Which is the strength of Google and Microsoft‘s artificial intelligence?

Both Microsoft and Google‘s parent company, Alphabet, play an important role in the frontier software world of machine learning. But Valley Singer holds the current machine learning framework TensorFlow, which gives it an edge in the competition with Microsoft.

The stock movements of Microsoft and Alphabet are strikingly similar: this year‘s declines are exactly the same, growth is similar, and market capitalization and valuation are similar.

Which one would you choose if you choose one?

Both companies are outstanding companies, and any one is a reliable choice. Ignore the difference in revenue between the two – for example, one from operating system software and one from advertising – you will find that they all dominate the software ecosystem. In this regard, both Microsoft and Alphabet have a strong cloud computing business, both of which are the leaders in machine learning, the most important computing software of today.

However, the Google division of Alphabet can say that it has the upper hand in Microsoft in several ways.

In September, Microsoft‘s market capitalization was $751 billion, surpassing Google‘s $707 billion, with an annual increase of 13%, while Alphabet fell by about 4%. The two companies have similar P/E ratios, and Microsoft (19x) is a bit cheaper than Alphabet (22x). Obviously, these two stocks are quite expensive compared to the overall situation of the market. Google’s revenue growth is relatively fast, with an expected growth of 19% in 2019, while Microsoft’s 2020 fiscal year revenue growth is 11%.

In the final analysis, the difference between the two companies lies in the development of cloud business. Microsoft has a size, but its position in the community may be relatively less important.

Machine learning is one of the most important trends in today‘s programming. Through machine learning, a computer can automatically form some code by extracting patterns from the data.

Both companies and Amazon have deployed machine learning tools in their respective cloud computing businesses. In terms of revenue scale, Microsoft‘s Azure cloud service is generally regarded as the second only to Amazon‘s AWS, and Google‘s Google Cloud, which ranks third, is still far behind. However, since neither Microsoft nor Google specifically disclose the revenue of cloud services, the comparison between the two is not rigorous.

However, it is safe to say that their cloud business is an important asset for both companies to attract and retain machine learning neural network developers.

However, Google has control over the most important "framework" TensorFlow for building machine learning applications. A framework is a set of pre-built code that software developers can use to piece together a program without having to write code from scratch. Currently, TensorFlow is the mainstream framework for machine learning.

It is not a machine learning framework. In fact, Microsoft also has some important products of its own. They include infer. The .NET Framework and the framework known as the Microsoft Cognitive Toolkit. Both of these frameworks are used by developers to launch many interesting projects.

However, much of the work on machine learning is done through TensorFlow; or through other frameworks developed by companies outside of Microsoft, such as "PyTorch" from social networking giant Facebook; originally developed at the University of California, Berkeley The campus‘s framework "Caffe"; and new projects developed based on these programs, such as the San Francisco startup fast. Ai‘s frame of the same name.

Microsoft’s $7.5 billion acquisition of the community’s GitHub deal in October did not change the situation. The success or failure of machine learning frameworks depends on their strengths and weaknesses and whether they are widely supported. Even with a programming library, GitHub doesn‘t change the developer‘s choice in the machine learning framework.

The control of the framework is crucial, and Google has always been based on the use of TensorFlow. They use self-developed custom chips in cloud business operations. Its custom chip, called TensorProcessingUnits ("TPU"), has become a cutting-edge silicon chip that Google offers in its cloud computing services. The combination of TPU and TensorFlow software means that Google has a combination of chips and software, which is an advantage over Microsoft today.

TensorFlow has rapidly expanded into the field of mobile computing. In this area, Google dominates the world with its Android operating system, and Microsoft has no sense of existence. In November last year, a study released by researchers from Peking University and Microsoft‘s research institute in Beijing showed that half of the most popular mobile applications on the Android platform use TensorFlow, "TFLite" (Google specifically A small subset of TensorFlow created by the mobile device) or "ncnn" (a framework created by Tencent). On the mobile machine learning application, Microsoft‘s code can be said to have no sense of existence.

The entire machine learning field is still a very new field and is still in its infancy. In the future, there will be a lot of innovations, and Google‘s advantages will shrink and even disappear over time. An interesting move by Microsoft is that it supports something called OpenNeural Network Exchange (ONNX), which over time may reduce developers‘ reliance on a single framework.

Time will tell us the answer, while at the same time, Alphabet‘s Google is at Microsoft for core tools for application development types. So the company‘s stock is worth betting.