cryptobriefing.com
Apple is collaborating with Broadcom to develop its first in-house AI server chip, internally codenamed Baltra, according to The Information, citing people with knowledge of the matter.
Expected to be ready for mass production by 2026, the Baltra chip is designed to accelerate AI tasks, specifically focusing on inferencing. The core functionality will allow the chip to process new data and interact with large language models (LLMs) to generate outputs. This will set it apart from AI training models.
Apple has chosen Taiwan Semiconductor Manufacturing Company (TSMC) to manufacture the chip. The world’s leading semiconductor foundry will use its advanced 3-nanometer process technology for production, the report notes.
The tech giant started working on developing its own server chips three years ago, but has ramped up efforts since OpenAI released ChatGPT in December 2022, according to a previous report from Bloomberg. ChatGPT’s success has prompted Apple to prioritize the development of its own AI chip to stay competitive in the AI market.
Apple wants to use high-performance chips, similar to those used in Macs, to power its cloud-based AI services. As part of the strategy, the company plans to integrate new AI features across its devices, enhancing functionalities like Siri and introducing advanced capabilities that improve the user experience.
Following news of the AI chip development, Apple’s stock briefly rose, touching $250 at market opening Wednesday, according to Yahoo Finance. Broadcom shares surged 5%.
Apple and Broadcom have a long-standing partnership, particularly in the area of semiconductor components. Last May, Apple announced a multibillion-dollar, multi-year deal with Broadcom to develop 5G radio frequency components and other wireless connectivity technologies. The agreement is part of Apple’s strategy to source more components from US manufacturers and reduce its reliance on foreign suppliers.