Apple Could Bring More Powerful AI to iPhones Through PrismML Partnership
Apple may be preparing for a major leap in on-device artificial intelligence. According to recent reports, the tech giant is in early discussions with AI startup PrismML, a company that has developed technology capable of dramatically shrinking large AI models so they can run directly on an iPhone instead of relying on cloud servers.
Why This Matters
One of the biggest challenges in AI today is that advanced language models require enormous amounts of memory and computing power. Most smartphones simply aren’t capable of running these models locally, forcing companies to process requests in the cloud.
PrismML claims to have solved a large part of that problem by compressing a 27-billion-parameter AI model from roughly 54GB to under 4GB, making it small enough to run on modern iPhones while still retaining much of its intelligence.
What Apple Could Gain
If Apple adopts this technology, future versions of Apple Intelligence could become significantly more capable while continuing to prioritize user privacy. Running AI directly on the device would offer several benefits:
Faster responses with reduced latency.
Better privacy since more data stays on the iPhone.
Less dependence on internet connectivity.
Lower cloud computing costs for Apple.
Improved battery and memory efficiency through optimized AI models.
These improvements align closely with Apple’s long-term strategy of performing as much AI processing as possible on-device.
The Talks Are Still Early
PrismML CEO Babak Hassibi confirmed that Apple is evaluating the company’s technology, but emphasized that discussions remain in the early stages. There is currently no confirmation that Apple will license the technology or acquire the startup, though conversations are reportedly progressing positively.
What It Could Mean for Future iPhones
If the partnership eventually moves forward, future iPhones could feature:
Smarter and more conversational Siri.
Advanced Apple Intelligence features that work entirely offline.
More capable AI tools for writing, coding, translation, and image understanding.
Faster AI performance without sending personal data to remote servers.
Such a move could help Apple differentiate itself in the growing AI race by combining powerful performance with its long-standing focus on privacy.
Geeksterr’s Take
The AI battle is no longer just about building bigger models—it’s about making those models efficient enough to fit inside the devices we carry every day. If PrismML’s compression technology lives up to its promise, Apple could unlock a new generation of iPhone AI that is faster, more private, and available even without an internet connection. While nothing has been finalized, this is one of the most intriguing AI developments to watch as Apple continues shaping the future of Apple Intelligence.
