We are building the most brain-like hardware ever imagined
We are inspired by the elegance, scale, and efficiency of the human brain. Our technologies focus on in-memory processing, analog computation, and scalable systems.
The world's first algorithm for end-to-end analog training and inference
We are developing new algorithms for fully analog neural networks, utilizing fundamental physics to further scale artificial intelligence.
We look forward to sharing more about this technology soon.
The world's first scalable, analog processor for artificial intelligence
Analog neurons
We tile artificial neurons across the entire chip surface, enabling higher neuron density in CMOS hardware than ever before. We can fit millions of artificial neurons in a square centimeter, enabling massive neural networks to fit onto a single chip.
Memristive synpases
The memristor is widely recognized as the ideal artificial synapse. Memristor is a combinaton of "memory" and "resistor"; it is a resistor capable of changing its resistance. Our memristive synapses exist in the shells of our wires, stacked vertically on top of our neurons.
Nanowire axons
Our nanowires are randomly arranged on the surface of the chip. Emerging from this random placement is a powerful, sparse, brain-like topology: a small-world network. This connectivity allows our network to remain well connected even at massive scale.