Companies

Synapse Semiconductor

synapsesemi.org

Synapse Semiconductor builds retina-inspired AI image sensors combining light capture and neural computation on-chip.

HQDurham, North Carolina, United States
Employees1-50
1 active role
Jobs checked 4h ago
AI / MLAI InfrastructureInfrastructure

About

Synapse Semiconductor is developing retina-inspired image sensors that capture light and run neural-network computation directly within each pixel. It targets robotic-perception, drone, satellite, and other physical-AI applications, differentiating itself by integrating sensing, compute, memory, and processing into a single wafer-scale substrate to reduce GPU dependence, power consumption, and latency.

Market

Synapse Semiconductor competes in edge AI vision hardware, intelligent image sensors, and neuromorphic or compute-in-sensor systems for physical AI. Its positioning is unusually vertically integrated: rather than pairing a conventional camera with a separate processor, it places neural computation at the pixel and aims to consolidate the broader vision stack into one substrate or wafer. This can differentiate it through lower power consumption, lower latency, and reduced sensor-to-processor data movement, although several competitors address adjacent parts of the same market through intelligent sensors, event-based vision, neuromorphic chips, or edge AI processors.

Target Customers

Robotics, drone, satellite, autonomous-vehicle, and other physical-AI companies that need low-power, low-latency visual perception, especially teams operating in energy-constrained environments or without reliable cloud access. Likely buyers are engineering, robotics-perception, and embedded-hardware teams at technology startups, OEMs, and advanced research organizations.

At a Glance

Problem

Synapse Semiconductor is addressing a central bottleneck in physical AI: today’s computer-vision systems separate the camera from storage, processing, memory, and GPU compute. That separation forces data to move between components and creates system complexity, power consumption, latency, and dependence on external processors. The economic pain is most acute where energy, weight, response time, and onboard computing capacity are constrained, such as autonomous vehicles, drones, humanoid robots, and machines designed to operate in environments humans cannot reach.

The company’s core use case is robotic and autonomous perception: enabling a machine to interpret its surroundings locally and in real time rather than sending camera data through a conventional camera-to-GPU pipeline. Synapse frames this as eliminating the traffic between the sensor, GPU, and memory so that intelligence can be generated where light hits the device.

Product / Service

Synapse is developing RETINA, an integrated image-sensor and computing architecture in which each pixel captures light and performs neural-network computation in the same device. The company describes this as collapsing the edge-AI vision stack—sensing, compute, processing, storage, and interconnect—onto one substrate or wafer. Its compute transistors are also the photosensors, removing the conventional separation between a camera and a Jetson-class processor.

The public materials present Synapse as a frontier semiconductor research lab developing a chip technology rather than as a software or services provider. The intended benefit is lower-power, lower-latency edge vision without a separate GPU or external processor, allowing neural-network operations, including computer-vision models, to run directly in the sensing hardware.

Market

Synapse competes at the intersection of application-specific semiconductors, edge-AI vision, computer vision, and physical-AI hardware for robotics, drones, autonomous vehicles, and related systems. The clearest incumbent alternative is the conventional stack of a camera such as Intel RealSense paired with an NVIDIA Jetson-class computer. Adjacent specialists include neuromorphic-vision companies such as iniVation and Prophesee, although the available evidence does not establish that they offer the same single-substrate architecture.

As of August 2026, Synapse appears to be an early-stage, likely pre-revenue company rather than a scaled commercial supplier. Y Combinator lists it as an active Summer 2026 company with a two-person team and no listed jobs; PitchBook classifies it as private and accelerator-backed, reports a $500,000 January 2026 accelerator deal and Y Combinator as its investor, and provides no current revenue figure. The company publicly invites builders of robotic perception, drones, and physical-AI vision systems to make contact, but the reviewed materials do not disclose customers, production deployments, or revenue.

Founders & Leadership

Sanjeev ChauhanFounder
CEO & Co-founder
Tania RoyFounder
Co-founder

Funding History

2025-02
Seed$500K

Y Combinator, Entrepreneur First

Recent News

2026-07-26partnership
Synapse Semiconductor: Collapsing the edge AI vision hardware stack into one wafer

Y Combinator lists Synapse Semiconductor as an active company in its Summer 2026 batch. The company is developing a single-substrate vision system that integrates sensing, storage, compute, and processing for edge AI.

2026-06-08
Duke ECE Hardware Startup Selected for Y Combinator

Duke reported that founders Tania Roy and Sanjeev Chauhan had been selected for Y Combinator’s summer batch. The article describes Synapse’s retina-inspired “retina on a chip,” which processes visual information directly at the sensor and has attracted interest from potential robotics, drone, and satellite partners.

2025-11-26product
Synapse Semiconductor: We put a neural network inside the camera pixel

Synapse Semiconductor presented its AI-on-the-sensor product concept, in which each pixel captures light and performs neural-network computation in the same device. The approach is intended to replace the separate camera-and-GPU architecture and reduce power and latency for edge vision.

Active Roles

1
San Francisco, CA, US / Durham, NC, US/Engineering/9d ago

Business Model

Synapse Semiconductor appears to be pursuing a business-to-business semiconductor model, commercializing integrated AI image-sensor chips for robotics, drones, satellites, and physical-AI companies through hardware sales and potentially technology licensing. Public materials do not disclose pricing, contracts, or current revenue.

Products

RETINA, a retina-inspired chip that runs neural-network operations directly in camera pixelsAI image sensors that capture light and process neural-network computations in the same deviceAn integrated one-wafer computer-vision platform combining sensing, compute, memory, and processing

Tech Stack

Compute-in-sensor and in-pixel neural-network processingRetina-inspired neuromorphic vision architectureIntegrated image sensors, compute, memory, and sensing substrateUltrathin molybdenum sulfide semiconductor materialsEdge AI hardware optimized for reduced power, latency, and memory movement

Competitors

Sony Intelligent Vision Sensors
Prophesee Metavision
SynSense
iniVation
Ambarella
Hailo