On May 1 2018, I’ve joined Gartner Inc. – the worldwide leading analyst firm – as a member of the Managed Business & Technology Services team. In my role as a Research Director, I am covering Infrastructure Services & Digital Operations.
History Has Shown That Digitization Is More Evolution Than Revolution.
In November 2017, I met Christian Lorentz from NetApp during Cloud Expo in Frankfurt. Based on my experience, he wanted to find out if artificial intelligence (AI) is eating the data infrastructure.
Today, the artificial intelligence (AI) hype wouldn’t exist without cloud computing. Only the easy access to cloud-based innovative AI services (machine learning etc.) and the necessary and fast available computing power enable the developments of novel “intelligent” products, services and business models. At the same time, AI services ensure growth of public cloud providers like Amazon Web Services, Microsoft and Google. Thus, one can observe a “Cloud-AI interdependency”.
After more than 10 years, cloud computing has evolved into a fertile business for providers such as Amazon Web Services or Microsoft. However, competition is getting stronger from laggards like Google and Alibaba. And with the massive and ongoing introduction of AI-related cloud services, providers have increased the competitive pressure themselves, in order to raise attractiveness among their customers.
The Cloud Backs AI and Vice Versa
To build and operate powerful and highly-scalable AI systems is an expensive matter for companies of any size. Eventually, training algorithms and operating the corresponding analytics systems afterwards need oodles of computing power. Providing the necessary computing power in an accurate amount and on time via the own basement, server room or data center is impossible. Computing power that afterwards is not required anymore.
Looking into the spheres of Amazon, Microsoft or Google, all three providers built up an enormous amount of computing power in recent years and equally own a big stake of the 40 billion USD cloud computing industry. For all of them, expanding their portfolios with AI services is the next logical step in the cloud. On one side, developing AI applications respectively the intelligent enhancement of existing applications requires easy access to computing power, data, connectivity and additive platform services. Otherwise, it is necessary to obtain attractiveness among existing customers and to win new customers. Both are looking for accessible solutions to integrate AI into their applications and business models.
Amazon Web Services
Amazon Web Services (AWS) is not only the cloud pioneer and innovation leader, but still by far market leader of the worldwide public cloud market. Right now, AWS is the leading cloud environment for developing as well as deploying cloud and AI application, due to its scalability and comprehensive set of platform services. Among other announcements, AWS presented Amazon Cloud 9 (acquisition of Cloud9 IDE Inc. in July 2016) at the recent re:Invent summit. A cloud-based development environment that is directly integrated into AWS cloud platform to develop cloud-native applications. Moreover, AWS announced six machine learning as a service (MLaaS) services, including a video analysis service as well as a NLP service and a translation service. In addition, AWS offers MXNet, Lex, Rekognition and SageMaker, powerful services for the development of AI applications. SageMaker, in particular, attracts attention, since it helps to control the entire lifecycle of machine learning applications.
However, as with all cloud services, AWS pursues the lock-in approach with AI-related services as well. All AI services are tightly meshed with AWS’ environment to make sure that AWS remains the operating platform after the development of an AI solution.
Amazon also sticks to its yet successful strategy. After Amazon made the technologies behind its massive scalable ecommerce platform publicly available as a service via AWS, technologies behind Alexa, for example, has followed to help customers integrate own chatbots or voice assistants into their applications.
Microsoft has access to a broad customer base in the business environment. This along with a broad portfolio of cloud and AI services offer basically good preconditions to also establish oneself as a leading AI market player. Particularly because of the comprehensive offering of productivity and business process solutions, Microsoft could be high on the agenda of enterprise customer.
Microsoft sticks deep in the middle of digital ecosystems of companies worldwide with products like Windows, Office 365 or Dynamics 365. And that is exactly the point where the data exist respectively the dataflows happen that could be used to train machine learning algorithms and build neural networks. Microsoft Azure is the central hub where everything runs together and provides the necessary cloud-based AI services to execute a company’s AI strategy.
In the cloud, Google is still behind AWS and Microsoft. However, AI could become the game changer. Comparing today’s Google AI services portfolio with AWS and Microsoft you can see that Google is the clear laggard among the innovative provider of public cloud and AI services. This is astounding if you consider that Google invested USD 3.9 billion in AI so far. Compared to the competition, Amazon has invested USD 871 million and Microsoft only USD 690 million. Google simply lacks in consistent execution.
But! Google already has over 1 million AI user (mainly through the acquisition of data science community “Kaggle”) and owns a lot of AI know-how (among others due to the acquisition of “DeepMind”). Moreover, among developers Google is considered as the most powerful AI platform with the most advanced AI tools. Furthermore, TensorFlow is the leading AI engine and for developers the most important AI platform, which serves as the foundation of numerous AI projects. In addition, Google has developed its own Tensor Processing Units (TPUs) that are specifically adapted for the use with TensorFlow. Recently, Google announced Cloud AutoML, a MLaaS that addresses unexperienced machine learning developer, to help creating deep learning models.
And if you keep in mind where Google via Android OS has its fingers in the pie (e.g. Smartphones, home appliances, smart home or cars) the potential of AI services running on the Google Cloud Platform is clearly visible. The only downer is that Google is still only able to serve developers. The tie-breaking access to enterprise customers, something that Microsoft owns, is still missing.
AI Becomes the Game Changer in the Public Cloud
The AI platform and services market is still at an early stage. But in line with the increasing demand to serve their customers with intelligent products and services, companies are going to proceed to search for the necessary technologies and support. And it’s a fact that only the easy access to cloud-based AI services as well as the necessary and fast accessible computing power is imperative for developing novel “intelligent” products, services and business models. Hence, for enterprises it doesn’t make any sense to build in-house AI systems since it is nearly impossible to operate them in a performant and scalable way. Moreover, it is important not to underestimate the access to globally distributed devices and data that has to be analyzed. Only globally scalable and well-connected cloud platforms are able to achieve this.
For providers, AI could become the game changer in the public cloud. After AWS and Microsoft started leading the pack, Google wasn’t able to significantly play catch-up. However, Google’s AI portfolio could make a difference. TensorFlow, particularly and its popularity among developers could play into Google’s hands. But AWS and Microsoft are beware of it and act together against this. “Gluon” is an open source deep learning library both companies have developed together, which looks quite similar to TensorFlow. In addition, AWS and Microsoft provide a broad range of AI engines (frameworks) rather than just TensorFlow.
It is doubtful that AI services are enough for Google to catch up with AWS. But Microsoft could quickly feel the competition. For Microsoft it is crucial, how fast the provider is able to convince its enterprise customer of its AI services portfolio. And at the same time to convey how important other Microsoft products (e.g. Azure IoT) are and to consider them for the AI strategy. AWS is going to stick to its dual strategy and focus on developers as well as enterprise customers and will still lead the public cloud market. AWS will be the home for all those who solely do not want to harness TensorFlow – in particular cloud-native AI users. And not to forget the large customer base that is innovation oriented and is aware of the benefits of AI services.
The World of IT Automation.
In November 2017, I was interviewed by Judith Ellis from dotmagazine, which is powered by eco – Association of the Internet Industry. In this podcast we are talking about digitization, the significance of the cloud, the advantages of Artificial Intelligence (AI) in enterprise solutions and the power of digital correlation.
State-Of-The-Art IT Stack: Cloud Occupies Center Stage.
The Four Pillars of Artificial Intelligence.
The Four Pillars of Building a General AI.