Sensing.
Edge AI.
Silicon designed in Riyadh. Sensor and mixed-signal SoCs in tapeout — the first completing in 2026 — and edge-AI processors in design.
A RISC-V sensor & mixed-signal SoC
EQ Semi’s first silicon. DUNE S1 is an analog sensor SoC with an embedded RISC-V processor, built for smart sensing of temperature, humidity and voltage. Analog inputs go in; calibrated digital readings and threshold alerts come out — so a battery, a cabinet or a power rail is monitored by one chip rather than a board of discrete parts. It is completing its first tapeout in 2026, the milestone that takes a design from verified RTL to a physical chip.
- Sensing
- Temperature, humidity, voltage and current on one mixed-signal front end.
- On-chip logic
- An embedded RISC-V core runs the sensing and control logic on the device.
- Thresholds
- Programmable alert levels per channel, evaluated locally and reported as events.
- Output
- A digital command and data interface, for integration into an existing system.
- Efficiency
- High-efficiency operation for battery-powered and always-on monitoring.
Inside the chip
- RISC-V processor core with instruction fetch, decode and execution blocks
- Analog data-acquisition unit with a multiplexed converter across four channels
- Configuration and threshold registers for temperature, humidity, voltage and current
- I²C command decoder for configuration and readout
Application areasClean energy · Consumer electronics · Electric vehicles · Smart homes & IoT · Industrial automation
Enquire about DUNE S1
Edge-AI processor with an on-chip CNN accelerator
EQ Semi’s second chip, now in design. DUNE E1 is an AI edge processor that runs lightweight vision networks directly on a camera feed, enabling real-time detection for traffic monitoring, drones and industrial inspection. Inference happens on the device, so only the result leaves it — not the video. Centralized GPUs train the models; DUNE E1 runs them where latency, privacy and cost all point to the edge.
- Edge intelligence
- Vision networks run on the camera feed itself, in real time.
- On-chip acceleration
- A CNN accelerator handles inference alongside the RISC-V host.
- Local processing
- Detections leave the device; raw video does not — lower bandwidth, better privacy.
- Connectivity
- Standard peripherals and a high-speed path for bulk data off-chip.
- Programmable
- Model parameters are loaded from external memory, so the network can change.
Inside the chip
- RISC-V host subsystem for control, boot and program execution
- CNN accelerator (NPU) built on a systolic array, with weight, feature and control datapaths
- AXI4 interconnect linking host, accelerator and peripherals
- Camera pipeline front end for capture and resize, plus SPI, UART and I²C peripherals
Application areasTraffic monitoring · Security and inspection drones · Industrial inspection · Smart-city cameras · Robotics and automation
Enquire about DUNE E1
Have an application in mind?
Include the product name, operating conditions and integration requirements in your enquiry.
Contact EQ Semi