CEO Keynote with Matt Garman

AWS CEO Matt Garman covers the newest features and technologies that are revolutionizing the cloud space. AWS keeps reinventing scalability and efficiency for companies of all sizes with its advanced AI integrations and improved security solutions. These developments demonstrate their dedication to giving businesses the resources they need to innovate more quickly and accomplish more.

Amazon EC2

The most powerful and cost-effective instances in the cloud.

• Largest amount of storage with D3en | 336 TB

• Largest in-memory footprint with U7i | 32 TiB

• Fastest networking with P5en | 3.2 Tbps network bandwidth

• First to bring technologies from intel, AMD and NVIDIA to the cloud.

• First and only major cloud provider to offer Mac-based instances.

AWS Nitro System

• Enhanced Performance: High-performance computing is provided by the AWS Nitro System, which transfers virtualized tasks to specialized hardware.

• Improved Security: By separating software and hardware resources, it improves security for delicate tasks.

• Cost Efficiency: Better resource utilization is made possible by Nitro, which lowers expenses for clients without sacrificing functionality.

AWS Graviton

• High Performance: Cloud-native and AI applications are among the workloads for which AWS Graviton processors are tuned for performance.

• Cost Efficiency: Their price-performance ratio is up to 40% better than that of conventional CPUs.

• Energy Efficiency: Graviton processors, which are based on the ARM architecture, are energy-efficient and promote environmentally friendly computing methods.

• Pinterest's move to AWS Graviton demonstrated the processors efficiency and sustainability benefits for contemporary workloads by reducing infrastructure costs by 47% and carbon emissions by 62%

AWS Graviton 4

• A few months ago, AWS unveiled the Graviton4 processors, which provide 75% more memory bandwidth and 30% greater compute performance than the Graviton3.

• They are perfect for applications that require a lot of memory because they can power instances like X8g with up to 192 vCPUs and 3 TiB of DDR5 memory.

• AWS's most energy-efficient CPU, Graviton4, is built for sustainability and scalability.​

AWS GPUs

• AWS offers a variety of GPU instances, including CG1, G2, P2, G3, P3, G4, P4, P5, G5, G5g, P5en, G6, and G6e, tailored to different AI, machine learning, and high-performance computing needs.

• Additionally, AWS introduced the P6 instance with NVIDIA Blackwell GPUs, delivering enhanced capabilities for demanding workloads​.

Amazon EC2 Trn2 Instances

• AWS announced the next generation of Trn instances, which are now widely available and powered by Trainium2, providing the best price-performance for generative AI on AWS.

• Performance is up to 30–40% better than that of GPU-based instances of the current generation.

• 16 Trainium2 chips

• Up to 20.8 FP8 Petaflops

Amazon EC2 Trn2 UltraServers

• It’s available in preview today.

• 4 Trn2 instances connected with Neuronlink.

• 64 Trainium2 chips

• Up to 83.2 FP8 petaflops

AWS Trainium3

• AWS announced Trainium3, a platform designed specifically for the upcoming generation of generative AI tasks. and it will happen soon.

• 3NM Process Node

• 40% More energy efficiency

• 2x performance

Amazon Aurora DSQL

• The fastest distributed SQL database was announced by AWS and is currently available for preview.

• Virtually unlimited scalability.

• Zero infrastructure management.

• 99.999% multi-region availability.

• Strong consistency

• PostgreSQL-compatible

Amazon DynamoDB global tables

• According to AWS, multi-region strong consistency is now supported and is available in preview today.

Amazon Bedrock Model Distillation

• AWS declared Make models more quickly and affordably, a preview is accessible right now.

• Transfer information from a large, complex model to a smaller one with ease.

• Distilled models are 75% less expensive and 500% faster.

• Anthropic, Meta and Amazon models.

Amazon Bedrock multi-agent collaboration

• AWS revealed that it is now available in preview form and that it is simple to design, deploy, and coordinate groups of agents that collaborate to manage intricate, multi-step operations.

• Orchestrate agents without complex coding.

• Accelerate tasks with agents working in parallel.

Amazon Nova

• AWS revealed Amazon Nova, a brand-new family of cutting-edge foundation models (FMs) that are only available on Amazon Bedrock and offer industry-leading pricing performance and frontier intelligence.

• For practically any generative AI task, Amazon Nova can reduce latency and expenses. With Amazon Nova, you can create sophisticated AI agents from a variety of intelligence classes tailored for enterprise workloads, analyze intricate documents and videos, comprehend charts and diagrams, and create captivating video content.

• Three understanding models are included in the Amazon Nova models (a fourth will be released soon) and are intended to address various needs:

Amazon Nova Micro

• An inexpensive, speed-optimized text-only model. It does exceptionally well in tasks like conversation, content classification, translation, text summarization, and basic reasoning.

