AWS re:Invent 2024
Las Vegas
From the 2nd of December to the 6th. Las Vegas transformed into the ultimate hub for the global cloud computing community — it was AWS’s flagship annual conference: reInvent. Packed with inspiring keynotes, in-depth technical sessions, exciting product launches, and endless networking opportunities.
Keynote Sessions
A keynote session is one of the most important and highly anticipated parts of any major conference, including AWS re:Invent. It’s a spotlight moment where influential leaders, visionaries, or experts in a field share big-picture ideas, groundbreaking announcements, and insights to set the tone for the event.
At AWS re:Invent, keynote sessions often feature high-level executives, who unveil major cloud innovations, new services, and the company’s future vision. These sessions provide attendees with a broad understanding of emerging trends in cloud computing and AWS’s strategy for staying at the forefront of the industry.
Think of a keynote session as the main stage performance at a music festival: it’s where the spotlight shines brightest, the crowd gathers, and the most exciting revelations happen!
SVP, Utility Computing, AWS
Monday Night Live with Peter DeSantis
Keynote Creativity with AI
Peter DeSantis used an AI assistant to make his keynote more engaging by generating numerous analogies, allowing him to choose the most fitting one. This introduction emphasized the benefits of GenAI, encouraging attendees to embrace AI collaboration without feeling diminished. The key message: AI is here to stay, so leverage it while you can.
DeSantis ultimately chose the analogy “Below the Surface,” which uses trees and their roots to represent his theme. Interestingly, not all trees have taproots, a fact he tied into AWS’s ability to innovate across stacks. He drew comparisons between the interconnectedness of roots and AWS’s innovation ecosystem, particularly the Nitro system.
Enter David Brown
VP of Compute and Networking at AWS
David Brown took the stage, beginning with a nod to the birth of EC2 in Cape Town — something I think he has mentioned this before at the Johannesburg Summit, and he should always mention it, because South Africa is awesome.
Brown discussed:
Custom Silicon Development: The journey began with Graviton in 2018, leading to Graviton 2, Graviton 3, and now Graviton 4, introduced in 2023. Graviton 4 is the most powerful chip yet, purpose-built for high-performance computing.
Graviton Optimization: Most services perform significantly better when powered by Graviton.
The AWS Nitro System: Nitro enhances server architecture, enabling features like running macOS on the cloud, improving security, and ensuring hardware integrity. Nitro solves problems that traditional on-premises data centers can’t address.
Hello Barge, Bye Barge: Brown highlighted AWS’s constant innovation in storage systems, referencing Barge, an ambitious attempt to push storage density boundaries. Despite its potential, AWS learned valuable lessons and moved on from Barge, focusing on building more resilient servers with reduced blast radius impacts.
The Return of Peter DeSantis
DeSantis returned to discuss AI workloads, describing them as a “scale-up” rather than a “scale-out” system due to the increasing size and capability of models.
Highlights included:
Data Parallelism: He explained how it optimizes AI training.
Trainium 2 Chip: DeSantis showcased the micro-level design of Trainium 2, including a live demonstration of the Trainium 2 server, a powerful infrastructure designed to meet customer demands for advanced AI workloads.
Guest Spotlight
Tom Brown, Co-Founder and Chief Compute Officer of Anthropic
Tom Brown introduced Claude, Anthropic’s AI model, and discussed its latest iteration, Claude 3.5 Haiku. He highlighted its smaller size, reduced costs, and compatibility with Trainium 2, reinforcing the message that AI is the way forward.
Announcements from Peter DeSantis’s Keynote
- Latency-Optimized Inference Option for Amazon Bedrock
- Anthropic Claude 3.5 Haiku
AWS continues to innovate at every level, from silicon to AI infrastructure, ensuring a future-ready cloud ecosystem.
CEO, AWS
CEO Keynote with Matt Garman
A Learning Conference for Builders and Developers
Matt Garman began by emphasizing that re:Invent is a learning conference dedicated to builders and developers, bringing together the AWS community from around the globe.
AWS Community
He celebrated the growth of the AWS community, which now boasts 600 user groups across various countries. He also highlighted the contributions of AWS Heroes, showcasing their vital role in driving community innovation.
Startups and GenAI
Garman recognized the agility of startups and the transformative impact of generative AI (GenAI) as a disruptor in the startup ecosystem. He announced AWS’s plan to invest $1 billion in startups globally in 2025, reinforcing AWS’s commitment to fostering innovation.
