will XIAOMI-W deploy edge AI big models? And when will it happen?

Wang Bin said, "We maintain close communication with our partners and collaborate on development, including Qualcomm and other hardware platform manufacturers, to tackle relevant issues together."

Zeng Xuezong revealed to Wall Street News, "It's not just Qualcomm; we also have deep cooperation with MediaTek. We used to be more proactive, and now both sides are very proactive. Their chips are also rapidly evolving and iterating. They share their thinking based on AI edge technology with us and hope that we can provide corresponding cooperation."

As for the timeline for deployment, Zeng Xuezong said, "It should be said that it will happen very soon, not far away." According to Dolphin Research's supply chain sources, Qualcomm is set to launch the Snapdragon 8Gen3 5G SoC mobile platform in September this year. This generation of Snapdragon chips will integrate the computational module for AI large-scale models on the edge.

Therefore, if we combine what Zeng Xuezong revealed to Wall Street News - "The deep integration between our two parties is very strong" - it is highly likely that XIAOMI-W's true on-device AI large-scale model application capability will be realized in the next generation flagship devices of the XIAOMI-W 14 series.

So the question becomes, when will the XIAOMI-W 14 series be released?

From achieving breakthroughs in the high-end segment, to implementing the four pathways of technological strategic upgrading, and to the core principle of "technology-oriented" while prioritizing user experience and adhering to the "two long-term" approach, a brand new XIAOMI-W is steadily moving towards a visible future.