Apple Advances On-Device AI Capabilities
Apple Inc. (AAPL) is markedly intensifying its commitment to on-device artificial intelligence, a strategic move poised to enable robust AI models to operate locally on iPhones.
The Chief Executive Officer of the Silicon Valley-based AI startup, PrismML, disclosed that Apple is pursuing a technical collaboration.
The objective is to harness advanced model compression technologies that would facilitate the execution of high-performance large models directly on iPhones, aiming to surmount existing hardware limitations in on-device AI.
PrismML has made headlines with the launch of a compressed iteration of Alibaba’s Tongyi Qianwen (Qwen) model.
This newly optimized version reduces the model’s size from an impressive 54GB with 27 billion parameters to under 4GB, thereby enabling it to function entirely on the iPhone 15 and subsequent models.
This transformation not only conserves memory usage by over 90% but also enhances inference speeds by six to eight times and significantly lowers power consumption by three to six times.
Currently, Apple is undertaking live testing and validation of this cutting-edge technology, concentrating on metrics such as execution speed, energy efficiency, and compatibility for on-device applications. Negotiations between the two entities remain in their infancy, yet progress appears to be steady.
Market analysts postulate that the adoption of this lightweight model technology could enable Apple to transition demanding functionalities—such as computational photography, video generation, and health data analysis—to the device level itself.
This shift would further bolster its privacy-centric advantages. With its vertically integrated ecosystem of proprietary hardware and software, Apple possesses an intrinsic advantage in adapting and deploying these models.
There is a strong likelihood that Apple will adopt a hybrid framework incorporating local processing for everyday tasks coupled with cloud support for more intricate operations.
While some experts argue that lightweight models might diminish the need for cloud computing resources, a dominant perspective among institutions suggests that advancements in technology will merely reallocate processing power from the cloud to individual devices.
This transition is not anticipated to lessen overall chip demand; rather, enhanced efficiency will likely propel an augmentation of AI application scenarios, stimulating broader growth in demand.
Nevertheless, industry insiders caution that this technology necessitates thorough testing through large-scale practical applications; questions regarding its impact on battery longevity, stability, and sustained performance remain unanswered.
For Apple, prospective demand for hardware upgrades represents a compelling investment narrative. Estimates from Morgan Stanley indicate that nearly 850 million iPhones globally lack the capability to support Apple Intelligence, and an additional 1.3 billion devices will be incompatible with the forthcoming AI-driven Siri. This potential upgrade spectrum significantly surpasses previous cycles of product iteration.
As AI interaction evolves into a fundamental mode of engagement between users and their smart devices—encompassing tasks like placing orders, managing schedules, and coordinating digital responsibilities through Siri—the imperative for hardware enhancements will surpass that of mere software updates.
This paradigm shift aligns perfectly with Apple’s core growth philosophy, where software innovation indeed drives hardware sales.
Source link: Tradingkey.com.
