Amazon Web Services re:Invent 2023 | Key Announcements and Insights

Home / Blog / Amazon Web Services re:Invent 2023 | Key Announcements and Insights
Sourced blog AWS re:Invent 2023 - key announcements and insights

This article provides a comprehensive overview of the key announcements from Amazon Web Services’ (AWS) re:Invent 2023, highlighting the latest advancements in cloud technology, Artificial Intelligence (AI), and Machine Learning (ML). It offers insights into AWS’s strategic direction and the implications for cloud computing in the coming year.

AWS re:Invent 2023, the much-anticipated annual event, has once again set the stage for a series of groundbreaking announcements and updates. This year’s event, held in Las Vegas, was particularly notable for its focus on generative AI, maching learning, and the continuous evolution of cloud technologies. As in previous years, re:Invent 2023 was a showcase of new products and services and an indicator of the trends and directions AWS is taking in the cloud computing space.

The excitement surrounding re:Invent is palpable each year as it brings together a global community of AWS users, partners, and enthusiasts. The event has consistently been a platform for AWS to unveil its latest innovations, and this year was no exception. Attendees and observers eagerly anticipated the announcements, knowing they often set the tone for cloud computing trends in the following year.

Key Announcements

AI and Machine Learning

  • Amazon Bedrock
    GA This machine learning service geared towards Large Language Models (LLMs), and by extension Generative AI (GenAI), is now generally available.
  • Model Evaluation on Amazon Bedrock
    A new feature to evaluate, compare, and select the best foundation models for specific use cases offers automatic and human evaluation options.
  • Guardrails for Amazon Bedrock (Preview)
    A feature enabling customers to implement safeguards across foundation models based on use cases and responsible AI policies.
  • Bedrock Foundation Model Fine-Tuner
    Amazon Bedrock now supports fine-tuning for Meta Llama 2, Cohere Command Light, and Amazon Titan Text Lite and Express foundation models, allowing organisations to use labelled datasets to increase model accuracy for specific tasks.
  • Agents for Amazon Bedrock
    Fully managed Agents for Amazon Bedrock are now generally available, enabling generative AI applications to execute multi-step tasks across company systems and data sources.
  • Amazon Transcribe with Generative AI
    Enhancements to the transcription service.
  • Titan Multimodal Foundation Models
    Announcement of new Titan Image Generator and Titan Multimodal Embeddings.
  • Amazon Q
    A new generative AI-powered assistant for various AWS services, like a ChatGPT integrated into the AWS console.
  • Amazon Q for Your Business (Preview)
    Amazon Q can be tailored to business needs for conversations, problem-solving, content generation, and actions using company data and enterprise systems.
  • SageMaker Large Model Inference DLC
    New version supporting TensorRT-LLM.
  • SageMaker HyperPod
    General availability of Amazon SageMaker HyperPod, offering purpose-built infrastructure for distributed training of foundation models (FMs) at scale, reducing training time by up to 40%
  • Nvidia GH200 AI Chip Deployment
    AWS to deploy powerful Nvidia AI chips.
  • SageMaker Automated Notebooks
    Tools to create and manage noninteractive notebook jobs
  • AWS Clean Room ML
    A new feature to apply privacy-enhancing machine learning for generating predictive insights without sharing raw data.
  • Reduced Inference Costs and Latency for Amazon SageMaker
    New capabilities to reduce model deployment costs by 50% on average and achieve 20% lower inference latency.


  • Lambda Improvements
    Enhanced scaling for handling high-volume requests.
  • Graviton4 for R8g EC2s
    New memory-optimised Amazon EC2 R8g instances powered by AWS Graviton4 processors are available in preview, offering enhanced performance for memory-intensive workloads.
  • EKS Pod Identity
    Simplified IAM permissions for applications on EKS clusters.


  • S3 Express One Zone
    A new high-performance storage class at lower costs designed for AI/ML training and financial modelling.
  • EFS Archive
    A new storage class for Elastic File System optimised for long-lived data.
  • Amazon FSx updates
    New features for FSx for NetApp ONTAP and OpenZFS.
  • Amazon S3 Connector for PyTorch
    Optimised connector for reading/writing to S3 buckets.


  • Aurora Limitless Database
    A new capability enables scaling Amazon Aurora clusters to millions of write transactions per second and managing petabytes of data. It uses Aurora Serverless underneath to distribute data and queries horizontally.
  • ElastiCache Serverless
    General availability of a serverless option for Amazon ElastiCache, simplifying cache management and instantly scaling to support demanding applications. Supports Redis version 7.1 and Memcached version 1.6.22 and above.
  • Amazon RDS for IBM Db2
    General availability of Amazon RDS for Db2, simplifying the setup, operation, and scaling of Db2 databases in the cloud. It supports AWS Database Migration Service (DMS) for easy migration.
  • Redshift Materialised Views
    Amazon Redshift now supports incremental refreshes for materialised views on Apache Iceberg and standard AWS Glue tables, enhancing efficiency and reducing the need for full refreshes.
  • Redshift Zero-ETL Lake House Connectors
    Simplified data absorption from various sources.
  • Amazon DocumentDB Vector Search
    Amazon DocumentDB now supports vector search, enabling the storage, indexing, and searching of millions of vectors with millisecond response times.
  • Amazon OpenSearch Serverless Vector Engine
    General availability of vector engine for Amazon OpenSearch Serverless, offering a scalable and high-performing vector database for machine learning–augmented search experiences and generative AI applications.
  • Amazon MemoryDB for Redis Vector Search (Preview)
    Amazon MemoryDB for Redis now supports vector search in preview, allowing the development of real-time machine learning and generative AI applications.
  • Amazon OpenSearch Service Zero-ETL Integration with Amazon S3 (Preview)
    A new way for customers to query operational logs in Amazon S3 and S3-based data lakes without switching between tools.
  • Amazon DynamoDB Zero-ETL Integration with Amazon OpenSearch Service
    Advanced search capabilities for Amazon DynamoDB data, allowing seamless synchronisation with Amazon OpenSearch Service without custom code.
  • Amazon Redshift Serverless Enhanced Manageability and Usability Features
    New features for Amazon Redshift Serverless, including cross-account cross-VPC, custom domain name (CNAME), snapshot scheduling, and cross-region copy (CRC).


