At 1 a.m. on June 11, Beijing time, once a year$Apple (AAPL.US)$WWDC, the developer event, is about to kick off.

The most remarkable thing about this event was Apple's layout in the AI field. It is expected that about half of the keynote speeches will be related to AI.

Last Wednesday, Nvidia, whose market capitalization has been advancing by leaps and bounds, briefly surpassed Apple to become the second-largest US stock market capitalization, and is now a popular fried chicken.

Although there was a lot of negative news about Apple's product sales in the first quarter, the crazy efforts in generative AI applications over the past month also made people see some sincerity in the new iOS 18 version, and Apple's stock price has also quietly rebounded.

This press conference may determine whether Apple will be left behind by Nvidia in future market capitalization competition.

Armed to the teeth with AI

The media quoted people familiar with the matter as reporting that at the press conference, Apple will launch a new AI system called “Apple Intelligence” and integrate it into the iPhone, iPad, and Mac operating systems.

AI = Apple Intelligence, isn't it very perverted?

According to the details of the new features revealed by the media, all native apps on the iOS system have the possibility of being armed to the teeth:

Camera: Take a photo/video in a specific mode, set the countdown and switch between front and rear lenses in one sentence

Photos: Edit and retouch images, organize albums, and even find specific people, pets, landscapes based on keywords...

Memo: transcribed audio to text, voice summary, formula recognition

Voice Memos: Record, transcribe, archive, and organize folders to make meeting notes more efficient

Reminder: Create a new shopping list, run errands, add tags, and organize categories

Email: Write emails, save drafts, set reminders, and automatically categorize ad spam

Books: Navigate to a specific book or chapter, turn pages and switch topics with one click

Safari: Summarize page summaries and organize tabs

Keynote: Insert audio, video, and images in one sentence to automatically edit slides

As for the above functions, in fact, AI phones released by many Android manufacturers last year have basically been implemented. Of course, we are still in the AI functionalization stage, but with the Apple faction's deep involvement, the AI application model is expected to evolve at an accelerated pace.

Last year, major manufacturers such as OPPO and VIVO all introduced new AI assistants, which far surpassed previous voice assistants in terms of functional support. And in Apple's upcoming iOS 18, the Siri upgrade is also one of the most widely watched features. According to foreign media reviews, Siri will be reborn in iOS 18.

In iOS 18, Siri will integrate LLM's large language model to optimize conversation continuity while allowing integration into real-time situations. Coupled with the previously reported collaboration with Open AI and Google, I don't know if the final product will bring us a Siri that not only supports complex instructions, but also responds fluently.

From the layout of AI to the upcoming launch of AI, in the future, Apple may create an operating system that is most suitable for implementing AI applications.

AI phones, don't just make money

However, if users want to get a new phone, there must always be a reason, such as insufficient memory or battery life. Advances in device hardware have gradually led to the trend of high-end mobile phone consumption.

The mobile phone market showed a very clear improvement in the first quarter of this year, with shipments growing for the third consecutive quarter. According to a report released by market research agency IDC, global smartphone shipments increased 7.8% year-on-year to 289 million units in the first quarter of 2024.

Looking at the smartphone market as a whole, the average global smartphone price in the first quarter was 370 US dollars (about 2,700 yuan), an increase of 4 US dollars compared to the same period last year.

Furthermore, judging from the sales of smartphones in various price ranges in the first quarter, the smartphone market with an average unit price of more than 800 US dollars grew the fastest, with a growth rate of more than double digits, accounting for 18% of smartphone shipments in the first quarter of 2024, an increase of 2 percentage points over the same period last year.

Although Apple's share of sales in the overall smartphone market fell 11%, the average unit price of an Apple phone rose to 900 US dollars, making it unique among the top five mobile phone manufacturers, and other manufacturers are also trying to develop high-end brands.

The picture above clearly shows the differences between Apple and other manufacturers. Hardware upgrades are becoming more and more homogenous, and it is really difficult to get higher rewards without competitive differentiation. As a chef, you have to learn how to cook high-end cuisine.

Before AI phones, people bet on folding screens. However, the folding screen track seems to roll up too much, and manufacturers can earn a rich return on investment from it; there are still more monks and less porridge. According to TrendForce market forecasts, shipments of folding screen phones may reach 17.7 million units in 2024, an increase of about 11% over the previous year.

What concept?

The share of folding screens in the overall smartphone market only increased slightly to 1.5% from 1.4% in 2023, which is still a long way from the mass market.

However, high-end folding screens cannot escape the fate of deflationary products. The price of the folding screen has dropped from the initial 10,000 yuan to the 7,000 yuan price range. Judging from the performance configuration, it has not had much impact on the Apple Pro series, but the appearance may cause significant usage costs.

On the other hand, Android manufacturers have accelerated the deployment of generative AI since last year, and more than 10 smartphone brands have launched more than 30 smartphones with generative AI functions.

Compared to folding screens, Apple's entry into the generative AI circuit can be described as fast, and the hardware base that has already been built can finally be fully effective.

According to the understanding of an AI mobile phone, according to the concept and actual form of implementation, first, the phone must be able to install a large model on the side; secondly, the AI processing speed must not be too slow, otherwise it will seriously affect the user experience, so it must have a chip unit that can specialize in AI inference calculation. The SoC contains NPU computing power.

The number of large model parameters that can be mounted on the terminal side can only continue to rise if the AI computing power of the mobile SoC increases. According to the definitions given by OPPO and IDC, the performance of an NPU using an SoC that can run the end-side GenAI model faster and more efficiently is at least 30 TOPS using the INT-8 data type. As of February of this year, only Apple A17 Pro, MediaTek Tianji 9300, and Qualcomm Snapdragon 8 Gen 3 meet this requirement.

