Amazon re:Invent Recap – Cloud, AI, Data and more

Amazon just held their annual Amazon Web Services (AWS) re:Invent conference in Las Vegas with more than 50,000 attendees in-person with many more attending online. The conference is so large that it spans the entire Las Vegas strip from the Wynn to Mandalay Bay and several hotel complexes in between. This year marked my tenth re:Invent conference and while AWS has evolved, the past couple of years have shown a marked shift for AWS.

AWS continues to hold the title as the leading public cloud provider. Since the start, cloud computing has matured and evolved in how it is consumed by both web scale companies and enterprises alike. While AWS has historically been a builder-focused company, this year AWS made a marked shift in their focus which should pique the interest of enterprises.

Sports in action

Prior to the official opening of the event, a small group of folks were invited to get a sneak peek into how AWS is involved in sports. This was a great way to see AWS’ solutions in action…literally. From golf to football simulators, you had an opportunity to see how this technology is used in professional sports and give it a try firsthand. For example, professional sports are able to leverage telemetry from the ball, uniforms and helmets to perform analytics in AWS to a) improve safety for the sport and b) improve player performance. If you play this out, one starts to wonder how to strike a healthy balance between science and art. AWS is partnering closely with professional sports organizations to play this out. To say the experience was impressive would be an understatement.

AWS four main areas of investment

Kicking off the event, AWS’ Dr. Matt Wood ran through key announcements and stated that there are four main areas of investment for AWS:

• Infrastructure Innovation: This includes AWS’ core services including silicon, compute, storage and databases.
• Bringing it all Together: Making it easier to use and manage applications on AWS.
• Data and Analytics: Simplifying data analytics for enterprises.
• Generative AI: Building solutions that leverage generative AI.

While infrastructure innovation is AWS’ bread and butter, AWS’ focus on making it easier to use and manage applications along with innovations in data, analytics and generative AI open the door to attracting more enterprise interest.

Key takeaways from AWS re:Invent announcements

Of the well over 150 announcements made at the conference, here are a few that stood out.

1. Custom Silicon: On the surface, enterprise leaders may question the importance of custom silicon. That would be a mistake as the consequences are of great interest to enterprises looking to tip the price/ performance balance. I spoke with AWS customers that are specifically using AWS because of their drive toward a strong price/ performance outcome. While AWS continues to partner with leading chip manufacturers such as Intel, Nvidia and AMD, at re:Invent, AWS announced their own new chips including Graviton 4, Trainium 2 and Inferentia 2 which continue to build on their multi-generational custom silicon experience. The improvements to Graviton 4 over the prior generation are impressive. According to AWS, Graviton 4 includes 50% more cores, 70% more memory bandwidth and 30% faster. In addition, Graviton 4 is 60% more power efficient…which is a huge improvement. More on that in a minute. Trainium 2 is four times faster with 65 exaflops than the prior generation. Inferentia 2 is also four times faster with 10 times lower latency than the prior generation. From a customer standpoint, over 50,000 customers are using Graviton with all 100 of AWS’ top 100 customers using Graviton. These enhancements are paired with AWS’ 5^th generation Nitro system to provide high-performance security options.
2. Amazon Q: Q is AWS’ generative AI powered chatbot. Q layers on top of custom AWS applications in addition to bringing natural language capabilities to existing tools such as AWS QuickSight, Amazon Connect and (soon) AWS Supply Chain. A key feature is that Q will leverage existing governance models to ensure that responses and data still adhere to policies. Interesting note about Q is that it demonstrates a continued move to provide tooling that includes operators and users, not just builders. This is part of the shift for AWS to address the operator and user personas. In fact, when AWS presented details about Q, they discussed it from the perspective of developer/ builder, then lines of business then specialists. Each persona will find different value in solutions like Q.
3. Expanding Personas: Historically, AWS focused their tools and services on the builder or developer persona. There are a few exceptions of end-user focused solutions including WorkSpaces and Chime. In the last couple of years however, AWS expanded their focus to include the operator and builder personas. In addition to solutions like WorkSpaces and Chime, newer solutions like QuickSight and Connect continue this expansion for AWS. This year, Q takes aim to provide a chatbot interface for users and builds on AWS’ existing success with tools like QuickSight and Connect. In addition, Q includes over 40 connectors to other systems such as Salesforce, Slack, Zendesk, ServiceNow and more. I believe this is a positive move for AWS as they more directly address the needs to enterprises.
4. Generative AI: In 2023, a conference would not be taken seriously without considering generational AI. In the past several months, AWS has taken some heat about not focusing on their own foundational model (FM) and large language models (LLM). When I speak with enterprises, they are more familiar with solutions from OpenAI, Cohere, Anthropic and Meta than AWS’ own generative AI models called Titan. While AWS has announced its commitment to invest up to $4 billion in Anthropic, AWS’ approach is to provide model choice to developers. At re:Invent, AWS stated that Anthropic is training their models on AWS which does suggest potentially stronger integration between AWS and Anthropic’s Claude models. I believe that model selection will be more important for enterprises than focusing on one model as each model has specific purposes and are not directly comparable. AWS is bringing image generation and multimodal embeddings to their Titan models.
5. Guardrails for Bedrock: In AWS’ continued drive toward responsible AI, Guardrails provide the ability to implement safeguards around how AI is used within orgs. There is a lot under the hood on this one as policy development and management is a complicated space.
6. Sustainability: Sustainability played a big role at AWS’ re:Invent conference this year. While there seems to be less interest broadly by CIOs right now around sustainability, I expect a marked change starting next year. Legislation, performance and customer requirements are going to be coming hot and fast to most enterprise organization. Unfortunately, most enterprises are not ready for it. AWS’ approach has been consistent and addressing these challenges for customers. From Graviton 4 custom silicon to data center design using free air and liquid cooling, AWS continues to improve their footprint. Amazon is on a path to powering its operations with 100% renewable energy by 2025. As of 2022, they are at 90%. To be clear, this is a complicated problem. AWS has also committed to being water positive by 2030 meaning they contribute more water to communities than they use in their direction operations. It includes both AWS’ operations in addition to their upstream and downstream partners.
7. FinOps Foundation: One of the key issues with cloud for enterprises is around cost management. AWS announced that it has joined the FinOps Foundation. The FinOps Foundation is a Linux Foundation project to advance the discipline of cloud financial management. While AWS does have cost management tools to assist customers, there is more work to be done both internally and externally. Bringing in Q with QuickSight along with best practices from the FinOps Foundation could accelerate this work.

In summary

Overall, AWS re:Invent is back to pre-pandemic attendance and engagement levels. It was great to see how AWS’ solutions play out in the real world with strong customer examples including professional sports, manufacturing, healthcare and financial services. What is interesting is that their use of AWS is not just as a ‘tool’ but as a solution that drives their business success. That’s a marked shift from prior years and a positive sign for AWS.

At the same time, there are strong indications that a) AWS is going more heavily after enterprise needs and b) is maturing in their thinking around how solutions are consumed. Both are positive moves for AWS. At the same time, AWS will need to continue their work to ensure they stay closely aligned with shifting enterprise needs.