31 AWS Announcements from Matt Garman’s AWS re:Invent 2024 Keynote

This year, I enjoyed attending the AWS re:Invent 2024 keynote session led by new CEO Matt Garman. This was his first re:Invent keynote since taking over from Andy Jassy. Crowds piled into the enormous theatre in The Venetian, Las Vegas, to hear AWS’s latest announcements. By sheer luck I was shepherded into a seat 3 or 3 rows back from the front, so I could capture all of the photos in this post myself! As you would expect, there were many AI announcements and some exciting new developments in the database portfolio—read on for more.

Compute

1. Amazon EC2 P6 instances

First up were the new EC2 P6 instances with NVIDIA Blackwell GPUs which have been added to the already extensive AWS GPU instance portfolio.

2. Amazon EC2 Trainium 2 Instances

Next, the new Amazon EC2 Trainium 2 instances launched into General Availability. They offer up to 30-40% better price performance than current generation GPU-based instances. 16 Trainium 2 chips offer up to 20.8 FP8 petaflops.

3. Amazon EC2 Trn2 Ultra Servers

In addition to the Trainium 2 GA launch, Matt also announced the preview launch of Amazon EC2 Trainium 2 Ultra Servers offering 4 Trainium 2 servers connected with NeuronLink, offering 64 Trainium 2 chips with up to 83.2 FP8 petaflops.

4. AWS Trainium 3 is coming next year

In addition to the Trn2 announcement, Matt also announced that the next generation Trainium 3 will be available next year. Expect a full announcement on this launch before Invent 2025.

Storage

5. Amazon S3 Table Buckets

S3 Table Buckets for Apache Iceberg tables were launched into General Availability, offering up to 5x faster query performance and up to 10x higher transactions per second.

6. Amazon S3 Metadata

Amazon S3 Metadata – the easiest way to manage your metadata in S3, was launched in Preview.

Databases

7. Amazon Aurora DSQL

Amazon Aurora DSQL was launced in preview. Amazon are claiming this to be the fastest distributed SQL database which is up to 4x faster than Google Cloud Spanner.

8. Amazon DynamoDB Global Tables

Amazon DynamoDB global tables were announced in preview, supporting multi-Region strong consistency.

Inference

9. Model Distillation in Bedrock

Amazon Bedrock Model Distillation was launched in preview, enabling faster, more effective model creation. This feature will enable users to easily transfer knowledge from a large, complex model to a smaller one. Distilled models can be up to 500% faster and 75% less expensive. Currently, Model Distillation supports Anthropic, Meta, and Amazon LLMs.

10. Amazon Bedrock Automated Reasoning Checks – Preview

Amazon Bedrock Automated Reasoning checks was launched in preview. Automated Reasoning Checks verifies the accuracy of LLM responses using mathematical proof, preventing factual errors due to hallucinations. The service provides recommendations for creating factual errors.

11. Amazon Bedrock multi-agent collaboration

Amazon Bedrock multi-agent collaboration was launched in preview. This enables users to easily build, deploy and orchestrate teams of agents that work together to handle complex multi-step tasks. Agents can be orchestrated without complex coding, and tasks can be accelerated with agents working in parallel.

LLMs

Matt Garman handed the stage to Amazon CEO Andy Jassy to discuss the new Amazon Large Language models. Jassy then announced several new Amazon Nova models.

12. Amazon Nova

First up was Amazon Nova, with Micro, Lite and Pro launched into general availability, with Amazon Nova Premier coming soon. Amazon Nova Micro is a text-only model that delivers the lowest latency responses at a very low cost. Nova Lite is a low-cost multimodal model that is lightning-fast for processing image, video and text inputs. Amazon Nova Pro is a competent multimodal model with the best accuracy, speed, and cost combination for various tasks. Amazon Nova Premier will be the most capable multimodal model for complex reasoning tasks and for use as the best teacher for distilling custom models.

13. Amazon Nova Canvas

Amazon Nova Canvas launched into General Availability, offering state-of-the-art image generation. It allows users to edit images using natural language inputs. It provides controls for colour scheme and layout. It has built-in controls for the safe and responsible use of AI, including watermarking for traceability and content moderation to limit the generation of harmful content.

14. Amazon Nova Reel

Amazon Nova Reel was launched into General Availability, offering state of the art video generation model. Amazon Nova Reel provides camera motion control, including pan motion, 360-degree rotation and zoom. It also has built in controls for safe and responsible use of AI, including watermarking and content moderation. Amazon Nova Reel supports 6-second videos at launch, with 2-minute videos coming soon.