• It allows for fine-tuning and customisation, and its context length is 128K tokens.

Amazon Nova Lite

• A quick, low-cost multimodal model that can process video, picture, and text inputs. It is perfect for activities like document analysis, visual question answering, and real-time customer service.

• Up to 300K tokens, several photos, or 30 minutes of video can be processed by it.

Amazon Nova Pro

• For complicated jobs, a strong multimodal model that provides the optimal ratio of accuracy, speed, and cost.

• It can process textual and graphic data, including financial documents and vast code bases, and can manage up to 300K tokens. It can also be used to teach distillation of other models.

Amazon Nova Premier (Coming soon)

• The most complex multimodal model, intended for training and challenging thinking problems.

• It is intended for the distillation of custom models and will be accessible in early 2025.

The Amazon Nova models also include two creative content generation models:

Amazon Nova Canvas

• With exact control over style and substance, this advanced picture creation model produces photographs of studio quality.

• It performs exceptionally well in text-to-image jobs and human evaluations because to features like inpainting, outpainting, and backdrop removal.

Amazon Nova Reel

• A state-of-the-art video generating model for producing brief videos using images and text instructions. It provides professional-quality videos for promotion and entertainment by giving users control over visual style and pacing.

• It performs better than other models in terms of consistency and video quality.

Amazon Nova Speech-to-Speech

• Coming soon

Amazon Nova Any-to-Any

• Coming soon

Amazon Q Capabilities

Amazon Q generates and applies unit tests

• Automate Unit Tests: For your project files, you may automate the process of finding test cases and creating unit tests that cover scenarios such as numerous input types, boundary conditions, and null values.

• Test Generation: Unit tests are created directly in your source file by Amazon Q Developer, which also handles error self-debugging and places them in the appropriate test file.

• Review & Approval: You can use the code editor's "View diff" feature to examine the tests once they've been generated and determine whether to accept or reject the modifications.

Amazon Q generates accurate documentation

• Generate Documentation: While guaranteeing high-quality output, Amazon Q Developer helps developers save time by automating the development of documentation such as readmes and data flow diagrams.

• README Creation: In your project, a README file can be created or updated. The program generates the documentation, summaries the material, builds a knowledge graph, and reads source files.

• Review & Accept: You can examine the README in the code editor after it has been generated and decide whether to approve its use in your project.

Amazon Q performs code reviews

• Automated Code Review: Code smells, anti-patterns, bugs, logical mistakes, code duplication, inadequate documentation, and security flaws are among the problems with code quality that Amazon Q Developer finds and fixes in both IDEs and GitLab repositories.

• Boosts Productivity: The automatic review speeds up feature releases while following best practices by saving time, improving consistency, and assisting teams in maintaining high code quality.

• Review & Fix: Before committing, Amazon Q Developer examines your project or chosen file, identifies any problems, and patches code as needed. A straightforward "Accept Fix" option allows you to examine and implement fixes right in the code editor.

Amazon Q Developer Transform .Net applications from windows to linux in a fraction of the time

• It’s available in preview today.

• Update.Net applications to run up to four times quicker on Linux and Windows.

• Hundreds of apps can be transformed simultaneously by agents.

• Improve security and performance.

• Reduce costs by up to 40%.

Amazon Q Developer Transform VMware workloads to cloud native architectures

• It’s available in preview today.

• Agents plan, determine dependencies, and change network configurations on their own.

• Increase stability and security.

• Reduce costs and license fees.

Amazon Q Developer Transform mainframe applications to accelerate migrations

• It’s available in preview today.

• Transform IBM z/OS mainframe applications.

• Automate analysis, discovery, planning and refactoring

• Generate documentation for millions of lines of code.

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

• It’s available in preview today.

• Because Q in the console has a thorough awareness of your AWS resources, troubleshooting is accelerated.

• Offers a supervised process for troubleshooting problems that is available from any location where you work with Amazon Cloudwatch data.

• Receive recommendations for AWS created documentation and automated runbooks to address problems fast.

ISVs can now integrate with the Amazon Q index

• It’s generally available.

• Easily access data from multiple applications.

• You fully control the permissions to your data.

• Better security with a single index.

Amazon SageMaker Capabilities

Amazon SageMaker Unified Studio

• These services are currently offered by AWS: machine learning, data processing, SQL analytics, and general artificial intelligence development.

• These services, which include streaming and business intelligence & search analytics, will be available soon.

Amazon SageMaker Lakehouse

• Simplify analytics and artificial intelligence with an open, unified, secure data lake house that is widely accessible.

• Integrated access to your data from multiple data sources, SaaS, Amazon Redshift, and S3 accounts.

• Dependable, fine-grained data management access controls.

• Apache Iceberg-compatible.

References

• Amazon Nova Next Generation

• Amazon Sagemaker Next Generation