Sagence Learning

About Sagence Learning

Sagence AI, a company operating from Santa Clara, California, is focused on developing specialized semiconductors for artificial intelligence workloads. Formerly known as Analog Inference, the company emerged from stealth in November 2024 to introduce its advanced analog in-memory compute architecture. The firm was established by Vishal Sarin, a semiconductor industry veteran with over 25 years of experience and more than 100 patents, who now serves as President and CEO. Sarin's background includes extensive work in analog in-memory computing, which forms the technical foundation of the company.

The company's core business revolves around addressing the significant power consumption and cost challenges associated with AI inference, which is the process of using a trained AI model to make predictions. Sagence AI is engineering analog chips that it claims can offer substantial improvements in energy efficiency and cost-effectiveness compared to the high-performance GPUs that currently dominate the market. Its technology executes AI operations directly within multi-level memory cells, a true in-memory compute approach that eliminates the data movement bottleneck between separate processing and memory units. This method is designed to drastically reduce power consumption, with the company stating its technology can perform with 10 times lower power and at a 20 times lower price point than leading GPU-based systems for specific large language model tasks.

Sagence AI's product is an inference machine built on a modular chiplet architecture, designed to scale from large data center applications for generative AI to smaller edge computing use cases. The system is built to work with standard AI frameworks like PyTorch, ONNX, and TensorFlow, importing already-trained models and converting them to its native format. Having secured US $58 million in funding from notable venture and strategic investors, including Khosla Ventures, TDK Ventures, and Alumni Ventures, Sagence AI is targeting a 2025 market launch for its initial chips focused on vision systems, with plans to address larger generative AI applications thereafter.