Small Models, Scarce Memory: Edge AI as the Answer to the AI Supply Crunch

AIoTHub (formerly IoT North) Community meetup are pleased to present Manisha Arora in "Small Models, Scarce Memory: Edge AI as the Answer to the AI Supply Crunch"

Abstract:
AI data centres are absorbing around 70% of the world's high-end memory output, and the bill is arriving everywhere else: DRAM contract prices up 90–95% in a quarter, eMMC allocations cut to 10–20% of requested volume, memory lead times out to 52 weeks. For MSMEs, SMEs, telecom vendors and manufacturers, that is a design problem, not just an inflation problem. This session explains what is happening to memory, energy and water, then sets out three practical responses, smaller and quantised models with LoRA/QLoRA adaptation, deliberate hardware selection, and smarter memory sourcing

Bio:
Manisha Arora is a Cloud, Data, AI and Enterprise/Solution Architect and Consultant with 19+ years of international experience spanning technology, product innovation and enterprise transformation.
She advises organisations on Generative AI, Agentic AI, RAG, multi-cloud architecture, AI governance and sovereign AI, championing responsible, scalable adoption. A guest speaker at leading UK universities on AI and Ethics, she also contributes to mentoring, STEM and NHS initiatives.
Recognised among Women in Tech 100 (2024), 100 Power Women to Watch and Women in AI Ethics, Manisha explores the intersection of technology, psychology and human cognition to help shape an AI future that is intelligent, ethical and genuinely human.