Takeaways and Ramblings from AWS re:Invent 2016

Last week, Amazon held its premier AWS re:Invent conference in Las Vegas. The conference essentially took over the Venetian, Mirage, Wynn, and Encore hotels, with over 30,000 attendees in attendance. This was my first time attending re:Invent and I was blown away by the quality of speakers, attendees, and content that existed there. It was awesome.

I spent about 48 hours at re:Invent, splitting my time between attending sessions, meeting with interesting people, wandering the expo hall, and, of course, hanging out with startups and other VCs at cocktail hours/dinners. And there were maybe a few hands of black jack in there, but I digress. Here are some of my takeaways...

Takeaway #1: Amazon (namely AWS) has an incredible knack for continuing to innovate even as a massive company.

Amazon Web Services has undeniably changed the game for software infrastructure. AWS originally made its name by replacing massive amounts of on-premise server deployments with cloud deployments. A good example of this happening: Netflix announced during the conference that this year it has replaced its last standing on-prem server with AWS, so is now 100% in the cloud.

However, AWS has been making serious moves beyond just providing basic storage and compute services in the cloud. The newest tools announced at re:Invent this year have moved higher up in the stack and wider across new use cases (like IoT and AI). The ethos of the organization seems to be “make developers’ lives easy.” This attracts developers from the bottom up, improves developer productivity, and in result convinces companies that they must use AWS as their infrastructure provider of choice.

The higher up the stack that AWS goes, the more use cases and workloads are being pushed into AWS. And — the more Amazon begins to compete with other infrastructure and developer tools providers. Which brings me to my next takeaway…

Takeaway #2: The increasing depth and scope of the AWS feature set is concerning for many infrastructure / developer tools startups (and VCs like me investing in this space).

The fear is that in some cases, AWS tools are essentially “good enough” to compete against the more specialized infrastructure products. You can find examples of this is in application and server monitoring, where Amazon has released products such as CloudWatch, or in container management with Amazon ECS. At re:Invent this week, Amazon made a series of new product and feature announcements in key areas of AI / machine learning, IoT, and serverless computing that I believe could make real impacts in each of those markets.

However, I am personally not in the same camp as some investors who are choosing to ignore investing in infrastructure startups altogether because of competitive threats from the major cloud providers (AWS, GCP, and Azure included). In certain cases, there is a lot of tangible value in having a third party system that eliminates vendor lock-in. Additionally, it is unreasonable to assume that one vendor alone will be best in class at literally all layers of application infrastructure. That said, it is important as a player in this space to constantly be reminded of where the major cloud providers are making their investments and be cognizant of the “good enough” factor. My next three takeaways touch on some of the major areas of AWS investment that came out of the conference this year.

Takeaway #3: AWS is investing in AI / machine learning functionality, reducing the need for developers to build infrastructure to power their AI models.

Amazon released a series of new products (Rekognition, Polly, and Lex) that enable its customers to easily launch capabilities such as image recognition, text-to-speech, and speech-to-text. These features leverage Amazon’s expertise and technology used in both its e-commerce and Alexa products to essentially democratize certain AI functionality (using common machine learning and deep learning techniques) to all AWS users. This is good for Amazon customers and, in turn, great for Amazon, as machine learning and deep learning models require a massive amount of data and compute to be effective, which increases spend on AWS when using these products.

Takeaway #4: AWS is building out a robust IoT platform, competing more directly with IoT infrastructure companies and lowering barriers for IoT application companies.

Amazon released a series of new infrastructure products that enable IoT customers to easily build on top of AWS in both the cloud and at the edge. AWS Greengrass is a new software product that enables embedded compute (using Lambda) in connected devices. AWS Snowball is a new hardware + software product that provides edge computing at Petabyte-scale on a small portable device. Both of these products are ideal for many IoT use cases that cannot easily utilize the cloud due to a lack of available connectivity (no WiFi or cellular connection) or a need for action to be taken in real-time (without needing to send data to and from the cloud).

In my opinion, the push towards AWS enabling more edge capabilities in addition to its cloud capabilities is an important move and could drive a meaningful increase in AWS adoption by IoT companies. It may also be a positive for some IoT startups by having lower infrastructure barriers to developing IoT applications.

Takeaway #5: AWS is investing heavily in its serverless computing capabilities, which could have significant downstream impacts on the infrastructure ecosystem.

Software infrastructure has shifted from on-premise servers > VMs > containers, and now it may be heading into “serverless” architecture. Amazon is building an entire ecosystem around serverless, which is the concept of abstracting every aspect of infrastructure away from the developer. Of course the application is still being run on servers at a cloud provider. But in a serverless architecture, all the developer has to do is write the application code, and AWS Lambda (Amazon’s serverless execution engine) will handle all of the compute and associated operations that are required. Each time the application calls Lambda, AWS only charges the customer for the compute time required to service the function.

Serverless is a powerful concept in that it nearly eliminates the need for an operations team to support application infrastructure. If we imagine a world that runs on serverless architecture, there could be significant impacts to how companies think about building, deploying, and managing their applications. This is yet another step into a world where developers “set it and forget it,” enabling companies to focus on their own core competencies and eliminate the headaches of managing software infrastructure. Amazon has set the stage for a serverless world, and the other cloud providers are developing their own competing approaches.

Conclusions: Amazon is a company that seems to defy the innovator’s dilemma concept, and it does so by continually investing in the development of new products even after it has grown to become one of the largest companies in the world. The AWS re:Invent conference this year was an impressive display of some of the world’s top engineering talent and thinkers coming together to discuss what is next. It was a friendly reminder that “the only thing that is constant is change.” As a VC, new developments and innovations don’t scare me, and I came away energized with ways to take advantage of the new normal.