Tenwatts builds fully integrated AI stacks—model, compiler, and runtime—optimized for on‑device execution to achieve an order‑of‑magnitude reduction in power consumption. Their hardware‑agnostic solutions use advanced quantization and pruning to deliver low‑latency, accurate inference on battery‑powered or low‑power devices, and they offer engineering support to help manufacturers integrate these efficient AI pipelines.
Funding
Funding not disclosed
Founders
Product
Problem
Deploying advanced AI models on edge devices is limited by high power consumption and hardware constraints, preventing real-time inference and widespread adoption in power‑sensitive applications.
Solution
Tenwatts develops fully integrated AI systems that are optimized for on‑device execution with a focus on dramatically reducing energy use. By applying custom hardware‑aware model design, quantization, and compiler techniques, the company targets an order‑of‑magnitude improvement in power efficiency compared to conventional AI stacks. The resulting solutions enable continuous, low‑latency inference on battery‑operated or energy‑restricted platforms while maintaining model accuracy. Tenwatts also provides engineering support to help partners integrate these efficient AI pipelines into their products, accelerating time‑to‑market for edge AI deployments.
Target Audience
Primary customers are device manufacturers, IoT product developers, and OEMs seeking to embed AI functionality into battery‑powered or low‑power hardware.
Features
- End‑to‑end AI stack (model, compiler, runtime) co‑optimized for on‑device power efficiency
- Advanced quantization and pruning methods that preserve accuracy while cutting energy use
- Hardware‑agnostic deployment framework supporting CPUs, microcontrollers, and specialized accelerators
- Real‑time inference capabilities with latency suitable for interactive applications
- Engineering consultancy for custom integration and performance tuning on client hardware