Since the launch of ChatGPT, generative AI models have become a new direction of competition for technology companies at home and abroad, such as Baidu, Alibaba, Tencent, Google, Meta, and OpenAI. However, in the face of the rising wave of generative AI, Apple, as the world's leading technology company, has a very low presence, like a bystander, silently watching the development of events.
"Later this year, I'll share with you how we're breaking new ground in generative AI, a technology we believe can redefine the future."
It wasn't until February this year that Cook's speech at Apple's annual shareholder meeting really released a signal to the outside world that Apple was fully committed to generative AI technology. Although Apple did not disclose any AI progress, as the WWDC2024 Worldwide Developers Conference approaches, more and more information about Apple's AI has been exposed.
Apple's latest AI strategy was exposed, abandoning purely local processing
Bloomberg reporter Mark Gurman posted that Apple plans to put high-end chips, including M-series chips, into cloud computing servers to process computationally difficult AI functions, while relatively simple AI-related functions are directly processed locally on terminal devices such as iPhones, iPads and Macs. In addition, Mark Gurman revealed in today's public news that Apple has reached an agreement with OpenAI, which will provide it with cloud-based AI technology support.
According to reports, Apple's first batch of AI server chips use the M2 Ultra released last year, and the M4 chip on the recently released new iPad Pro will also be used in AI servers in the future. At the same time, Apple is developing its own exclusive chip for AI servers, which is expected to use TSMC's 3nm process technology, and will be mass-produced in the second half of 2025 at the earliest.
Source: Apple
Independence and closure have always been synonymous with Apple's ecology, and many users have chosen Apple products for this purpose. That's why Apple has been prioritizing the route of local device processing when embedding generative AI to ensure user privacy and security. At this stage, the hardware of mobile devices such as mobile phones and tablets is difficult to bear the computing pressure brought by the pure device-side large model solution, and Apple's AI strategy of preparing for cloud computing servers means that Apple will abandon the pure local processing solution and adopt a more efficient cloud combination solution.
The shift in AI strategy does not mean that Apple does not take the security of user data seriously, on the contrary, Apple has invested hundreds of millions of dollars in a data security solution called "Secure Enclave" in the past three years, which aims to protect user privacy by insulating data from security vulnerabilities by components inside the processor.
There are many strong competitors in AI terminals, and what does Apple rely on to achieve a breakthrough
We don't know the level of data security of this solution until it is actually implemented, but it is likely that Apple's AI strategy change is to accelerate the AI of devices to catch up with the AI process of other mobile phone and computer manufacturers.
Taking smartphones as an example, various mobile phone manufacturers have embraced the AI wave, and the large model on the side of the mobile phone has almost become the flagship standard, and even AI mobile phones have appeared in the mid-range and thousand-yuan machines. In the past, Apple's iOS ecosystem was self-contained, and the rest of the manufacturers mostly used Android, so that the difference between them was not obvious, but in the context of the AI wave, manufacturers in the Android camp have begun to build their own exclusive AI ecology, seeking opportunities for corner overtaking.
Glory magic model, OPPO's Andean model, vivo's blue heart model...... Among the top five brands in the domestic mobile phone market, except for Apple, the rest of the brands have released their own end-side large-scale model products. This is not the case in the domestic mobile phone market, not to mention the wider global mobile phone market.
Source: Lei Technology
If Apple wants to break out of the encirclement in the AI terminal market, which is full of strong competitors, it may be difficult to follow the old path of other manufacturers, and it may be wise to find another way.
According to industry sources, Apple does not plan to launch a chatbot to compete directly with the ChatGPT-like large model, and the Siri voice assistant after AI recasting will introduce a new generative AI system to help Siri better understand user intent, understand complex instructions, and provide detailed and personalized responses in a natural conversational way.
Siri is one of the earliest intelligent assistants in mobile phones, and its intelligence level was still stuck in the past when it was born in 2011, so it will make many jokes about misunderstandings. Therefore, reinventing Siri from the underlying technology is a necessary step for Apple, which aims to embed generative AI.
Source: UNsplash
The new Siri will have the ability to have continuous conversations, no longer in the initial stage of asking and answering questions. At the same time, Siri will integrate system applications such as notes, text messages, memos, and Safari embedded with generative AI functions, so that users can better call different functions.
