en.Wedoany.com Reported - Apple has extended its custom application-specific integrated circuit (ASIC) supply partnership agreement with U.S. semiconductor company Broadcom to 2031, aiming to enhance its artificial intelligence infrastructure competitiveness.
To improve the cloud computing efficiency of its AI service "Apple Intelligence," Apple has disclosed a multi-year collaboration with Broadcom in the ASIC field. Broadcom, which previously supplied cellular and wireless communication chips to Apple, has now extended its partnership to co-design core computing silicon components for future generations of Apple products. Through this agreement, Apple secures a stable supply line for server and device components into the early 2030s, amid global semiconductor procurement supply chain uncertainties.
Establishing long-term supply chain alliances is part of Apple's strategy to reduce reliance on external GPUs from companies like NVIDIA and achieve full vertical integration of its data center infrastructure. Currently, Apple Intelligence servers still use general-purpose consumer-grade processors such as the M2 Ultra from Macs. With the surge in AI inference workloads, early deployment of large-scale custom server silicon chips has been prioritized. Apple is developing its first data center AI inference chip, Baltra, using TSMC's advanced 3nm N3P process, and is integrating Broadcom's high-speed interconnect and networking technology assets into the chip architecture. The chip is slated for mass production in the second half of 2026 and will be installed in Apple's self-built data centers by 2027.
Apple has also established a buffer against potential component shortages from its in-house wireless chip roadmap. Apple recently equipped the iPhone 16E model with its first self-developed Wi-Fi/Bluetooth integrated chip, N1, and its self-developed 5G modem, C1, in an effort to reduce dependence on Broadcom. After realizing the difficulty of matching Broadcom's high-frequency RF filters and wireless communication yield rates in the short term, Apple continued its in-house wireless chip projects while breaking negotiation barriers by adding high-value AI custom chip collaborations.
Apple's active adoption of external design assets also aligns with its software diversification strategy, such as building a generative AI joint front with Google. Apple recently introduced Google's large language model, Gemini, to enhance cloud-based AI and next-generation Siri performance, signing a $1 billion annual licensing agreement. In its new AI server infrastructure, assembled at Foxconn's North American factories, Apple plans to simultaneously leverage Broadcom's custom chip assets and Google's software engine to maximize performance advantages.
