Artificial intelligence is invading every part of the cloud experience for organizations. There are many vendors in the space out to provide the most robust, well-integrated, secure, and usable cloud-based AI services. And the four biggest cloud vendors are revving their engines in this race to see which will be the biggest AI winner.
Microsoft has just wrapped up its 2021 Build conference, where a key focus was Azure’s AI capabilities. Azure has introduced new applied AI services built to make AI more operational and make it easy for companies to adopt AI. This combines Azure’s cognitive services, task-specific AI, and business logic. It is meant to allow businesses to leverage AI even if they don’t have the best machine learning and AI chops. Some featured use cases are chatbots, document digitization, video analysis, and more.
Microsoft is almost entirely an enterprise company today. With its consumer businesses on the decline, the part of Microsoft that is making up for the loss is its Azure section. And enterprises that have been with Microsoft for decades have become conditioned to expect point-and-click, drag-and-drop wizard user experiences that are foolproof. Microsoft is indeed the least daunting of the top cloud vendors. Its platform is built to be simple to operate, and there is good support from the Azure team. It’s no wonder Azure is No. 2 after AWS, and in the enterprise segment, it may even have a lead over AWS.
Azure wants to keep up its record of bringing great technology to the masses. In this case, AI to the masses. This wouldn’t mean expecting customers to build complex ML models with TensorFlow and invest heavily in a huge data lake that is distributed. It means to store all data in Cosmos DB, point Azure’s AI services to it and get AI-based output in a matter of minutes. It skips all the in-between steps of preparing data, optimizing ML models, creating dashboards to view results, and more. This would enable organizations to operate at greater speed and be more agile with their AI efforts.
If you’re a Microsoft shop, it makes sense to go with Azure AI. However, if you are not heavily invested in Azure, you still have good reason to consider it for its ease of use, support, and integration with other cloud services from Azure.
IBM Watson and Cloud Pak
IBM has two components to their AI strategy — Watson and IBM Cloud. Watson needs no introduction. As one of the earliest AI projects, Watson has made it to headlines more than any AI product. IBM has invested in Watson heavily over the years, even acquiring many AI startups to improve Watson’s capabilities. In fact, acquisitions have been a key strategy of IBM. IBM acquired Red Hat in a landmark deal of $34 billion to bolster IBM Cloud. IBM combines the power of these two platforms to deliver an extensive list of AI services.
IBM Studio is where customers go to build ML models. IBM has taken pains to create a very visual experience to building models in Watson. It’s complete with flows, connectors, and a very easy drag-and-drop feature. There is extensive support for open-source ML frameworks like TensorFlow, scikit-learn, and more, if you’d rather build models with one of those tools than with Watson.
IBM has allowed one-click deployments to the IBM cloud if you choose. It is critical that IBM makes this effortless as they compete with three other cloud vendors with a bigger market share. IBM Cloud Pak for data is an AI management platform that lets you deploy, manage, and scale AI models on IBM Cloud or other cloud services. IBM takes an open approach here, making sure to mention that they are built on RedHat OpenShift and encourage open, hybrid, multicloud setups.
While IBM focuses on a single product (Watson) doing many AI jobs well, AWS has a different approach where they have created numerous AI services, each with a specific focus. There’s CodeGuru for code reviews, Lex to build chatbots, Rekognition for image and video analysis, Polly for text to speech, Translate for, well, translation, and you get the idea. Each of these services is laser-focused and fine-tuned for the exact purpose they were built. Being a very developer-centric platform, AWS has organized its AI services in a way that is easy to integrate into applications being built and hosted on the AWS cloud. The list of services is exhaustive but well-organized and easy to apply. AWS is able to price the different services accordingly, so you can easily correlate what you get with what you pay for.
AWS’s claim to fame with AI is their extensive use of it for Amazon.com and with their Alexa smart assistant devices. This is a strong footing from which AWS can deliver well-optimized and scalable AI services that are battle-tested. Though developer-centric, AWS’s AI services also cater to organizations with little or no machine learning talent in-house.
The biggest benefit of opting for AWS AI is its vast universe of related cloud services. Being the leader in the cloud, AWS has the widest range of services. If you have very mature AI cloud operations and need full control end-to-end, there likely isn’t a better place than AWS. Its services are well-integrated, and you can’t really go wrong. If there is a downside, it’s just that the sheer vastness of services can be overwhelming and daunting for some.
Google Cloud AI
Google, as an organization, has a rich history of being data-driven. The Google search engine itself is one of the earliest implementations of data science and AI. Further, Google’s work on large-scale data led to the rise of Hadoop and the big data ecosystem. Finally, Google’s prowess with Kubernetes has spawned a new revolution with container technology. Google was late to the cloud party, but it is doing all it can to beat AWS and Azure. Google Cloud isn’t as enterprise-friendly as Azure, nor is it as well-integrated as the AWS-verse. What Google excels at is simple user interfaces, extremely functional products, and unmatched data prowess. The last one really matters with AI.
Google has created Vertex AI as the hub of its AI services. This is where customers can build and deploy ML models. While Vertex is the focus for building custom ML models, Google Cloud also offers solutions for specific use cases. Contact Center AI offers services aimed at customer support. This includes text to speech, speech to text, and more. Document AI handles the analysis of documents and forms.
You can run your AI models on a Google range of instances such as CPUs, GPUs, and TPUs. The latter is especially interesting as it is purpose-built for AI. Tensor Processing Units (TPUs) are especially good at processing large batches of data and gleaning predictive insights from it.
Cloud AI services: All’s good
Technology has come a long way, and AI is all set to take it further than we’ve ever imagined. As organizations build AI solutions of tomorrow, they need reliable infrastructure and AI products and services in the cloud to run them on. The four cloud vendors mentioned here are innovating in their own way with little to separate one from the other. You can’t go wrong choosing one of them. With the cloud AI services taken care of, the only thing left to be seen is how organizations use these platforms and what they build on top of them.
Featured image: Shutterstock / TechGenix photo illustration