AnalogAI Selects Silicon Storage Technology's memBrain SAGE IP for Real-World Edge AI Processors

AnalogAI Partners with Silicon Storage Technology for Ultra-Low-Power Edge Computing

AnalogAI has selected memBrain SAGE intellectual property from Silicon Storage Technology, a Microchip Technology subsidiary, to power its first real-world edge AI processors1

1

2

2

. The memBrain Synaptic Analog Generative Engine (SAGE) neuromorphic hardware IP will serve as the core inference engine, enabling AnalogAI to deliver analog compute-in-memory performance at or below one watt for ultra-low-power edge AI applications1

1

. This partnership marks a significant step in bringing adaptive AI capabilities to resource-constrained edge devices.

Enabling Simultaneous Training and Inference for Dynamic Environments

AnalogAI edge AI processors feature a hardware-aware proprietary algorithm designed to simultaneously train and run inference on AI models in real-world environments1

1

. This capability addresses a critical gap in current edge computing solutions, where devices typically rely on pre-trained models that cannot adapt to sudden environmental changes. By leveraging the silicon-proven memBrain SAGE IP based on SuperFlash technology, AnalogAI is developing new capabilities for real-world inference applications such as environment-adapting humanoid robots, drones and vehicles1

1

2

2

. The on-chip AI hardware approach eliminates the need for constant cloud connectivity, enabling truly autonomous operation in dynamic scenarios.

Technical Architecture Behind the memBrain SAGE Solution

The memBrain SAGE intellectual property includes a comprehensive suite of components optimized for analog compute-in-memory operations. At its core, the Tensor In-Memory Logic Element (TILE) provides the computational foundation1

1

2

2

. The architecture features optimized digital-to-analog converters (DACs) and analog-to-digital converters (ADCs), summator and high voltage bias circuitry, and nanoamp level bitcell control logic1

1

. The solution includes memBrain proprietary test circuitry, comprehensive technology documentation and simulation models, plus full IP integration service and support1

1

. Silicon Storage Technology has developed and deployed the memBrain SAGE IP in 40 nm and 28 nm foundry processes using production-ready SuperFlash memory, with a technology roadmap that includes 22 nm development1

1

2

2

.

Why This Partnership Matters for Edge AI Development

Jaejun Lee, AnalogAI's chief executive officer, explained that the company selected Silicon Storage Technology's memBrain SAGE IP after conducting an industry-wide search of available offerings1

1

. The silicon-proven nature of the IP enables AnalogAI to accelerate development time while achieving the ultra-low-power and high performance required in their market segment1

1

. Mark Reiten, senior vice president of Microchip Technology's Intelligent Compute business unit, noted that AnalogAI joins a rapidly expanding ecosystem of memBrain IP licensees1

1

. This growing adoption signals increasing industry recognition of analog compute-in-memory as a viable path forward for edge AI processors that must balance performance with severe power constraints. Watch for AnalogAI's product announcements targeting robotics and autonomous systems, where real-time adaptation could differentiate their solutions from competitors relying on static inference models.