Press Release
July 7, 2026
Nota AI, a pioneering leader in AI model hardware-aware optimization and compression technology, announced today that it is enabling enterprise customers to maximize the performance of their AI models on Amazon Web Services (AWS) dedicated AI chips. Holding status as a Validated AWS Partner, Nota AI offers specialized compression and tuning services designed to unlock the peak performance of customer workloads on AWS custom silicon environments, including AWS Trainium.
The newly launched service targets enterprise clients facing a shortage of specialized AI optimization talent. By tuning models precisely to AWS’s high-performance AI chips, Nota AI helps businesses achieve significant cost efficiencies and boosted inference speeds. Nota AI conducts rigorous technical diagnostics and tuning tailored to each client's specific model architecture. This allows enterprises to fully capitalize on the superior price-performance and energy efficiency that AWS custom silicon brings to production-grade workloads.
Specifically, Nota AI’s new service accommodates enterprises evaluating or adopting AWS Trainium—purpose-built by AWS for high-performance deep learning training with optimal token economics—and AWS Inferentia, engineered for low-cost, high-throughput inference.
The service is powered by NetsPresso®, Nota AI’s proprietary hardware-aware AI optimization platform that automates the entire life cycle from model compression to target hardware deployment. NetsPresso® can shrink AI model sizes by up to 90% or more while maintaining strict baseline accuracy, delivering multi-hardware optimization within a single, streamlined workflow. The newly rolled-out service follows a structured three-stage roadmap: Proof of Concept (PoC) & Diagnostics, Model Porting& Compression, and Target Performance Tuning.
AWS Trainium and Inferentia have already established a proven track record among global industry leaders. Leading companies such as Anthropic, Apple, Databricks, Uber, Ricoh, and Decart leverage AWS custom silicon. Notably, as of Q3 2025, AWS's Trainium-related business experienced explosive growth, surging 150% quarter-over-quarter into a multi-billion dollar segment. Nota AI’s service focuses on eliminating technical friction, allowing businesses to integrate and deploy their custom workloads onto these powerful, market-proven AWS AI chips rapidly.
This service launch stands on the foundation of Nota AI’s deep, proven engineering expertise. The company has accumulated extensive experience tuning Large Language Models (LLMs) natively within AWS Trainium and Inferentia environments. In a recent notable benchmark,Nota AI applied its compression technology to a high-performance 32-billion-parameter language model, successfully reducing the model size by 68% while keeping the accuracy loss below 1%. The company also holds a vast portfolio of porting and tuning diverse model architectures, allowing clients to tap into the capabilities of AWS custom silicon instantly.
"AWS AI chips offer unparalleled price-performance, and Nota AI is uniquely positioned to help enterprises fully unlock that performance potential for their custom models through expert tuning," said Myungsu Chae, CEO of Nota AI. "We are committed to delivering tangible cost-reduction outcomes and performance breakthroughs to corporate clients seeking to transition their infrastructure to AWS AI silicon."
Nota AI has consistently validated its market-leading technology through strategic partnerships across a diverse hardware ecosystem, collaborating with Samsung Electronics in mobile, Arm in semiconductor IP design, Furiosa AI in data center accelerators, and Mobilint in edge devices. The roll out of this AWS-centric optimization service marks a significant expansion of Nota AI’s technology footprint into the cloud infrastructure domain. Moving forward, Nota AI intends to solidify its position as the ultimate model compression and tuning partner across every environment where AI workloads run.
.png)