Customer Obsession
Acknowledging AWS’s roots in customer obsession, Garman urged attendees to always focus on business needs and customer satisfaction when driving innovation.
Compute Innovations
Flexibility of EC2: He highlighted the flexibility of EC2, which enables a broad range of workloads.
Nitro System: AWS remains the only cloud provider capable of running macOS, thanks to the Nitro system’s flexibility.
Graviton 4: The progression from Graviton to Graviton 4 was emphasized, with a success story from Pinterest, which moved to Graviton 4 and reduced costs by 47%.
NVIDIA: Garman also underscored AWS’s 14-year collaboration with NVIDIA, noting that NVIDIA runs its GenAI workloads on AWS.
Enter Benoit Dupin
Senior Director of AI and ML at Apple
Dupin’s appearance marked a rare confirmation of Apple’s use of AWS services. He shared insights into Apple’s collaboration with AWS over the years.
Compute Announcements
- P6 NVIDIA Blackwell GPUs, arriving early next year.
- Trainium 2 Instances and UltraServers for AI workloads
- Preview of Trainium 3, the next-generation AI infrastructure.
Storage Innovations
Garman discussed the importance of scalable storage and announced:
- S3 Table Buckets for improved scalability and query performance with Iceberg Tables.
- S3 Metadata for enhanced file discovery.
Database Innovations
Celebrating the 10th anniversary of Amazon Aurora, Garman highlighted:
- Amazon Aurora DSQL for advanced database queries.
2. Multi-Region Capabilities for DynamoDB Global Tables, ensuring resilience and performance.
AI as a Building Block
Garman referred to AI as a foundational element of cloud computing and invited Andy Jassy, President and CEO of Amazon, to the stage.
Andy Jassy: The Godfather of the Cloud
Jassy reaffirmed the transformative power of AI in solving customer problems, boosting productivity, and reinventing applications. He emphasized that AI is here to stay and will only get bigger and better.
AI Announcements
- Amazon Bedrock: New features like Model Distillation, Automated Reasoning Check, and Multi-Agent Collaboration.
2. Amazon Nova: Tools like Canvas, Reel, Speech-to-Speech, and Any-to-Any communication.
3. Amazon Q: Revolutionary developer tools to automate and accelerate tasks, including code reviews, migrations, and environment assessments.
Analytics and AI Convergence
Garman emphasized the seamless integration of analytics and AI, unveiling:
- The Next Generation of SageMaker.
- Zero-ETL Integrations for efficient data processing.
- The Amazon SageMaker Lakehouse, combining analytics with AI capabilities.
Announcements from Matt Garman’s Keynote
Compute
01. P6 NVIDIA Blackwell GPUs (early 2025).
02. Trainium 2 Instances and UltraServers.
03. Trainium 3 for advanced AI workloads.
Storage
04. S3 Table Buckets for Iceberg Tables.
05. S3 Metadata for file discovery.
Database
06. Amazon Aurora DSQL.
07. Multi-Region Capabilities on DynamoDB Global Tables.
AI and Developer Tools
08. Amazon Bedrock updates: Model Distillation, Automated Reasoning Check, Multi-Agent Collaboration.
09. Amazon Nova: Canvas, Reel, Speech-to-Speech, and Any-to-Any communication tools.
10. Amazon Q: Automation tools for unit testing, documentation, code reviews, and migrations.
Analytics
11. The Next Generation of SageMaker.
12. Amazon SageMaker Studio
13. Zero-ETL Integrations.
14. Amazon SageMaker Lakehouse.
AWS continues to innovate across compute, storage, databases, and AI, solidifying its position as a leader in cloud technology.
VP, AI and Data, AWS
Dr. Swami Sivasubramanian Keynote
Dr. Swami Sivasubramanian focused on AI innovation and the growing adoption of AI across industries. He also unveiled exciting advancements in the next generation of Amazon SageMaker and other key AI offerings.