  • Inspector Updates
    New open-source plugins and API for assessing container images for vulnerabilities. Amazon Inspector now offers agentless vulnerability assessments for Amazon EC2 instances in preview, expanding vulnerability assessment coverage across EC2 infrastructure.
  • Amazon Inspector Snapshot-scanning
    New capability for instance and snapshot scanning.
  • AWS Security Hub
    New capabilities in AWS Security Hub allow security teams to centrally enable and configure Security Hub standards and controls across accounts and Regions.
  • AWS Security Hub Tag Enrichment
    New metadata enrichment for findings in AWS Security Hub, adding resource tags, a new AWS application tag, and account name information to every finding.
  • Amazon One Enterprise
    A palm-based identity service for enterprise access control, offering a secure, convenient, and contactless experience for accessing physical locations and digital assets. It combines palm and vein imagery for biometric matching with an accuracy rate of 99.9999%


  • CloudWatch Logs Insights and AWS Config Queries
    GenAI capabilities for enhanced query generation.
  • CloudWatch Logs Anomaly Detection
    General availability for anomaly detection and pattern recognition.


  • Application Load Balancer Updates
    Support for Automatic Target Weights and mutual authentication.

Business Applications

  • WorkSpaces Thin Client
    A new device providing secure access to virtual desktops.

Quantum Computing

  • Quantum Computing Chip
    A new chip designed to reduce quantum error correction.


  • AWS Backup Restore Testing
    Automated periodic restore testing for compliance.
  • Elastic DR Drill Validation
    Enhanced validation for EC2 disaster recovery.
  • EBS Snapshots Archive for AWS Backup
    Support for transitioning EBS Snapshots created by AWS Backup to EBS Snapshots Archive for low-cost, long-term storage.
  • AWS FIS Multi-Account Experiments
    AWS Fault Injection Service (FIS) now supports multi-account experiments, allowing the setup and execution of real-world failure scenarios across multiple AWS accounts.
  • Amazon EFS Replication Failback
    Amazon EFS Replication now supports failback, allowing easier and more cost-effective synchronisation between EFS file systems after Disaster Recovery (DR) and other failover events.

Other Updates

  • AWS SDKs for Rust and Kotlin
    General availability of the AWS SDK for Rust and Kotlin, providing idiomatic, type-safe API and support for modern language features, including support for asynchronous AWS service calls using coroutines in Kotlin.
  • Glue Data Catalog Multi-Engine Views
    AWS Glue Data Catalog now supports the creation, management, and access control of SQL views that support multiple engines, allowing querying from Amazon Athena, Amazon Redshift, and Spark with Amazon EMR on EC2.
  • QuickSight and Glue Additions
    Numerous updates, including CloudTrail Lake support for Athena.
  • FinOps 
    Several new features to support FinOps monitoring, alerts, and reporting in AWS. For example, a prebuilt cost monitoring dashboard for QuickSight and cost optimisation hub. 

Wrapping Up

The dominant theme at this year’s re:Invent was undoubtedly the emphasis on GenAI and machine learning, signalling AWS’s commitment to lead in these areas. The introduction of tools like Amazon Bedrock and Amazon Q reflects a strategic move to integrate AI more deeply into cloud services, offering customers new ways to leverage AI for business innovation.

The advancements in compute and storage, particularly with the EC2 R8g instances and S3 Express One Zone, demonstrate AWS’ ongoing efforts to enhance performance and efficiency. These developments will likely be well-received by organisations dealing with data-intensive applications, such as AI/ML training and financial modelling.

AWS continues to innovate in databases and security, ensuring that its offerings are robust but also secure and compliant with evolving industry standards. The updates to Aurora and ElastiCache and the new Inspector and CloudWatch features indicate AWS’s commitment to providing robust and secure cloud solutions.

The focus on AI and machine learning at re:Invent 2023 suggests that AWS is positioning itself at the forefront of these technologies. This strategic direction will likely influence the cloud computing landscape significantly as AWS continues to push the boundaries of what’s possible in the cloud.

Given AWS’ focus, enterprises should consider their big data and ML strategy in the coming year. Many new tools and capabilities are available, and teams will want access to them, aiming to improve their productivity and innovation. Integrating and making these available within a large enterprise in a secure and compliant way however, takes time and will need the right approach to avoid unnecessary risk exposure.

Thomas is a Lead Consultant at Sourced with over 18 years of experience in delivering digital products. He gained a passion for evangelising Serverless architecture through his enthusiasm for cloud and entry into the AWS Ambassador programme.