Key points
- AI companies require phones to run 100-billion-parameter models by 2028.
- Memory price surges have reduced smartphone sales by 20% this year.
- Qualcomm plans to open-source its Modular platform under Apache 2.0.
Qualcomm chief executive Cristiano Amon stated that leading artificial intelligence companies are demanding the ability to run models with at least 100 billion parameters directly on smartphones by 2028. Amon shared these requirements during an interview with Alex Heath for the Sources podcast, recorded at the Snapdragon Summit in Maui. The companies reportedly want these large models running continuously on the device, rather than relying solely on cloud infrastructure.
By the numbers
- 100 billion: the parameter count AI firms want on phones by 2028
- 20 percent: the decline in the smartphone market this year due to memory costs
- 5 to 6 times: the increase in memory prices affecting device manufacturing
- 1.2 billion: the annual number of phones purchased by consumers globally
- 50 to 60 million: the estimated market size for smart glasses
What the figures mean
Amon explained that AI companies are beginning to create their own phones, citing ByteDance’s Doubao device from Nubia as an example. He argued that personal agents need direct access to user context, sensor data, and local apps, which cannot always be handled by cloud-based virtual machines. While some processing will remain in data centers, Amon emphasized that critical user interactions must happen on the device. He noted that memory shortages have severely impacted mid-range and cheaper phones, though premium models remain resilient.
Operational impact
Technology leaders must assess how on-device AI requirements will affect their hardware procurement strategies and supplier relationships. Teams usually need to evaluate whether current device fleets can support increased local processing loads or if upgrades are necessary. Check whether existing security protocols account for local model execution and data handling. Communicate with engineering and procurement teams about potential supply chain constraints related to memory components. Ensure that vendor contracts address performance guarantees for emerging AI workloads.
What to do and how to stay safe: Qualcomm
- Review current device specifications to determine if they can support larger local AI models.
- Monitor memory component pricing trends to anticipate budget impacts on hardware refresh cycles.
- Evaluate the security implications of running large language models locally on employee devices.
- Identify vendors offering solutions that balance on-device processing with cloud offloading.
Step-by-step guide: Shadow AI Explained: 8 Questions Answered for Secure Operations
General security guidance from the BitShield News newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Which AI companies want 100-billion-parameter models on phones?
Qualcomm CEO Cristiano Amon would not name the specific companies but said they are at the forefront of AI.
How have memory prices affected the smartphone market?
Memory prices have risen five to six times, causing the smartphone market to drop 20% this year.
What is Qualcomm doing with its Modular platform?
Qualcomm plans to make its Modular platform open source under the Apache 2.0 license for any chip.
