Qualcomm CEO says AI firms want 100B-parameter local LLMs on phones by 2028
Cristiano Amon reports that artificial intelligence companies are designing mobile devices and requesting support for massive local models.
Key points
- Qualcomm CEO Cristiano Amon says AI companies are starting to design smartphones.
- These firms are requesting support for running 100B-parameter local large language models continuously.
- The target timeframe for this continuous local model execution is the year 2028.
Qualcomm CEO Cristiano Amon says that artificial intelligence companies are starting to design smartphones. According to a report by Techmeme citing Alex Heath and unnamed sources, these AI firms are asking Qualcomm to support the continuous operation of 100B-parameter local large language models by 2028.
This shift indicates a change in how mobile hardware is being conceived. Amon highlighted that AI companies are moving beyond software integration and are now starting to do phones. They are specifically requesting that Qualcomm enable these devices to run massive local models without interruption.
Strategic Shift in Mobile Hardware
The request for 100B-parameter model support represents a significant increase in computational requirements for mobile processors. Traditionally, such large models have been hosted in data centers due to their intensive resource needs. By pushing for continuous local execution, AI companies are demanding that mobile chips handle workloads previously reserved for server-grade infrastructure.
Amon provided this update while discussing broader industry trends, including the evolving sentiment regarding AI development in China versus the United States. He also addressed how artificial intelligence is changing smartphones and wearable technology.
Impact on Data Centers and Connectivity
Qualcomm’s data center strategy was also a topic of discussion. Amon discussed the company's approach to data centers alongside other industry trends. The integration of advanced AI into consumer devices is driving changes in both mobile and server computing architectures.
Looking ahead, Amon discussed what 6G will unlock. The next generation of wireless connectivity is expected to bring new capabilities, though specific details on interactions between local models and cloud resources were not provided in the source material.
What should you do?
Security and technology professionals should monitor Qualcomm’s hardware roadmaps for changes in processing capabilities. As local AI models grow in size, the attack surface for mobile devices may expand. Organizations should assess their mobile device management policies to ensure they can handle increased computational loads and potential security risks associated with local AI processing.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: Qualcomm
- Review mobile device specifications to understand current and future AI processing capabilities.
- Assess security policies for devices running local large language models.
- Monitor vendor announcements for updates on hardware support for AI workloads.
- Evaluate the impact of increased local processing on device battery and thermal management.
Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What are AI companies asking Qualcomm to support?
AI companies are asking Qualcomm to support running 100B-parameter local large language models continuously by 2028.
Who is designing smartphones according to Cristiano Amon?
Cristiano Amon stated that AI companies are starting to do phones.
When will these local LLMs be running continuously?
The target timeframe for continuous local LLM execution is 2028.
