Recapping Matt Garman’s AWS re:Invent 2024 Keynote Presentation

One of the best parts of AWS re:Invent every year is the CEO keynote presentation. This is when we learn about exciting AWS updates and how the company is thinking about the future overall. Former AWS CEO Adam Selipsky spoke extensively last year about how AWS is set apart from other cloud service providers. He spent a lot of time on AWS’s generative AI capabilities and how the AWS team is working hard to make GenAI accessible to the masses.

This year, CEO Matt Garman welcomed everyone to the 13th annual AWS re:Invent and took us through major updates across key AWS building blocks – compute, storage, databases, AI/ML, and much more. He brought in leaders from industry giants – Apple, JPMorgan, PagerDuty – to the stage to share how each is leveraging AWS infrastructure today. We also heard from various startups about how AWS enables cutting-edge innovation and growth.

It’s clear from Garman’s AWS re:Invent 2024 keynote address that AWS is pouring energy into the intersection of AI and cloud computing. The company continues to invest heavily in both its infrastructure and services in collaboration with developers, strategic partners, and the broader community. Now more than ever, organizations can tap into the cloud – either for highly specific use cases or for wide-ranging digital transformation – to deliver better experiences, products, and services.

AWS Compute

AWS provides more compute instances than any other cloud platform in the world. Customers can access 850 different instance types, each designed to fulfill specific workloads and use cases. This year, AWS has been working on several major upgrades. Garman highlighted or introduced the following:

• AWS Graviton 4: 40% better price performance with 60% less energy consumption over Graviton 3
• P6 family of instances: will utilize Nvidia’s Blackwell GPUs and be available in 2025
• Trainium2 (Trn2) instances: next-gen compute designed for AI workloads with Trn3 also in development
• Trn2 UltraServers instances: for combining up to 4 Trn2 instances together to build larger AI training clusters

On the compute side, AWS is always trying to achieve maximum price and cost performance. The company is doing this while still meeting the unique needs of different organizations with specific business requirements.

AWS Storage

Every year, storage on AWS gets better. 2024 was no exception, and things are only becoming more exciting. Garman announced the following releases:

• Amazon S3 Table Buckets: new bucket type specifically made for Iceberg tables to improve query performance and latency across many data sources
• Amazon S3 Metadata: enables users to better understand S3 data quickly. Combined with S3 table buckets, learning about stored objects in AWS is easier and faster.

AWS Databases

AWS’ database offerings took another step forward this year. Ten years after releasing Amazon Aurora, Garman announced the following:

• Amazon Aurora DSQL: new managed and distributed SQL databases with unlimited scalability, low latency, high consistency, and high availability
• Dynamo DB Global Tables upgrade: similar improvement to Amazon Aurora DSQL to enable strong multi-region consistency with minimal latency

GenAI & AI/ML

Building on last year’s keynote, Garman had a lot to share on the GenAI and AI/ML side during his AWS re:Invent 2024 keynote. He mentioned several times the potential for GenAI to completely transform every industry and organization. AWS’s new and upcoming releases in this area bring us closer to this reality:

• Amazon Bedrock Model Distillation: allows users to quickly transfer knowledge from large and complex models to smaller models
• Amazon Bedrock Automated Reasoning: checks the factual accuracy of model outputs to minimize hallucination risk and impact
• Amazon Bedrock multi-agent collaboration: enables teams to orchestrate many AI agents with minimal complexity
• Amazon Q Business workflow automation: for building detailed, automated workflows using natural language
• Amazon Q Developer Transformation for VMware workloads: for generating VMware to AWS migration plans automatically for existing VMware workloads
• Amazon Q Developer agents: three new autonomous agents to help speed up mainframe modernizations
• Amazon SageMarker Unified Studio: next-gen SageMaker release with a broader capability set for data and AI development
• Amazon SageMaker Lakehouse: for unifying data across all S3 data lakes and Redshift data warehouses

Amazon CEO Andy Jassey also made an appearance to share about Amazon Nova, a new generation of foundational models for frontier intelligence. He introduced Nova Canvas for generating images and Nova Reel for generating short videos, both from text inputs. Amazon Nova is available in three understanding models today – Micro, Lite, and Pro — with a Premier multimodal coming soon.

As we learned during the AWS re:Invent 2024 keynote, it’s an exciting time to take advantage of AI capabilities in the cloud. AWS recognizes this special moment and keeps delivering better tools to empower customers in every industry.

Maximize Cloud Performance on AWS with ClearScale

Since 2011, ClearScale has focused on delivering exceptional cloud projects on AWS exclusively. No other AWS professional cloud services company has the same breadth and depth of services available, and our AWS experts know how to guide teams to take full advantage.

Our AWS Premier Tier Services Partner status gives us early access to new AWS tools and best practices, which we then put to use in client engagements. Companies that work with us reduce execution risk, speed up time-to-value, and improve operation efficiency across even the most complex of IT footprints. Explore our case studies for a sample of some of the work we’ve done for customers across all industries.

Then, schedule a call to discuss your project with one of our cloud engineers or solutions architects.