Apple is in talks with the startup shrinking a 27B AI model onto an iPhone
What happened
Apple is in early discussions with PrismML, a startup that compresses large AI models to run directly on smartphones. PrismML’s technology shrinks a 27 billion parameter AI model to fit and operate entirely on an iPhone. This on-device AI approach aims to reduce reliance on cloud computing for tasks like powering Apple’s Siri voice assistant. PrismML’s CEO Babak Hassibi confirmed that Apple and other tech companies are actively evaluating the startup’s compression and inference solutions.
Why it matters
Running large AI models on-device changes the economics and speed of personal assistants and AI features. By avoiding the cloud, Apple can reduce latency and energy costs, improve user privacy by keeping data local, and lower Siri’s dependency on internet connectivity. For Apple, a stronger push toward on-device AI tightens control over the user experience and data security, areas where hardware and software integration matter most. This also flags a shift in the competitive landscape as cloud-first AI solutions face pressure from shrinking model technologies.
What to watch next
The key question is whether PrismML’s technology meets Apple’s strict efficiency and performance standards for consumer devices at scale. Watch for announcements of broader integration with iOS features beyond Siri. Additionally, other hardware vendors and AI startups will likely accelerate their own on-device AI compression to avoid falling behind. Investors and operators should monitor how this move influences cloud service demand and the pace of AI hardware innovation in mobile markets.
AI Quick Briefs Editorial Desk
