Apple Eyes Startup That Squeezes AI Models onto iPhones
Apple is in talks with PrismML, whose compression tech runs AI models using up to 15x less memory on-device.
Apple is reportedly in talks with PrismML, a startup that has cracked a genuinely interesting problem: making large AI models small enough to run natively on an iPhone. If you've been frustrated watching Apple play catch-up on AI, this deal could be the move that changes things.
PrismML's core trick is aggressive model compression. Their compressed version of Alibaba's Qwen model runs using up to 15 times less memory than the original. That's not a rounding error — that's the difference between an AI feature that lives in the cloud and one that fires instantly on your device, with no server ping and no privacy trade-off.
Read more Zipline Recruits Tesla, Uber, Waymo Vets to Scale Drone Delivery →
On-device AI is Apple's whole strategy right now. The company has been betting hard that users will prefer private, local AI over cloud-dependent alternatives. The problem has always been raw horsepower — or more precisely, memory. PrismML's compression approach directly attacks that bottleneck, which is exactly why Apple would be circling.
For traders and tech investors, the angle here is straightforward. If Apple closes this deal, it accelerates the timeline on meaningful on-device AI capabilities across the iPhone lineup. That's a potential catalyst for upgrade cycles — and a direct challenge to competitors leaning on cloud infrastructure to power their AI features. Watch how this plays out heading into Apple's next major hardware announcements.
Continue reading at US Top News and Analysis
