The global volume of data and the rapid growth of AI are driving unprecedented demand for compute power. In today's digital systems, performance and efficiency depend on continual hardware and software gains. Yet conventional digital computing is approaching practical limits in throughput, latency, and power consumption. Modern AI models increasingly face scalability constraints because processor performance is bottlenecked by data movement and memory bandwidth. In many workloads, more energy is spent transferring data than performing computation. The result is limited model efficiency, escalating power requirements, and growing environmental impact. To address this challenge, SEMIQA has developed an Analog Neural Network (ANN) technology inspired by the human brain. This neuromorphic approach significantly reduces energy consumption while improving performance.
Low Latency
Event-driven analog computation cuts response times for time-critical edge inference workloads.
Energy Efficiency
Minimized data movement and native analog execution dramatically lower total power consumption.
High Performance
Dense parallel analog operations deliver high computational throughput at milliwatt-level power.
SEMIQA develops two licensable ANN IP cores for external vendors. Both are built on CMOS-compatible neuromorphic technology and optimized for production deployment with strict power and efficiency targets.
ANN1000 focuses on edge intelligence where ultra-low-power and low-latency operation are critical.
ANN2000 is tailored for cloud-scale workloads, delivering high throughput for data center environments with demanding bandwidth requirements.
ANN1000 targets the growing demand for edge computation and is based on event-driven signal propagation. It is an ideal solution for applications requiring ultra-low-power, low-latency inference. The design is dedicated to direct communication with local data collectors, such as cameras.
ANN2000 is designed for integration with data centers. Its neuromorphic approach significantly reduces power consumption while delivering higher computational efficiency than conventional architectures.
Our Process Design Kit enables foundry partners and design teams to integrate SEMIQA's analog neural network into their own chip designs.
The PDK includes verified cell libraries, simulation models, and layout rules optimized for our novel analog materials, enabling tape-out with confidence.
Our Software Development Kit provides a straightforward workflow for deploying AI models on our hardware. Load standard models, quantize them for analog execution, and deploy in minutes.
Our compiler handles the conversion from floating-point to optimized integer representations automatically.