Apple rushes towards AI, senior executives closely visit TSMC. According to industry reports, Apple's Chief Operating Officer Jeff Williams recently made a low-key visit to TSMC (2330) in Taiwan. TSMC President Wei Zhejia personally received the visit, and both sides focused on Apple's development of self-developed AI chips and will subcontract TSMC's advanced process production capacity to produce related chips, adding strong momentum to TSMC's orders.
Faced with inquiries from reporters, Apple did not respond immediately. TSMC has consistently not commented on market rumors or single customer messages, and did not comment.
Corporate analysis shows that Apple is actively developing innovative applications and needs more support from the latest semiconductor technologies. Previously, it had taken the lead in contracting the first batch of TSMC's 3-nanometer production capacity. If the first batch of production capacity for 2-nanometer or even more advanced processes is scheduled in the future, it is estimated that Apple's contribution to TSMC's annual revenue and output value will steadily increase. This year, there is a chance to reach a new high of NT $600 billion, which is a long-term challenge of trillions of yuan, setting a new milestone.
Regarding data center investment, Apple's CFO Luca Maestri recently stated at a financial report meeting that he has his own data center in this field and also uses third-party technology. This model has shown good results and the plan continues to advance. He also mentioned that he is excited about the generative AI business opportunity, and the company has invested over $100 billion in related research and development over the past five years.
It is understood that in addition to the A-series processors used in iPhones, Apple has been cooperating with TSMC for many years. In recent years, Apple Silicon has launched a long-term plan to create M-series processors for MacBooks and iPads, and the key driver is Williams. This time, Williams came to Taiwan to discuss a new wave of package production capacity and cooperation with TSMC for Apple's self-developed AI chips, which has attracted special attention from the industry.
The M-series processors have successfully sparked a market whirlwind, and since the end of 2020, Apple has gradually converted Intel processors from its Mac series product line into its own M-series chips. This successful model has enabled Apple to focus on designing and developing more competitive products. At present, we are developing our own AI chips for our own data center servers, with the aim of optimizing the performance of some of the current Huida architecture data center chips.
Apple has begun to enter the AI server side and develop its own AI computing processors in a semi customized form using the Arm architecture. In addition to mass production using TSMC's advanced processes, the computational efficiency of AI servers will affect computing power. At that time, it will be built using TSMC's most expensive SoIC advanced packaging process, and the processors will be integrated in a 3D stacking manner. However, due to high costs, Apple should not have plans to expand to terminal devices in the short term.
However, in response to the booming development of new technologies in generative AI, in addition to a significant increase in cloud computing demand, terminal devices are also beginning to have edge computing needs. Apple is following this trend, especially in the M4 system built with TSMC's 3-nanometer process, which integrates more powerful neural network engines in a small chip manner. Although the integration method is similar to the previous generation, the computing speed is expected to be twice as fast as before.
Next, Apple will also develop three different levels of M4 processors, codenamed Donnan, Brava, and Hidra, to fully leverage the AI PC business opportunity. It is expected to enter the mass production phase of TSMC in the second half of this year, which will be controlled and tested by Sun and Moon.