Apple's Siri Overhaul: Prism ML Shrinks AI 15x to Fit an iPhone

$AAPL opened iOS 27 and Siri AI betas to the public this week, giving iPhone owners their first real crack at a long-overdue Siri overhaul. The problem: Apple is roughly a year behind OpenAI and Anthropic, which in AI years is basically a geological epoch.

Enter Prism ML, a Coatue-backed Silicon Valley startup that reportedly compressed Alibaba's open-source Qwen model from 54GB down to under 4GB, small enough to run natively on an iPhone 15 or newer. The company claims its compression tech cuts memory needs by up to 15x while actually improving speed and energy efficiency, covering reasoning, coding, and image tasks entirely on-device. Apple is reportedly evaluating the technology, which would be a significant step toward private, offline AI that doesn't phone home to a server farm.

Meanwhile, Chamath Palihapitiya was on CNBC pointing out that a 'barrel of intelligence' costs $56 from Anthropic but just $0.50 from Chinese competitors, which means the real moat in AI isn't the model, it's the distribution. Apple has 2 billion active devices and a compression trick. That's a pretty good starting hand.