About SEMIQA
Pioneering analog signal processing technology with revolutionary ANNET solutions for real-time AI applications
Our Mission
To pioneer a paradigm shift in AI by developing sustainable and scalable analog neural network solutions that provide real-time data processing, low energy consumption, and minimal latency. We seek to eliminate the efficiency and speed limitations of digital systems and offer cost-effective alternatives for industrial and commercial applications.
Our Vision
To leverage the advantages of analog signal processing to create products that deliver superior performance in real-time applications. We aim to close the gap between edge computing and human perception across domains—from wearable health trackers to autonomous sensor networks—by offering solutions with high sensitivity and low latency.
The Minds Behind Innovation
Our diverse team of experts brings together decades of experience in analog computing, AI research, and cutting-edge technology development.
Our Core Values
Innovation
Pushing the boundaries of analog computing to create revolutionary solutions.
Excellence
Committed to delivering the highest quality technology and services.
Collaboration
Working together with partners to drive technological advancement.
Impact
Creating technology that makes a real difference in the world.
Technology Focus
ANNET Technology (Analog Neural Networks)
Our revolutionary ANNET technology processes signals continuously in analog form, eliminating the conversion overhead of traditional digital systems. By bypassing A/D and D/A conversions, ANNET treats time as a continuous variable and enables temporal integration similar to biological neural networks, delivering unprecedented real-time performance.
50× higher speed than digital approaches
5× lower power consumption
100× lower latency for real-time applications
ONNX compatible with existing models
Core Architecture Components
ANNET technology is built on four fundamental innovations that work together to deliver unprecedented analog processing performance:
Analog Memory (aMEM)
Stores weights, bias values, inputs, and outputs as analog voltages. Each cell can hold many more bits than digital memory, offering major space and efficiency advantages.
Analog Compute Unit (ACU)
Performs summation, multiplication, and comparison operations directly on analog signals using corresponding weights and biases, all without digital logic elements.
Management Unit (MU)
RISC-V based microprocessor responsible for loading parameters into aMEM and managing analog signal propagation during operation.
Digital Memory (dMEM)
Stores configuration settings and network structure data, ensuring system programmability while maintaining analog processing advantages.