IoT at AWS re:invent 2018 – disappointments & discoveries
When I attended the latest re:invent in Las Vegas, I was primarily looking to get the latest news on IoT from AWS. In a nutshell, I was a little disappointed about IoT, but discovered plenty of news from areas related to IoT. Let me outline here my 3 disappointments and my 3 main discoveries.
Let me start with the disappointments, because this part is much shorter. My first disappointment was that neither Andy Jassy nor Dr. Werner Vogel talked about IoT in their big keynotes (3h/2h long). This was a real surprise for me and gave me the impression that IoT is currently not a big topic for AWS. The second disappointment from an IoT perspective was the big Expo hall, because I could only find one single player in this hall with a strong IoT focus – greetings to C3IoT. This gave me the impression that AWS currently has only a limited IoT partner ecosystem. The third disappointment was the fact that there were not so many IoT announcements at re:invent 2018, and that the newly announced IoT functions, such as AWS IoT Events (an event-processing tool), AWS IoT Things Graph (a low-code/rapid app dev tool), or AWS IoT SiteWise (edge SW to monitor machine KPIs) are only available in preview mode. I’m under the impression that the development of new IoT tools is progressing more slowly than developments in other areas of AWS’s business.
On the other hand, I discovered interesting new things at re:invent – not directly in the IoT space, but closely linked to IoT – around machine learning (ML), edge computing, and robotics. “We put machine learning in the hands of every developer, independent of their skill level”. This statement by Andy Jassy reflects AWS’s ambitions in the ML space, and they delivered on this at re:invent through multiple announcements. Let me point out just two of them: You can now find hundreds of machine-learning algorithms and models on the AWS Marketplace and deploy them directly on Amazon SageMaker. And you can start building out your machine-learning capabilities by buying a small, self-driving racing car ($400), training it with ML methods in a cloud-based, 3-D racing simulator, and finally taking part in a global, fun-based competition during 2019, the “AWS DeepRacer League”.
The second interesting discovery was in the edge computing space. AWS Snowball Edge Compute Optimized (available now) is built to run compute-intensive applications at the edge. You can also add an NVIDIA Tesla V100 GPU for scenarios such as full-motion video processing. This can also be used as an edge computing engine for use cases within connected/autonomous cars. AWS clients in this space are Toyota (one of the biggest car OEMs) and Denso (one of the world’s largest suppliers of auto components and software in Japan). In 2019, AWS will add a new chip – called AWS Inferentia – to this range to improve edge capabilities for ML workloads, but this chip may also quickly become an option for replacing the above-mentioned NVIDIA chip. AWS is not alone in this space, with Google, Microsoft, and Alibaba going in the same direction, providing their own ML chips and working with big automotive suppliers on connected/autonomous cars – this is a big market opportunity for integrated edge & cloud computing solutions. AWS is taking another step in the edge direction with AWS Outposts, a new AWS service that brings AWS servers to the edge and allows the fully integrated management of hybrid environments via AWS or VMware tools.
The third interesting discovery was about robotics. AWS is already quite active in warehouse robots, and rumor has it that they will soon move into smart home robots (controlled by the Alexa virtual assistant). AWS announced RoboMaker, a new service that utilizes the open source software Robot Operating System (ROS) and offers developers a place to develop and test robotics applications in the cloud. The demo of robot dog aibo from Sony, which is based on AWS services, was a funny early example in this context.
My conclusion: AWS builds and combines more and more services around all kinds of technologies (IoT, ML, robotics, edge, cloud) to provide broader building blocks for its developer community to build connected, automated, and increasingly autonomous vehicles and robots. I expect to see more from AWS in this space in the coming years – watch out!