15. Amazon Nova Speech-to-Speech

Amazon Nova Speech-to-Speech is coming soon. It will understand streaming speech input in natural language, interpret verbal and nonverbal cues (like tone and cadence), and deliver natural human-like interactions.

16. Amazon Nova Any to Any – coming soon

Amazon Nova Any-to-Any will be capable of processing text, images, audio, and video as both input and output.

Amazon Q Developer

17. Amazon Q generates and applies unit tests

18. Amazon Q generates accurate documentation

19. Amazon Q performs code reviews

Three new Amazon Q Developer features were launched into General Availability. Amazon Q can now generate and apply unit tests, generate accurate documentation and perform code reviews.

20. GitLab Duo with Amazon Q

GitLab Duo with Amazon Q was launched into preview, offering a seamless developer experience with new capabilities to accelerate feature development and transform workloads, all within the same familiar GitLab environment that developers are used to.

21. Amazon Q Developer – Transform NET applications from Windows to Linux in a fraction of the time

Announced in preview, Amazon Q Developer now enables the transformation of .NET applications from Windows to Linux in a fraction of the time. This will enable businesses to modernise Windows .NET applications to Linux up to 4 times faster and reduce costs by up to 40%.

22. Amazon Q Developer – Transform VMware workloads to cloud-native architectures

Also launched into preview was the capability to transform VMware workloads to cloud-native architectures. Agents autonomously plan, identify dependencies, and transform network configurations. This enables improved security and resilience while reducing costs and license fees.

23. Amazon Q Developer – Transform mainframe applications to accelerate migrations

Sticking with modernisation, attention turned to mainframes, with Amazon Q Developer being able to transform mainframe applications. This feature can transform IBM z/OS mainframe applications. It automates analysis, discovery, planning and refactoring, and can generate documentation for millions of lines of code.

24. PagerDuty Advance with Amazon Q

PagerDuty CEO Jennifer Tejada took the stage to announce the PagerDuty Advance with Amazon Q integration, which offers incident response teams a streamlined way to receive incident data.

Business Intelligence

25. Combining QuickSight and Amazon Q Business data

Matt Garman took back the stage to focus on new Business Intelligence offerings. First up, Amazon QuickSight and Amazon Q Business data can now be combined, giving a global view of data from databases, data lakes and data warehouses, combined with documents, purchase orders, customer emails, call logs and client no

26. ISVs can now integrate with the Amazon Q index

Next, Matt announced that ISVs can now integrate with the Amazon Q index, enabling easy access to data from multiple applications. Users can fully control data access permissions; a single index offers better security.

27. Amazon Q Business can automate complex workflows

Coming soon, Amazon Q Business will be able to automate complex workflows based on documentation or recordings. Users can navigate changes to workflows in real-time, reducing breakage.

28. Amazon Q Developer – Investigate issues across your AWS environment in a fraction of the time

Launched into preview, Amazon Q Developer can now investigate issues across AWS environments in a fraction of the tome. Q in the console accelerates troubleshooting with a deep understanding of AWS resources. Q can provide a guided workflow for investigating issues accessible from wherever you interact with CloudWatch data. Q will also offer suggestions for AWS automated runbooks and curated documentation to quickly resolve issues.

Analytics

29. SageMaker Unified Studio

Sagemaker Unified Studio now offers unified notebooks and query editors for SQL Anayltics, Data Processing, Machine Learning and AI Development. Soon it will include Streaming for Amazon MSK and Kinesis, Business Intelligence with Amazon Quicksight and Search Analytics with Amazon OpenSearch.

30. Zero ETL for 3rd Party SaaS Apps

AWS has announced a number of zero ETL integrations recently, and the latest offering is zero ETL integration with third-party SaaS applications.

31. Amazon SageMaker Lakehouse

The final announcement of the keynote was Amazon SageMaker Lakehouse launching into General Availability. SageMaker Lakehouse gives unified access to data across Amazon S3, Redshift, SaaS and federated data sources. It offers consistent, fine-grained access controls for data governance and is Apache Iceberg compatible.

31 announcements over 3 hours kept us on our toes, and there will be plenty more to follow throughout re:Invent 2024. Speaking to AWS Community Builders after the Keynote, the announcement that seemed to generate the most interest was Aurora DSQL, so it will be interesting to watch this space and see how well this service is adopted by AWS customers.