Table of Contents
The Machine Learning as a Service (MLaaS) space is changing at a dizzying pace, and as a result, the future outlook for the platform is as important as its current state. This Gigaom Sector RoadMap describes the market, comparing solutions by identifying several Disruption Vectors that will drive the space in the coming 12 to 24 months.
Five Disruption Vectors were chosen that highlight the suitability of each vendor to both the current and future of MLaaS platforms. Tech buyers can use the Disruption Vector analysis to assist in selecting products that best match their own situation.
Key findings of this analysis include the following points.
- A breakdown of the various aspects of machine learning (ML), and how each vendor scores in the disparate categories for image recognition and analysis, voice and speech analysis, and voice response shows that many vendors differentiate themselves with not only the number of raw services, but the capability of each service.
- It is important to compare the different vendor clouds on which these MLaaS services run, especially with the perspective of how that platform affects MLaaS workloads and costs, taking into account raw number of services.
- Developer productivity should be a paramount concern when deciding upon a vendor. Ancillary concerns like network speed, streaming capabilities, analytics, and other factors can dramatically impact developer productivity as well as end-user satisfaction. Additionally, many vendors like IBM and Microsoft have started to bundle their MLaaS services into solutions for particular use cases or verticals, reducing the time-to-market for apps that fit into those boxes.
- As a part of developer productivity, support may also be a concern, especially if machine learning is a discipline new to your organization. Most vendors roll their ML support into a general “cloud support” contract, but Google has created a dedicated services team specifically for ML.
- Google, Microsoft, and IBM are in a dead heat for top provider. AWS is showing up as 2nd in the MLaaS arena, with their newly announced ML-specific services. It’s still too early to determine who will emerge as the clear leaders in the future. Although significant differences exist between the vendors, as features and capabilities coalesce around a standardized feature list, vendor offerings will start to closely resemble each other.
- Machine Learning is a broad umbrella under which many individual projects and use cases are grouped. Therefore, at the end of each vendor’s section, we call out several specific use cases and rate the individual platforms as useful for particular use cases.
- Number indicates company’s relative strength across all vectors
- Size of ball indicates company’s relative strength along individual vector