Announcements from Dr. Sivasubramanian’s Keynote
Amazon SageMaker Updates
01. Hyperpod Flexible Training Plans: Streamlined training workflows with flexibility.
02. Hyperpod Task Governance: Improved task management for AI training.
03. AI Apps from AWS Partners Now Available in Amazon SageMaker: Expanding the ecosystem with partner integrations.
Amazon Bedrock Updates
04. Poolside AI: Coming soon to Bedrock.
05. Stability Diffusion 3.5: Enhanced generative AI capabilities.
06. LumaAI: coming soon to Bedrock
07. Bedrock Marketplace: Centralized marketplace for AI models and tools.
08. Prompt Caching: Faster and more efficient prompt handling.
09. Intelligent Prompt Routing: Optimized prompt workflows.
Knowledge Bases
10. Kendra GenAI Index: Advanced search and indexing capabilities.
11. Structured Data Retrieval Support: Enhanced data querying in Bedrock Knowledge Bases.
12. GraphRAG Support: Integration of graph-based retrieval-augmented generation.
Data Automation and Safety
13. Amazon Bedrock Data Automation: Streamlining data workflows.
14. Guardrails for Multimodal Toxicity Detection: Enhanced safeguards for content moderation.
AI for Business and Education
15. Amazon SageMaker Canvas: Simplified AI model creation for non-technical users.
16. Amazon Q in QuickSight Scenarios: Improved data visualization and insights.
17. AWS Education Equity Initiative: Leveraging AI for educational access and inclusivity.
Dr. Sivasubramanian’s keynote demonstrated AWS’s commitment to AI accessibility, innovation, and safety, ensuring tools and solutions are optimized for businesses, developers, and educators alike.
VP, Global Specialists and Partners, AWS
AWS Partner Keynote with Dr. Ruba Borno
Dr. Ruba Borno unveiled exciting updates, including four new specializations and advancements in the AWS SAP Partnership program, further emphasizing AWS’s commitment to empowering its partner ecosystem.
Announcements from Dr. Ruba Borno’s Keynote
New Specializations
AI Security: Addressing security challenges in AI implementations.
Digital Sovereignty: Ensuring compliance with local data governance requirements.
Amazon Security Lake: Specialization in centralized security data storage.
AWS Security Incident Reporting: Tools for managing and reporting security incidents.
Partner Program Enhancements
01. SaaS Co-Sell Incentive for Startups: Supporting startups with collaborative selling opportunities.
02. Buy with AWS: A streamlined purchasing experience for customers.
03. Partner Connections: Enhanced collaboration tools for AWS partners.
Dr. Borno’s keynote highlighted AWS’s dedication to fostering innovation and success within its global partner network.
VP and CTO, Amazon.com
Dr. Werner Vogels Keynote
As the saying goes, “Save the best for last,” sung by Vanessa Williams, Dr. Werner Vogels closed with his signature blend of storytelling and wisdom. The theme of his keynote?
Simplexity — the delicate balance between simplicity and complexity in system design.
Dr. Vogels emphasized core principles:
Build for the customer, build for the business
Building for the customer and the business requires striking a balance between meeting user needs and aligning with long-term organizational goals.
Plan for failure
In a cloud-native world, failure is not an “if” but a “when.” Planning for failure means designing systems that expect and gracefully handle errors.
Keep it simple (stupid)
The KISS principle emphasizes that simplicity is the ultimate sophistication. In cloud computing, simplicity minimizes errors, reduces maintenance costs, and enhances scalability.
The distinction between intended vs. unintended complexity
Understanding the nature of complexity in your systems is key to managing it effectively.
Enter Brendan Humphreys
CTO of Canva
Brendan, with his captivating voice, shared Canva’s journey of evolution and its growing need for scalability. His insights on scaling resonated deeply with the theme of managing complexity in the business while maintaining simplicity for the customer.
The Return of Dr. Vogels
Dr. Vogels reminded the audience that evolution is fundamental and change is constant. He explained that achieving simplicity demands discipline and unveiled his Six Lessons to Prevent Complexity:
1. Make Evolvability a Requirement
Your systems will grow over time. Design them to evolve in a controllable manner, accommodating future changes with ease. As Vogels put it, “A system’s ability to easily accommodate future change is key.”
2. Break Complexity into Pieces
Avoid ignoring warning signs. Decompose problems into manageable parts to maintain clarity.
3. Align Your Organization to Your Architecture
Structure teams in a way that mirrors your system’s architecture to streamline communication and decision-making.
4. Organize into Cells
Create autonomous, self-sufficient units to foster flexibility and resilience.
5. Design Predictable Systems
Ensure systems behave consistently, even under stress. Predictability reduces unintended surprises.
6. Automate Complexity
Use automation to handle complex tasks, minimizing human error and saving time.
This approach might just be the future of human intelligence, echoing principles of classic computational thinking. The keynote left the audience reflecting on the beauty of simplicity in a complex world.
