In the latest episode of IoT Unplugged, Pete Bernard, CEO, Edge AI Foundation discusses the evolution and future of Edge artificial intelligence (AI).
The Edge AI Foundation, originally established as the Tiny ML Foundation in 2018, aims to support the development of AI models for highly constrained environments – running small-scale models on devices such as microcontrollers and sensors. Over time, advancements in semiconductor technology and AI model sophistication have broadened the Foundation’s scope, leading to its rebranding as the Edge AI Foundation to reflect its expanded technical mission. Today, the Foundation is a non-profit focused on advocacy and education in the Edge AI sector, supported by leading technology companies, and dedicated to empowering professionals through scholarships, open-source university curricula, and global educational initiatives.
A core distinction Bernard draws is between Edge AI and Cloud AI. Edge AI refers to running AI models where the data itself is generated – outside centralised, multi-tenant data centres. This has tangible advantages, as approximately 75% of all data is created on the Edge (through sensors, cameras, manufacturing equipment, etc.), and processing data locally allows for higher impact, reduced cost, improved privacy, and lower power consumption. Unlike high-profile, Cloud-based Generative AI systems – like chatbots powered by massive data centres – Edge AI is embedded in real-world scenarios. Examples include agriculture, safety and security monitoring in municipalities, water quality analysis, healthcare wearables, and even in consumer tech like phones detecting scam calls.
Bernard reflects on the rapid acceleration of AI capabilities, noting that early applications were limited by data availability and computing power, but today just about every smartphone is capable of running complex AI models locally. He underscores how Edge AI is becoming increasingly prevalent in practical, behind-the-scenes scenarios – such as retail analytics, anti-theft systems, or infrastructure monitoring – improving efficiency, sustainability, and safety across sectors. Importantly, deploying AI on the Edge means solutions can be more cost-effective, power-efficient, and, in many cases, self- or solar-powered, enabling wider adoption and sustainability.
The discussion also addresses technical and ethical challenges. Edge AI development remains constrained by hardware resources, particularly memory and energy efficiency. Ethical concerns around privacy, transparency, and data stewardship are also highlighted. Bernard emphasises the need for regulation, though he acknowledges the inherent lag between policy and technological innovation. He calls for more transparency in AI systems – including disclosures about power usage and data provenance – to build greater public trust.
Looking to the future, Bernard predicts that within three to five years, more AI models will run on the Edge than in the Cloud, due to the compelling benefits of local computation: reduced cost, improved performance, and privacy. He advocates for Edge-first system design, stressing the importance of continued collaboration across startups, academia, and major tech firms. The Foundation remains committed to fostering an inclusive, innovative community to accelerate the responsible adoption and positive impact of Edge AI globally.
To hear more about what Bernard had to say about Edge AI, listen on Spotify, Apple Podcasts, and at the link below.
