Sergey Nivens - Fotolia

Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Machine learning platforms comparison: Amazon, Azure, Google, IBM

The platform war over machine learning tools is heating up. Use our features comparison chart to see how four top vendors stack up and help you decide which is right for your enterprise.

Data scientists who want to build machine learning models and put them into production have no shortage of available tools, but choosing the right one comes with some thorny decisions.

The chart below breaks down some of the most popular machine learning platforms by their key features and price tags. Note that many open source tools are available for machine learning, as well as other vendor offerings, but we focused exclusively on vendor cloud platforms that span the entire machine learning lifecycle from data ingestion to model development to production.

The market for machine learning platforms is heating up, and all of the leading vendors are looking to nab their share. Analyst firm Forrester expects this market to grow at a rate of 15% annually through 2021.

Several vendors have beefed up their offerings in recent months and now offer simple, cloud-based platforms for getting started with machine learning and developing models that can quickly be put into production.

But these machine learning platforms all come with their own downsides. There is a significant risk of vendor lock-in with each. They generally require users to bring their data to the broader cloud platform. Once all of an enterprise's data is located in one vendor's cloud, it's difficult for that business to use another vendor's services or to use open source tools for other tasks.

Machine learning vendor comparison chart
This chart compares top machine learning vendors on key points.

Since it's still early days for these types of platforms, it's hard to know which vendors will ultimately develop the best offerings. Users need to understand the various pros and cons of each before deciding to hop in with all their data. The machine learning platform war is on -- use the comparison chart to help your enterprise navigate the battlefield. Click on the diagonal arrows in the top-right corner to view the full chart.

Next Steps

Machine learning platform helped investors limit Brexit fallout

Cloud machine learning has yet to hit its stride

Ready-to-use algorithms make chatbot development easier

This was last published in June 2017

Dig Deeper on Advanced analytics software

PRO+

Content

Find more PRO+ content and other member only offers, here.

Join the conversation

8 comments

Send me notifications when other members comment.

By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

Please create a username to comment.

What factors do you think are most important in selecting a machine learning platform?
Cancel
When I had to do this 3 years ago it was all just kicking off in EMEA ie low cost, easier access via these players. I started with MS ML probably off a Google search. It was easy to use but then I moved to Amazon as I was more used to their database offer via AWS. Then went on to use SAS decision trees as my company then had a licence and data team were familiar. Nowadays its 1. ease of use and fast algo training and scoring 2. production ready ie actioning those predictions 3. real time scalable - hardest to find IMO. Its not as hard as people claim so crack on ;)
Cancel
Very good summary article and reflects my direct experience with Amazon and MS ML plus what I've read/heard from other experts re ML on Google and with IBM.
Cancel
Do not get caught up in the hype. This stuff has been around since before Nasa wrote the Rete Java libraries in the mid-90s, and it's not substantially improved since then.
Cancel
I can't stand for the other vendors but the IBM opinion is not based on the actual offering. IBM does model selection for example, you only took the service available on Bluemix (deployment) and not the actual IBM ML suite.
Cancel
Cancel
This is awesome. We are using Amazon ML and Google Cloud ML for training our data. Building smart web applications will now be easier for everyone.
Cancel
Hi, I am from Good AI Lab and we have created a new cloud platform exclusively for TensorFlow. Its called TensorPort (please see TensorPort.com.) We use git and git-lfs to load models and data so theres no need for ssh, sftp or reloading the the same versions multiple times. You can just push as many versions of your model as you want and run the single command 'tport create project', then enter our browser based interface to run, monitor, and view side by comparisons of jobs. We have a growing community and I hope to see you there. 
Cancel

-ADS BY GOOGLE

SearchDataManagement

SearchAWS

SearchContentManagement

SearchCRM

SearchOracle

SearchSAP

SearchSQLServer

SearchSalesforce

Close