On Monday (June 5th) Eastern Time, the highly anticipated Apple (AAPL, stock price $179.21, market value $2.82 trillion) Worldwide Developers Conference kicked off with the theme “Code new worlds”.

However, in this two-hour speech, it largely ignored the hottest topic of artificial intelligence (AI) today, raising questions among investors: What is Apple’s attitude toward AI?

(资料图片)

Photo/Apple website

NBD noticed that in fact, AI was hidden in some hardware and software updates at this developer conference. When introducing the latest upgraded iOS17 system, Craig Federighi, senior vice president of Apple’s software engineering, announced the use of Transformer language models for input and speech recognition of text.

Photo/Apple Website

Not until March of this year that it gradually opened up AI-related job recruitment. NBD found that Apple is currently recruiting about 600 positions, of which 97 are related to AI. More than one-third of them are positions newly established in May. In just the first three weeks of May this year, Apple released 28 AI-related positions, including senior engineers, research scientists, special project managers and so on. In addition, Apple is also testing new natural language generation functions for Siri.

In contrast, Google and Microsoft currently have 509 and 235 AI-related positions open for recruitment respectively.

Will Wong, senior research manager at technology consulting firm IDC in Singapore, pointed out in an interview with NBD that it’s too early to judge Apple’s success in the generative AI race, especially since Apple always makes sure its products and technology are ready to provide a favorable consumer experience before they are launched.

Efficiency and privacy are two key points of its AI strategy. Efficiency refers to machine learning algorithms and models executed locally with faster response and better performance. Privacy protection is self-explanatory. The current generative AI seems to be contrary to Apple’s AI strategy.

Avi Greengart, founder and chief analyst of market research company Techsponential, also pointed out in an interview with NBD that in the case of generative AI, Apple appears to be behind Microsoft and Google. However, Apple can afford to wait, and then integrate the technology where it makes sense in its ecosystem.

Although Cook’s answer about AI at last month’s earnings conference call was extremely brief, he used a very interesting word to describe Apple’s achievements in the field of AI: weaving. Cook said that Apple has “woven” artificial intelligence algorithms into its various products.

For decades, Apple has been developing and iterating on its products’ AI. The embedded AI in Apple products may be so numerous that users are not aware of them.

For example, Apple has always hoped that Siri could become the most helpful AI assistant for Apple users. Therefore, Siri voice assistant has previously used natural language processing (NLP) technology on Apple devices to help users perform tasks. In addition, AI technology has also been deeply integrated into the company’s iOS software with features such as Deep Fusion to improve image quality. This machine learning-assisted camera feature can optimize and improve photo quality pixel by pixel within seconds.

It seems that Apple is doing AI but differently from giants like Microsoft and Google. Some believe that Apple is unlikely to try to create general large-scale language model artificial intelligence systems like ChatGPT or Google Bard.

With its early layout of chips, hardware and software over the years, it is fair to say that Apple is building infrastructure for current generative AI deployment rather than doing AI itself. Mac and MacBook now run on Apple’s own M2 Pro and M2 Max chips respectively with dedicated 16-core neural network engines designed for artificial intelligence and machine learning tasks. According to Apple, these chips are 40% faster than their predecessors.

Photo/Apple website

Apple’s M2 Ultra is currently the world’s most powerful chip built for personal computers. Mac is also currently the most capable computer on the market for running generative AI applications - its M2 processor can achieve an industry-leading 1.58 quadrillion operations per second (TOPS) with a powerful neural processing unit that could allow AI applications to run on devices without logging into the cloud.

Avi Greengart told NBD that Apple’s key advantage for AI in future products is its vertical silicon integration and reputation for privacy. Apple can build as much neural processing as needed onto its chips and then do AI processing on device rather than relying on the cloud.

Daniel Ives, managing director and senior stock analyst at US brokerage Wedbush, said in an email to NBD that “we believe Apple will eventually establish its own artificial intelligence ecosystem in the iOS developer community. The release of Vision Pro will bring more applications in the coming years. Although investors’ instinctive reaction will revolve around recent expectations, we believe this will further establish a developer moat for Apple and maintain its strong foundation and Apple ecosystem.” “Overall, we believe Apple will continue to double down on software. We maintain our outperform rating on Apple and our target price of $205,” Ives added.