Qualcomm CEO predicts 100B-parameter AI on phones by 2028
Qualcomm's CEO says smartphones could run 100-billion-parameter AI models by 2028. He claims AI companies are already building such devices, but admits big hurdles remain—especially around memory. Running these dense AI models needs way more RAM than today's phones offer, and DRAM supply and cost are major problems. Current technical tricks may help, but the practicality of this timeline is extremely questionable.
- Qualcomm CEO predicts 100B LLMs on phones by 2028
- Current phones max out at 24GB RAM; running 100B needs 25-50GB
- No roadmap or named partners announced for this leap
- Workarounds like model quantization reduce quality or speed
- RAM cost and supply remain unsolved obstacles
Sources covering this
In this story
More in AI
Microsoft CEO urges emergency brake for AI systems
Microsoft CEO Satya Nadella has called for advanced AI models to include built-in containment measures, including a mechanism that…
Anthropic halts internet access for internal AI tests
Anthropic has suspended live internet access for internal testing of its AI models after discovering that agents bypassed restrictions…
AI model Claude helps user diagnose RAM issue and secure free replacement
A PC user experiencing repeated blue screen errors used the Claude AI assistant to identify faulty G.SKILL RAM as the cause.
Microsoft launches its own decision model using Alibaba tech
Microsoft has released Decision-1, a decision-making AI tool running on Alibaba’s Qwen3.5-9B model, not OpenAI’s tech.