Summary
Highlights
The episode begins by defining edge computing as any computation performed outside a cloud region, encompassing billions of devices from laptops to smartphones. It also immediately introduces the primary concern of hardware integrity due to components coming from various low-cost sources.
The discussion traces the evolution of edge technology, starting with Edge caching and CDNs in the 1990s, followed by peer-to-peer networks in the 2000s, micro-datacenters like AWS Outposts in the 2010s, and finally, the pervasive data collection and processing on devices like security cameras and self-driving cars in the 2020s.
The speakers illustrate the rising strategic importance of edge computing by showing a graphic that reveals how mentions of 'Edge Computing' in IT earnings calls have surpassed 'Cloud Computing' in recent years, indicating significant executive and investor interest.
Four key predictions for the future of edge computing are presented. The first two are the rise of Federated Inference, where data processing occurs on decentralized devices (e.g., drones communicating as a hive mind), and Federated Learning, where AI models are personalized and trained on individual devices, enhancing privacy and creating a 'super AI' without central data access.
The third prediction emphasizes the need for purpose-built infrastructure to handle the growing data generated at the edge, citing examples like new streaming data movement technologies (Redpanda, WarpStream) and Edge Hybrid Cloud (MotherDuck). The fourth prediction highlights the critical role of 6G, expected to be 100 times faster than 5G, in providing the necessary network backbone for this data transfer.
These four forces (Federated Inference, Federated Learning, purpose-built infrastructure, and 6G) are predicted to converge and lead to the 'cognitive internet,' where devices proactively act as co-pilots, collecting and processing data at the edge to make suggestions and enhance productivity.
The discussion shifts to security, noting that significant investment (22%) is now directed towards securing Operational Technology at the edge. The unique challenges include limited compute power on edge devices, making it hard to run additional security code, and concerns about the integrity of globally sourced, low-cost hardware components.
The episode compares security solutions in mobile and OT contexts. Mobile Device Managers (MDMs) have addressed data theft concerns in mobile devices through remote wipe capabilities, and robust operating systems like Android and iOS have effectively isolated applications and hardware. However, hardened operating systems have not seen widespread adoption due to their heavyweight nature. This contrasts with the OT space, where standardization and comprehensive security controls are still evolving.
Practical advice for founders and CISOs in the edge computing space includes ensuring hardware integrity through supply chain assurance, developing a minimal monitor control plane for awareness of attacks on resource-constrained devices, maintaining security and compliance for sensitive data moving to the edge (auditing, logging), and implementing classic data protection concepts like encryption at rest and in transit.