From: acquiredfm
The co-location of data and machine learning (ML) processing capabilities is a critical factor in the adoption and effectiveness of ML services. Running machine learning near the data it processes is considered the most important aspect of an ML strategy [00:01:08].
Amazon’s Strategy in Cloud Storage and Machine Learning
Amazon’s business is built on key pillars, with compute being a massive component and databases offering significant stickiness [00:00:06]. Amazon’s investment in machine learning has grown, evidenced by the announcement of SageMaker, dedicated ML keynotes, and a variety of cloud-hosted ML offerings [00:00:18]. While Amazon Web Services (AWS) provides support for technologies like TensorFlow (originally created by Google), they also offer their own container service (Amazon Container Service, ECS) and elastic Kubernetes service [00:00:26].
A key insight into Amazon’s strategy is that their machine learning capabilities, while needing to be good, do not necessarily need to surpass those of competitors like Google [00:00:53]. The core of this strategy is that customers will typically utilize the ML services available where their data is already located [00:01:01].
Running machine learning near your data is the most important thing [00:01:08].
The Role of Data Storage in ML Adoption
Once an organization selects a vendor like Amazon for data storage and migrates their data into the vendor’s data centers, they are unlikely to then shop around for different ML providers [00:01:12]. Instead, they will run their machine learning workloads on AWS [00:01:23].
This dynamic means that even if Google might be better positioned to offer superior ML capabilities, it becomes less relevant if customers are not storing their data with Google [00:01:31]. Therefore, for cloud providers, ensuring competitive, though not necessarily leading, ML offerings coupled with robust data storage solutions, is a powerful strategy for customer retention and service adoption. This highlights a significant aspect of the comparison of Amazon and Google’s ML capabilities in the cloud market.