Nota, a technology company specializing in AI model lightweighting and optimization, has expanded its business into the cloud by launching a model tuning service targeting AWS's proprietary AI chip environment. Following Samsung Electronics, Arm, Furiosa AI, and Mobilint, a collaboration framework with AWS has been established, expanding the scope of AI model optimization from mobile to the cloud.

Nota announced on the 7th that it has launched a dedicated model tuning service for AWS AI chips based on its platform NetsPresso® as an AWS Validated Partner.

The service is intended for enterprise customers considering the adoption of AWS Trainium, a chip dedicated to training, and Inferentia, a chip dedicated to inference.

Netspresso is a lightweight and automated deployment platform that can reduce the size of AI models by more than 90% while maintaining accuracy.

This service consists of three stages: proof of concept (PoC) and diagnosis, model porting and lightweighting, and performance tuning.

The company explained that it supports the entire process to help companies lacking expertise in AI optimization apply the price-performance and energy efficiency of AWS AI chips to their workloads.

Nota has extensive experience in large-scale language model tuning in the Treinium and Inferentia environments.

It is reported that lightweighting technology was applied to a language model with 32 billion parameters, reducing the model size by 68% while maintaining a loss of less than 1% in accuracy.

AWS's proprietary AI chips are being utilized by global companies such as Anthropic, Apple, Databricks, and Uber.

According to AWS, the business related to Treinium grew 150% from the previous quarter as of the third quarter of 2025, expanding to a scale of billions of dollars.

Nota CEO Chae Myeong-su said, “AWS AI chip-based infrastructure"We will provide tangible cost-efficiency results to corporate customers considering the adoption of [system name]," they stated.