In addition to computing power, once the number of parameters increases, storage capacity and power consumption will inevitably increase, and DRAM/NAND, power modules, and heat dissipation related materials will all require upgrades.

According to reports, currently only the iPhone 15 Pro series equipped with the A17 Bionic chip, and iPads and Macs equipped with M1 or higher chips can run “Apple Intelligence” perfectly. Therefore, if you want to use it, you can only replace it with the latest one; you still have to use the top one. Naturally, the price will not be low.

However, if there is a lack of applications that require higher performance adaptations, even if the configuration is full, it will eventually face excess, especially in today's environment where special emphasis is placed on cost performance.

The most famous Andy Beale law in the PC era means that no matter how much performance the hardware (CPU) increases, it will eventually be consumed by the software (operating system). The CPU performance of the old version is insufficient to support the new version of the operating system, which can drive the entire industry's demand for equipment changes.

Currently, AI application scenarios mainly focus on document summarization, image processing, information retrieval, and text translation. In fact, we can't talk about disruption, or even just what is needed, and the cost of implementing these functions is also quite expensive.

As a result, mobile phone manufacturers are now betting on AI. It is no longer just a simple feature update, but an iteration in line with technology, market, and application development trends. The aim is to promote a new wave of switching cycles, raise the ceiling on hardware configurations, and keep mobile phone prices moving upward.

AI phones are not mature enough; they require innovative application scenarios that users can sense and bring value. Can Apple, which already has the best hardware configuration, open a new door for the industry tonight?

Behind the market capitalization duel, will the AI trend tilt again?

Last Wednesday, Nvidia's stock market capitalization surpassed 3 trillion US dollars and officially surpassed Apple to become the second-largest technology giant by market capitalization in the US stock market. Who would have thought that a company that “sells shovels” could make great strides in nearly a year, defeating the consumer tech giant that has taken over hundreds of millions of end user portals.

But in the age of computing power, the shovel itself is gold. Nvidia's monopoly position in the field of AI training and limited chip production capacity gave Nvidia a ticket to the big model competition.

Meanwhile, Apple's stock price on the other side was once lost in poor sales. Amid negative news about price reduction promotions, the company's revenue for the first three months of this year reached 90.8 billion US dollars, down 4% year on year. Among them, Apple's revenue in Greater China fell 8.08% year on year.

Since there is always a discrepancy between the speed of iteration of new products and market expectations, although the demand for this batch of high-end upgrades has made Apple more profitable, in the end, it still has to bow to fierce market competition.

Apple's iPhone revenue fell by nearly 10% in the first three months of this year, and sales volume fell 9.6%. Among them, sales volume in China fell 6.6%, and due to promotional price cuts, the market share fell 4.4% from the previous quarter.

Apple switched 180 degrees from a car building project to a generative AI circuit last year in order to once again seize the wave of high-end upgrades and raise the ceiling again, and not only just mobile phones, but also a new upgrade to a basket of iPad and Mac products. The core is AI OS.

Based on past experience, huge changes in mobile phones often begin with the way they interact. When a smartphone replaces a feature phone, it uses a touch screen instead of a keyboard; if an AI phone replaces a smartphone, it will replace the touch screen with voice.

This year marks WWDC's 40th year.

At the first WWDC, Apple unveiled the world's first personal computer (Lisa) with a graphical interface. In 2007, the original operating system, iPhone OS, was officially unveiled. This is the starting point for Apple to build the App Store software ecosystem. Later, the App Store expanded from the initial 500 apps to 5 million, forming an ecological moat. Until 2010, after four major version updates, we finally met with us under the name iOS.

Because Apple has a relatively capable chip configuration and a natural and smooth operating system, exploring AI functions and new applications should be the easiest to implement quickly.

In October of last year, Apple released the Ferret multi-modal model in a low-key manner. Compared with other well-known multi-modal models, Ferret's main advantage is its ability to recognize and locate images in the operating interface. Simply put, when users talk to the AI assistant about information in the interface, it can quickly understand what you mean and then find the corresponding elements.

This UI-like model doesn't seem powerful, but it's extended a bit further. If the user's instructions, text, or speech are split into specific workflows to execute, eventually it can also be operated across apps, then an AI-native operating system will basically be implemented. This coincides with the final form of AI phones; Apple has already laid some foundation.

According to this concept, AI OS makes interaction between users and applications better, more efficient and faster. With the improvement of end-side model capabilities, AI gradually integrates deeply with operating systems from some native applications and connects to third-party apps, replicating app store innovations in 2007, and new rules may be ready to come out.

If third-party apps want to be used more in the future, or even increase the frequency of payments and advertising, the Apple ecosystem may take more initiative. As the penetration rate of AI applications increases in the future, whether it is charging users or taxing developers, Apple is on the strong side.

Although worried that sales will continue to decline, the market never questioned Apple's ability to create high-end products. In the first half of the year, Apple made frequent generative AI moves, and stock prices slowly recovered. Despite Nvidia's strong momentum, the gap was still not bridged.

However, whether the AI phone gimmick can guide a new wave of switching, allow the high-end price market to continue to grow, and whether Apple's entry into the market can catalyze the development of AI applications has many areas that need to be verified, so companies in the supply chain are still worth paying attention to.

If it goes well, will the more valuable part of the AI industry chain shift to operating systems and software ecosystems in the future?

This is a press conference on market capitalization prospects, and Apple has begun to respond! (End of full text)