Modern computer vision requires the complex orchestration of storage, compute, processing, and sensors, inherently limiting the flexibility of its applications. Synapse Semiconductor is the first company to integrate all components of the computer vision stack into one substrate.
We push machines to explore where humans can’t go.
We solve how machines
see the world.
We replace the disjoint architecture of a camera and a GPU with one chip: a compute substrate that acts as a sensor.
The whole stack. One wafer.
Scroll
Synapse Semiconductor is a frontier semiconductor research lab building senses for physical AI. Synapse Semiconductor builds image sensors that process AI on the sensor itself. Each pixel captures light and runs neural-network computation in the same device, replacing the need for a GPU while cutting the power and latency of edge vision.
Vision systems are already the core technology for self-driving cars, drones and humanoid robotics.
Like the New World once was, space is the next great frontier to explore. In the future, drones will navigate using vision alone, robots will scout the Moon for the next colony, and satellites will perceive their environment and react in real time. At the foundation of it all will be Synapse Semiconductor.
Manifestos
Sanjeev Chauhan
The innovation at the smallest scale yields the largest impact.
Solving problems in the hardest domains creates results no nation can replicate.
What the US needs is a semiconductor lab that imagines novel ideas for devices, transistors, compute, sensors, memory systems — the foundations of our society — and scales them all the way to a product. In house. That’s what Synapse Semiconductor is.
The globalization of semiconductors was great. It lowered prices and increased the ubiquity of technology. However, it slowed the cycle of innovation. Fabless companies can innovate at the circuit design level with little friction, but what happens when that is not enough? New ideas are limited by the selection of PDKs a foundry has.
So few can make new products where innovation is rooted in new transistor-like devices. So few can imagine where ideas are not constrained by the current market. Synapse Semiconductor is the solution — where ideas are constrained only by the imagination of the future.
Sanjeev Chauhan
Tania Roy
Our semiconductor research lab at Duke has built something strange and useful: a layer of compute that can also see the world around it, and remember what it saw. Each transistor is co-located with a high-bandwidth memory that is also sensitive to light!
The separation of the camera and the GPU is gone. Data does not have to travel from the sensor to the GPU to the memory and wait in traffic any more. Intelligence grows where the light hits.
America has to lead the AI race. It has to control semiconductor manufacturing.
Imagine if you could own the semiconductor manufacturing tools as easily as you can own a mechanical tool set. You could assemble different electronic materials and build your AI chip wherever you are! Imagine what American innovators could do with that. That will be the day when we will have the right amount of progress to build civilization in space.
Silicon is unbeatable. It is also a century-old tree — deep roots, perfect canopy, decades of tending. It demands high investments and constant care to deliver.
This industry is crying out for a disruptor.
Many challengers to silicon have risen, only to die because they did not receive the time and investment that was needed to nurture them.
2D materials have been in the sun for more than a decade. Slam two slabs of silicon together and hope for a diode? Never. Exfoliate a flake with Scotch tape, stack it on another — a diode, on a tabletop, without a billion-dollar fab! These materials deserve to be scaled. Matching silicon on day one was never the prudent path. What de-risks their growth better than vision chips the market actually wants?
Paradigms are already shifting.
I was taught that high-performance computing demands a single crystal — silicon grown with surgical precision.
In the last five years, amorphous oxide semiconductors have competed anyway. They break the rule we were handed: that crystallinity is everything. The garden keeps surprising you, if you stop grading every sprout against the oldest tree. Only when chip-making becomes accessible will we keep finding facts like that.
Our love for semiconductors, devices, and the desire to make life multiplanetary is the unique combination that will revolutionize physical AI. We are Synapse Semiconductor.
Tania Roy