March 01, 2018

Public Cloud Platforms - A.I. and Machine Learning

By Brian Cooke
A.I. and machine-learning applications are gradually being integrated in U.S. public cloud applications, with demand expected to rise during the next 12 months.

U.S. Demand Outpacing Europe
Although artificial intelligence/machine learning received numerous mentions as an up-and-coming business catalyst in recent OTR research, a closer look at adoption found actual demand for such functionality varied widely among cloud service partners’ customers. European partners reported little demand while U.S. partners’ responses varied based on their market focus and whether their business model relies more on custom development versus lift-and-shift projects. However, U.S. partners typically said most current and expected A.I. and machine-learning applications are part of much larger public or hybrid cloud projects. One U.S. partner said, “We’re using A.I. as part of custom applications: It’s integrated into other cloud applications. More and more larger-solution architectures will include A.I. components. Some of these are greenfield solutions, but many -- if not most -- of our planned A.I. implementations are part of larger cloud solutions.” In contrast, one German partner said, “A.l. looks promising but … currently everything is on a hypothetical level.”

U.S. Growth Expected in Next 12 Months
Overall, U.S. partners expect consulting projects that include A.I. and machine-learning components to account for almost 10% of their 1Q18 cloud revenues on average. In 3Q18, this figure is expected to exceed 15%, while the expected average for 1Q19 was about 30%. For each quarter, sources whose companies focus on custom development and A.I. expertise gave higher numbers. One source at a smaller, more specialized cloud consulting company expects the percentage to hit at least 40% in 1Q19, while another source at a larger, enterprise-focused company expects less than 10% of business to fit this description by 1Q19. “We’re not targeting the cutting-edge companies,” the source said.

Despite the growing interest in A.I. applications reported in OTR Global's January Public Cloud Platforms report, large A.I.-centric projects could be several years off, partners said. According to one, “This is a five-year -- maybe a 10-year journey. For example, voice recognition for A.I. is like optical character recognition on steroids -- it gets really compelling, really quick. People aren’t always aware that it’s not that easy to develop this stuff, but there’s incredible opportunity.”

A.I. Vendor Demand Expected to Depend on Cloud Platform Preferences
Partners said their customers’ A.I. vendor preferences tend to match customers’ cloud vendor preferences. One said, “For selecting an A.I. platform, our customers tend to gravitate to whatever vendor they’re using for public or hybrid cloud. It helps that [Microsoft Corp.'s] Azure and [ Inc.'s] AWS both have robust A.I. solutions. [Alphabet Inc.'s] Google is coming on fast as well.” Another said, “AWS marginally leads the pack in front of Azure in terms of capability and maturity. The platforms are pretty similar in a lot of ways.”

Smart-Home Applications Seen as Promising Long-Term A.I. Use Case
Partners cited a variety of promising A.I. business solutions, including smart-home applications that can be integrated with products such as Amazon Echo, Google Home and Apple Inc.’s HomePod. One partner said, “Most of the applications I’ve seen that are achievable and demonstrable have been home automation. Voice-controlled devices are becoming more prevalent, which makes it easier to picture those applications. It also doesn’t take a whole lot to impress people in that space.” However, other partners said smart-home solutions could take some time to mature. One said, “Smart home’s a hard market because you have to fit into existing houses’ infrastructure. We may have to leave that to the big guys like Belkin [International Inc.] and Intel [Corp.] to figure out.”

Operational Efficiency Applications May Catch on Faster
While many sources named smart-home applications first as a promising application, several said their A.I. business so far has been dominated by operational efficiency/automation, risk assessment and security/fraud detection applications -- what one partner called “mundane solutions.” One partner said, “Right now most of our A.I. business is internal-facing, but I think that’s because it’s new, developing technology. Once customers are comfortable with using A.I. internally, I think you’ll see more companies start offering A.I. services to their customers.” Another said, “We see the biggest opportunity as finding ways to improve workload efficiency: things like evaluating patterns and automating fixes. The goal is a self-healing system that can fix anomalies and trigger-type events.”

Healthcare and financial services were named most often as driving A.I. demand, followed by manufacturing, insurance and transportation/logistics. One partner said, “Any industry that’s got a lot of data is likely to be interested in A.I. -- healthcare, financial services and insurance all fit that description.” Another said A.I. demand extends across a variety of industries. “The consumer goods sector is interested in basic A.I. to reduce overhead for support and post-sale costs," the source said. "Healthcare is also doing some fascinating stuff, especially in conjunction with robotic surgery.”

Customer Service A.I. Solutions Also Seen as Promising
Partners also expect A.I. customer service applications to catch on, with several mentioning A.I.-powered chatboxes as a way to reduce reliance on human call centers. One partner said, “Chatboxes are a pretty easy sell. If you can cut out 50% of your call center calls because you’re using an A.I. chatbox to answer questions like ‘what is my password,’ you could be talking about millions of dollars in savings.” As a result, partners said call centers, managed service providers and other customer service providers could likely lose business as a result of A.I. trends. One partner said, “Any business that's focused on throwing bodies at problems could lose business as a result of A.I."

Contributors: Suzanna Kerridge and Hartmut Leuschner

What is an OTR Snapshot?

OTR Snapshots are direct feeds from our editors expressing their views on key issues and events in their industries, within and beyond our core coverage. Snapshots feature industry insights, source commentary, and general observations.