When Apple advertised the new Siri, it said that Siri would handle all requests on the device side, which means that Apple listed the new Siri as a relatively simple AI-related function, which only needs to be handled locally. At a special event in May, Apple showcased the iPad Pro's AI applications for video editing and painting, and although the iPhone 15 series upgraded its memory to 8GB ahead of schedule, the pressure of the iPhone to run purely locally could affect the user experience.
In this way, it can explain why Apple adopts the strategic layout of cloud integration, which may be prepared for other creative tools and productivity functions in the ecosystem.
New Siri, creative tools, and productivity features are Apple's new application directions in the AI ecosystem. In the era of AI, the relationship between mobile phone manufacturers and developers has a faint momentum of reversal, and mobile phone manufacturers represented by Apple previously grasped the power of whether third-party apps should join the ecosystem, and now it is the turn of mobile phone manufacturers to think about how to integrate more third-party apps into their own large models. Continuing the ecological advantages of the traditional smartphone era will determine whether Apple's AI can stand out in the new round of AI ecological warfare.
Apple is not willing to be a bystander anymore, and AI has already been laid out
In fact, although Apple has been free from the competition of generative AI, it has a lot of layout in the field of AI, including internal self-developed AI model frameworks, acquisition investment, hardware procurement, etc.
Looking back at Apple's exploration of generative AI, perhaps it can be traced back to the CoreML framework announced in 2017, which is used to deploy trained AI models to Apple devices, which is also considered by many media to be the beginning of Apple's relationship with generative AI.
In the past two years, Apple has established Ajax, a large language model framework, and released OpenELM's self-developed large model, which shows that Apple has not stopped developing its own AI model. Not only that, but Apple has also acquired and invested in a number of AI companies. As of 2023, Apple has acquired 32 AI startups, far more than Google's 21 and Meta's 18. Through the acquisition of many promising AI startups, Apple has gained core innovative technology and top talent, not to mention how many AI technologies are actually used in Apple devices, at least in the acquisition of AI startups, Apple can be said to be far ahead.
Source: Medium
In terms of hardware procurement, Apple is still continuing to use its banknote capabilities and actively purchasing chips in preparation for the promotion of generative AI. Analyst Ming-Chi Kuo said that Apple has invested a lot of money in the purchase of AI servers, and Apple will purchase 2000~3000 units in 2023, with procurement spending of at least 620 million US dollars, and it is expected that by 2024, this figure will reach 4.75 billion US dollars. It plans to purchase an additional 18,000~20,000 units, accounting for 5% of global AI server shipments.
Hoarding AI servers is naturally to prepare for cloud computing, which shows that Apple has long set an AI strategy for cloud collaboration, and Cook said that redefining the future of Apple AI is not just a slogan.
Judging from the above layout, Apple is trying to catch up with its competitors, and for a long time in the past, Apple has endured the impact of falling behind. Shortly after the start of 2024, Apple's sales in the Chinese market fell by 24%, and its market ranking was surpassed by vivo, Huawei, and Honor, falling from first to fourth. At the same time, Apple has been fined $2 billion in the European Union for antitrust issues, and the North American market is also facing strong competition from Samsung's S24 series. This series of questions highlights the unprecedented challenges that Apple faces in the global smartphone market.
Apple made up for the lost opportunity in the field of artificial intelligence and got rid of the label of "AI bystander", which determines the direction of Apple in the next decade. The answer to all this depends on the next WWDC2024.
In 2024, the implementation of AI large models will be accelerated, cars will become a golden scene, and "the second half is intelligent" will become the consensus of the industry.
On April 25, the Beijing Auto Show became a watershed for new energy vehicles to move towards the second half, with intelligence becoming the absolute protagonist, domestic brands rising strongly, entrepreneurs "new car models" becoming landscapes, and fuel vehicles, joint venture vehicles, foreign brands, and functional vehicles completely lonely.
During the auto show, Lei Technology's new media tram of "pay attention to electric vehicles and understand intelligence better" sent a luxury reporting group to Beijing to conduct a systematic front-line professional in-depth report, with a total of 8 videos, 8 in-depth analysis and massive short content, the whole network received millions + readings, and won a number of WeChat 100,000+ explosive models (real traffic), which attracted great attention in the professional media of the industry.