AWS has unveiled critical updates to its generative AI and cloud infrastructure portfolio, including the immediate availability of NVIDIA Nemotron 3 Super on Amazon Bedrock and a new SDK for customizing Nova models.
The announcements, made during the week of March 23, 2026, signal Amazon’s aggressive push to simplify enterprise adoption of large language models (LLMs) and accelerate data analytics performance.
NVIDIA Nemotron 3 Super Now Available on Amazon Bedrock
NVIDIA Nemotron 3 Super, a high-performance language model optimized for text generation, complex reasoning, summarization, and code generation, is now accessible via the unified Bedrock API. Customers can invoke the model alongside other foundation models without managing infrastructure.
“This is a game-changer for enterprises looking to deploy generative AI at scale,” said Daniel Abib, Senior Specialist Solutions Architect at AWS. “By integrating Nemotron 3 Super into Bedrock, we’re reducing the operational burden and enabling teams to focus on innovation.”
Amazon Redshift Delivers Up to 7x Faster Performance for New Queries
Amazon Redshift now accelerates first-time queries in dashboards and ETL workloads by up to 7x. The performance gain, achieved without depending on cached results, reduces wait times for interactive dashboards and speeds up ETL pipelines, especially in high-variability query environments.
Introducing Nova Forge SDK for Custom Enterprise AI
The Nova Forge SDK provides a streamlined way to fine-tune and customize Amazon Nova models for domain-specific data. Developers can deploy the customized models directly within Amazon Bedrock, offloading the complexity of model customization to AWS.
“Nova Forge SDK handles the heavy lifting of model customization,” Abib added. “You can adapt Nova models to your specific business logic and deploy in hours rather than weeks.”
Amazon Corretto 26 Now Generally Available
Amazon Corretto 26, the latest long-term support (LTS) release of the production-ready OpenJDK distribution, is now generally available. It includes new Java language features, performance improvements, and security patches, with long-term support from AWS for use across development and production environments on Amazon Linux and other platforms.
Background
The updates come as enterprises increasingly seek to integrate generative AI into core workflows. AWS has been expanding Bedrock’s model catalog, adding third-party LLMs like NVIDIA’s Nemotron series alongside its own Nova models.
Simultaneously, demand for faster analytics and cost-efficient query processing has driven improvements in Redshift’s architecture. The new Nova Forge SDK addresses enterprise concerns about data privacy and model customization, a key barrier to AI adoption in regulated industries.
What This Means
For developers and IT leaders, these launches reduce the complexity of building AI-powered applications. Nemotron 3 Super on Bedrock allows teams to experiment with state-of-the-art models without provisioning GPU clusters, while the Nova Forge SDK enables domain-specific tuning without deep ML expertise.
The Redshift performance boost directly impacts data engineering teams, cutting query times for new analytics workloads and reducing total cost of ownership. Corretto 26 ensures Java developers have access to the latest performance and security enhancements, backed by AWS’s long-term support commitment.
Industry analysts see Amazon consolidating its lead in cloud-based AI infrastructure. “AWS is making it easier than ever to move from experimentation to production,” said a cloud computing expert. “The combination of broad model access, customization tools, and high-performance data services will accelerate enterprise generative AI adoption.”