Neuromorphic computing promises to radically improve AI performance for always-on, real-time applications—from smart sensors to edge robotics—by mimicking the brain’s parallel, event-driven architecture. These systems can dramatically reduce power consumption while improving responsiveness, adaptability, and continuous learning capabilities.
While the research landscape is rich, viable commercial solutions remain limited due to technical, ecosystem, and market readiness challenges. One promising path forward is analog in-memory computing, which reduces data movement by computing directly where data is stored. Several memory technologies have been explored for this, including flash and MRAM, but these technologies face limitations.
ReRAM stands out with its ability to scale below 28nm, low-voltage operation, high endurance, and analog behavior suitable for synaptic weighting. However, it introduces new challenges around variability, endurance tuning, and circuit integration.
This session will introduce the neuromorphic computing landscape, explain why ReRAM is emerging as the key enabler, and outline what’s needed to bring brain-inspired AI from lab to product.
Attendees will gain a clear understanding of neuromorphic computing’s promise, the current state of hardware solutions, and what research and development is still required.