Led by Swami Sivasubramanian, AWS Vice President of Agentic AI, the keynote address highlighted AWS's commitment to enabling customers to deploy production-ready AI agents at scale. The summit showcased a series of groundbreaking announcements that underscored AWS's leadership in cloud innovation, generative AI, and scalable infrastructure. Below is a detailed overview of the top announcements from the event.

1. Amazon Bedrock AgentCore: Securely Deploy and Operate AI Agents at Scale (Preview)

Amazon Bedrock AgentCore emerged as a flagship announcement, designed to empower businesses to deploy and operate AI agents with enterprise-grade security. This new service facilitates rapid scaling of AI agents, offering features such as memory management, identity controls, and seamless tool integration. AgentCore is compatible with any open-source framework and foundation model, streamlining development processes and reducing complexity. It also includes AgentCore Observability, powered by Amazon CloudWatch, which provides real-time visibility into agent performance through built-in dashboards and telemetry. This feature enables developers to track and optimize AI agent actions in production environments, ensuring reliability and efficiency.

This innovation is poised to accelerate AI adoption across industries by providing a secure, scalable platform for building autonomous agents capable of reasoning, planning, and adapting to complex tasks. Companies like Warner Bros. Discovery Sports Europe and BMW have already leveraged Bedrock-based solutions for real-world applications, such as rapid fact-checking for sports commentary and diagnosing network issues for connected vehicles.

2. Extensive Customization of Amazon Nova Models in Amazon SageMaker AI

AWS introduced powerful customization capabilities for Amazon Nova foundation models through Amazon SageMaker AI. These capabilities allow customers to tailor Nova models across all stages of model training, including pre-training, supervised fine-tuning, direct preference optimization, reinforcement learning from human feedback, and model distillation. Available as ready-to-use SageMaker recipes, these tools enable businesses to adapt Nova models to meet specific industry requirements, enhancing their applicability in sectors like healthcare, finance, and manufacturing. SageMaker AI also introduced three new features: integration with Amazon QuickSight for dashboard creation, Amazon S3 Unstructured Data Integration for cataloging documents and media, and automatic data onboarding from Lakehouse, unifying structured and unstructured data management.

This announcement reflects AWS’s focus on providing flexible, industry-specific AI solutions, empowering organizations to build models that align with their unique business needs.

3. AWS Free Tier Update: Up to $200 in Credits for New Customers

AWS enhanced its Free Tier program, offering new customers up to $200 in credits to explore its services. This includes $100 upon sign-up and an additional $100 earned by completing activities with services like Amazon EC2, Amazon Bedrock, and AWS Budgets. This initiative lowers the barrier to entry for organizations and developers looking to experiment with AWS’s cloud and AI offerings, fostering broader adoption and innovation.

4. TwelveLabs Video Understanding Models in Amazon Bedrock

AWS announced the integration of TwelveLabs video understanding models into Amazon Bedrock. These models enable customers to search, classify, summarize, and extract insights from video content with high precision and reliability. This capability is particularly valuable for industries like media, entertainment, and security, where video data analysis is critical. By incorporating TwelveLabs’ technology, AWS strengthens its portfolio of AI-driven tools for multimedia processing.

5. Amazon S3 Metadata and S3 Vectors: Enhanced Data Management

AWS introduced Amazon S3 Metadata, which provides comprehensive visibility into all objects stored in S3 buckets through live inventory and journal tables. This feature supports SQL-based analysis with automatic updates within an hour of changes, enhancing data governance and management. Additionally, Amazon S3 Vectors was unveiled in preview, offering a cloud object store with native support for storing and querying vectors at scale. S3 Vectors delivers up to 90% cost savings compared to traditional approaches and integrates seamlessly with Amazon Bedrock Knowledge Bases, SageMaker, and OpenSearch for AI applications. These advancements simplify large-scale data management for AI and analytics workloads.

6. Enhanced Amazon EventBridge Logging for Event-Driven Applications

Amazon EventBridge now offers enhanced logging capabilities, providing comprehensive event lifecycle tracking for event-driven applications. These logs detail when events are published, matched against rules, delivered to subscribers, or encounter failures, enabling developers to monitor and troubleshoot applications effectively. This update is crucial for businesses relying on event-driven architectures, ensuring greater reliability and transparency.

7. Amazon EKS Scales to 100,000 Nodes per Cluster

Amazon Elastic Kubernetes Service (EKS) now supports up to 100,000 nodes per cluster, enabling massive AI and machine learning (ML) workloads. This scalability accommodates up to 1.6 million AWS Trainium accelerators or 800,000 NVIDIA GPUs, making EKS a powerhouse for large-scale AI/ML applications. This enhancement positions AWS as a leader in supporting ultra-scale computing needs for industries like autonomous vehicles, scientific research, and financial modeling.

8. AWS AI League: Upskilling Through Competition

The AWS AI League was introduced as a program to help organizations upskill their workforce through hands-on learning and competition. Focused on generative AI skills like fine-tuning, model customization, and prompt engineering, the program combines fun and practical training to build expertise in AI development. This initiative supports enterprises and developers in mastering the tools needed to create cutting-edge AI solutions.

9. AI Agents and Tools in AWS Marketplace

AWS unveiled AI Agents and Tools in AWS Marketplace, a one-stop shop for discovering, purchasing, deploying, and managing AI agents and tools from leading providers. This marketplace streamlines access to ready-to-integrate solutions, enabling businesses to accelerate their AI strategies. The announcement builds on AWS’s partnership with providers like Anthropic and Meta, offering tools like the open-source SDK Strands Agents, which simplifies the creation of coordinated AI agent teams for complex tasks.

10. $100 Million Investment in Generative AI Innovation Center

AWS announced an additional $100 million investment in its Generative AI Innovation Center, building on two years of success in helping customers like AstraZeneca, BMW, and Yahoo Finance achieve significant productivity gains through AI. The investment will focus on advancing autonomous agentic systems, supporting the next wave of AI innovation. Additionally, AWS partnered with Meta to offer 30 U.S. startups up to $200,000 in credits and technical support to build AI applications using Llama models, with applications open until August 8, 2025.

Conclusion

The AWS Summit in New York 2025 highlighted AWS’s strategic vision for agentic AI, scalable infrastructure, and accessible cloud services. From Amazon Bedrock AgentCore’s secure AI agent deployment to the enhanced scalability of Amazon EKS, these announcements demonstrate AWS’s commitment to empowering businesses with cutting-edge tools. The introduction of the AWS AI League and the expanded Free Tier program further democratizes access to AI and cloud expertise, while innovations like S3 Vectors and TwelveLabs integration push the boundaries of data and multimedia processing.