Probabilistic AI

Probabilistic AI

Today, a third wave of AI is needed to reconcile learning and expertise while tackling data explosion. Probabilistic programming has an important role to play to allow the development of models able to learn from few examples, to integrate expert knowledge, to generalize from what was learned, to reason, and to explain the decisions that are made. The dedicated hardware accelerators developed at will be a key enabler of this third wave of AI.

Trustworthy AI: analyze, understand, and control how the model reaches conclusions.

Sustainable AI: work with limited amount of data and reduced training time and energy.


The lack of efficient computing hardware is the main obstacle to pervasive use of probabilistic AI.

Our architecture optimizes key operations of probabilistic models such the sampling of random variables.

This allows running more complex models faster and with less energy than existing hardware.


Our accelerators will provide leading-class energy efficiency for probabilistic workloads.

Edge inference FPGA

Edge inference FPGA

Sense, Analyze & Act

x40 samples/watt @ 10W

Open-access in 2023

Contact us for closed-beta access

Edge inference & learning SoC

Edge inference & learning SoC

Sense, Analyze, Act & Learn

x500 samples/watt

Mass-production in 2026


We are a team of committed, passionate and expert people.
Entrepreneurs, data scientists, microelectronics engineers,
we strive to enable a more ethical and more sustainable AI.