Microsoft and AWS are no strange bedfellows. The Seattle-based tech giants had earlier pooled their resources together to make Cortana and Alexa smarter. In August, Jeff Bezos led Amazon and Redmond giant agreed to give users access to Cortana and Alexa. Recently, the tech behemoths put aside their rivalry to roll out Gluon – “an open source deep learning interface which allows developers to more easily and quickly build machine learning models without compromising training performance.”
In what could be construed as a move leading to democratization of AI, AWS and Microsoft have rolled out Gluon, that attempts to reduce barriers that exist in the process of building and training neural networks, which require parsing huge volumes of data and can take weeks on end. Besides being a time-consuming process, training neural networks require developers to write long lines of complex code that is difficult to change and debug.
So, what’s the need for two tech titans to launch another Deep Learning framework when frameworks such as Google’s TensorFlow, Caffe2, Cognitive Toolkit and Apache MXNet already exist? What prompted the biggest cloud vendor Amazon to team up with its Redmond rival to build a AI ecosystem?
With Gluon, the deep learning framework market has become a little more crowded and competitive. On the other hand, this means good news for enterprises and mid-sized companies who want to incorporate AI capabilities in applications without making a steep investment in AI infrastructure. Another major factor is that by launching Gluon, the Seattle tech titans have made an attempt to wrest the platform dominance from TensorFlow as the go-to AI tool gaining ground amongst the developer community.
TensorFlow is widely popular for its speed and precision and open-sourcing it allows the research community and programmers to use an industry-grade software. However, the most vital point is that today Google has the first mover advantage to release an industry-grade AI platform and become an authority in the field of artificial intelligence. Reportedly, TensorFlow reduces the time spent optimizing the neural networks by 100 times.
Is TensorFlow the Android of the AI world?
If you want to become familiar with how neural networks and deep learning works, TensorFlow, Google’s machine learning software is the best place to start. Released in November 2015, TensorFlow scores high on popularity by being the most popular machine learning framework on code repository GitHub. One of the most popular projects on GitHub is DeepDream, a neural network that analyzes images and renders them a psychedelic effect. TensorFlow was Google’s way to dominate artificial intelligence and also drive customers on Google CloudML and so it has, the deep learning framework has gained a large developer following.
Here’s what we think – machine learning had always been a big part of Google’s strategy to catch up with its competitors in cloud market and by controlling the leading machine learning platform (TensorFlow), Google wanted developers to run their machine learning workloads in Google Compute Engine. By open sourcing TensorFlow, Google can effectively build new use cases that can add to the framework’s capability in later releases. Eventually, the Mountain View headquartered company hopes to monetize TensorFlow by offering its own cloud platform. Besides, Google wants to replicate its success with Android and wishes to make it the “Android of AI”.
Today, Google’s open-source software library is being used by the world’s biggest companies – DropBox, Uber, eBay, Airbnb. According to sources, since it was open-sourced, TensorFlow has gained in popularity and is now helping the search giant corner a bigger share of the $40 billion cloud infrastructure market, where it trails behind Seattle behemoths Amazon and Microsoft.
Outlook: How can AWS & Microsoft benefit from open sourcing their AI tool
AI has become a part of our lexicon and companies are looking for ways to incorporate artificial intelligence capabilities to speed up business processes. That’s exactly what AWS & Microsoft wish to achieve by open sourcing its deep learning framework – giving enterprises another choice.
- And here’s another catch – deep neural networks require massive computing power and this is where Cloud GPU instances are required for training session. That’s where AWS and Microsoft can monetize by offering the most reasonable and least expensive GPU-enabled alternative. While we will not delve into the costing part, AWS is believed to be the least expensive.
- By putting neural networks within the reach of every developer and seasoned data scientists, Microsoft & AWS are promoting an open AI ecosystem where developers have another option to choose from.
- As machine learning becomes more important for enterprises, whichever company provides the best framework will stand to gain a lot of business
- By open-sourcing the Deep Learning framework, AWS and Microsoft will also jointly tout their machine learning and AI capabilities. According to news report, a future release will add support for the Microsoft Cognitive Toolkit and other frameworks as well.
- As neural networks gain popularity and find a wide array of use cases in character recognition, fraud detection, stock market prediction, AWS & Microsoft want to become the go-to approachable framework to define and train neural networks.
- As AI seeps into everyday products and powers the most compelling technology, it has become all the more important to bring down barriers around AI and make deep learning more accessible to the developer community. It also a way of opening the Deep Learning platform to the world and giving a peek into how Microsoft & AWS are developing their AI systems, just as publicly as Google does.
- Not to forget, how companies are eyeing another revenue stream. Microsoft’s core market is from software licensing, Google’s revenues are driven from ads and AWS is the biggest cloud player, but that is being threatened by Google Cloud ML chipping away at its #1 spot. Even getting a small percentage of the cloud market can lead to billions of dollars.
- Both AWS & Microsoft would like to woo more enterprise customers relying on intensive AI tasks to add to their roster
- By open sourcing Deep Learning software, Microsoft & AWS are also targeting the researcher community involved in this work. Researchers who are already familiar with Gluon can also add to the job pipeline.
- Here’s another upside of outsourcing, businesses can use and adapt the framework according to their need. Every single contribution to the code or software can become part of future releases.
- Lastly, the Microsoft-AWS partnership is definitely an industry first to check Google’s rising ambition to dominate in AI.
