【Abstract】Behind the rumors of ByteDance developing AI smartphones is actually the logic of wanting to be a large model supplier.

In other words, for ByteDance, the focus of AI smartphones lies in AI, not the phone itself.

The profit logic should be that ByteDance directly supplies mobile phone manufacturers, and the profits obtained are divided proportionately, similar to the APP Store model. At the same time, it implants its own professional software to improve its own products.

From a timeline perspective, the large-scale maturity of AI smartphones is expected in 2026 or 2027. Before then, various manufacturers need to go through a cycle of development, deployment, entering daily life, and then changing user habits. Therefore, the products at this stage cannot yet form sufficient willingness to pay.

Internet manufacturers with cloud, data, and computing power bases hold important resources on the path to AI large models. Based on this, ByteDance's entry into AI to empower smartphones makes sense.

It is foreseeable that ByteDance will not be the last manufacturer to enter the AI large model supplier market. In the next 2-3 years, each company will find its own characteristics and work quietly.

Search and add Xinliu WeChat Aristodemus0403 for in-depth discussions on the latest developments in the mobile phone industry (no group creation), market and project consulting, talent services, and decision-making judgments.

Below is the main text:

ByteDance is not making AI smartphones, but a large model supplier.

What AI smartphones lack today is not new opportunities, but a suitable definition. Similarly, what each company lacks is not a trend, but more importantly, finding its own area of advantage.

Not long after various mobile phone manufacturers shouted AI slogans, the industry also spread rumors that ByteDance was going to develop AI smartphones.

Previously, industry sources said that ByteDance secretly launched an AI smartphone research and development project in April. The core team participating in this project is mainly composed of two parts: one part comes from the Smartisan mobile phone research and development team acquired by ByteDance in 2019, and the other part comes from the PICO VR research and development team acquired in 2021.

AI smartphones are a new concept, and ByteDance has been eyeing hardware for a long time, undoubtedly attracting a lot of attention.

However, on June 12, a media reporter verified with ByteDance, and its relevant personnel said: "The information is not accurate. In fact, we are exploring large model software solutions based on mobile phones to provide reference for mobile phone manufacturers. Currently, there is no plan to make and sell our own phones."

The meaning of this statement is actually very clear. For ByteDance, AI smartphones focus on AI, not the phone itself. The former is ByteDance's area of advantage.

Under the current situation, there are actually two ways for manufacturers to enter the AI smartphone market.

One is that mobile phone manufacturers do it themselves, integrate it into the phone system, and may package it in monthly services later, similar to Apple's iCloud charging model. This part of the AI large model is mainly targeted at defining content related to mobile phone brands or optimizing based on the consumer group characteristics of different brands, which is mainly controlled by mobile phone manufacturers.

The second is for third-party large model manufacturers to do it. Based on their own advantages, they provide it directly to mobile phone manufacturers after completion, and the profits obtained are divided proportionately, similar to the APP Store model. Not only that, third-party large model manufacturers often have their own main business software that can be implanted to improve their own products.

This is the second approach that ByteDance intends to adopt.

As an internet manufacturer, ByteDance's AI large model can not only charge in its own professional apps, such as CapCut, but can also participate in revenue sharing with consumer electronics manufacturers and mobile phone manufacturers, effectively forming a competitive cooperation relationship.

Currently, even if mobile phones are made, the increment is not significant.

An industry insider from the consumer electronics industry revealed to Xinliu Think Tank that under the current AI trend, laptops are the first increment to be empowered, and cars are also an increment, but mobile phones have not yet had much opportunity.

An industry insider said that AIPC is the most noticeable trend at this year's computer show and may mature ahead of AI smartphones.

The main profit model is To B, assisting with programming, writing PPTs, organizing documents, etc., which can drive the replacement of 30 to 50 million computers in the next few years.

From a market environment perspective, foreign countries will also have a more significant effort. Their market replacement cycle is 4-5 years, and they replace their devices when they expire, with relatively fixed iterations. Most European and American employees' work laptops are recovered and disposed of by the company after they leave, and are not reused. Therefore, once AIPC enhances productivity, companies will easily promote large-scale device replacements.

Under this logic, the replacement of mobile phones relies more on personal habits, attractive features of AI large models, and other factors.

From a timeline perspective, the large-scale maturity of AI smartphones is expected in 2026 or 2027. Before then, various manufacturers need to go through a cycle of development, deployment, entering daily life, and then changing user habits. Therefore, the products at this stage cannot yet form sufficient willingness to pay.

Although various companies now have AI large model components and software, the market still hasn't changed the past impression of "artificial stupidity." This 2-3-year window leaves room for various mobile phone manufacturers, as well as new entrants with advantages in segmented areas.

Even Apple's first AI large model, Apple Intelligence, which was introduced at the iOS 18 launch event, is officially scheduled to officially launch this fall, and its specific effects remain to be seen. The national version of Apple's smartphones with large models will have to wait until next year at the earliest.

Under the current situation, a large amount of large models and data are difficult to enter China. Both mobile phone and automotive terminals require continuous optimization and iteration based on data. Domestic manufacturers are also conducting development work. Currently, it seems that the large model for cars has already been determined to be deployed on the end side.

ByteDance's advantage lies in its internet genes.

ByteDance may not have much experience in hardware, but when it comes to computing power and data, ByteDance has sufficient say.

The development of AI smartphones to a sellable version ultimately needs to land on paid scenarios, which requires large-scale identification of possible demands. Internet manufacturers have unique advantages and barriers in this regard.

Google has been promoting the concept of "smart assistants." For example, for travel scenarios, no matter how complex the user's input is, it can modify flight tickets, determine travel plans in emails, and recommend coffee brands based on whether the user likes coffee when identifying specific locations.

ByteDance's Doubao has also ventured into the field of children's education, helping parents guide homework and guide children to read.

Whether consumers will pay for efficiency software ultimately depends on whether these paid scenario identification, definitions, and services are perfect. This characteristic makes the company's existing data crucial, as it can directly identify users' group characteristics, behavioral patterns, emotional tendencies, etc., resulting in more comprehensive optimization.

Moreover, internet companies already possess a certain computing power foundation. Cloud, computing power, and data form a golden triangle for these manufacturers to access "AI+ products."

It is with such cooperation that there have been large-scale cloud price reductions and token plunges in recent times. Each company is seeking to bind partners and build its own new territory based on its own advantages.

ByteDance will not be the last manufacturer to enter the AI large model supplier market. Each company is quietly exerting its efforts by finding its own characteristics.