NeuroScale raises $85M Series B to scale neuromorphic chip for low-power inference

Tech · 5 min read

NeuroScale, a deep tech startup building neuromorphic processors for low-power AI inference, announced an $85 million Series B led by a consortium including Sequoia and a major telecom operator. The company plans to use the funds to finalize tape-out for its second-generation chip and build manufacturing partnerships for volume shipments in 2027.

Concurrently, NeuroScale released a developer SDK that abstracts spiking-neural-network topologies into familiar ML frameworks and supports model quantization and conversion pipelines. The SDK includes edge-first runtimes designed for battery-constrained IoT devices, drones, and AR wearables.

NeuroScale says its chips deliver 5–10x improved energy efficiency for specific sequential and temporal workloads compared with conventional accelerators. The company is partnering with several robotics and wearable manufacturers to run pilots later this year, positioning itself for a market where battery life and continuous sensing are strategic differentiators.