The IBM IoT story, starring AI, Watson and an industrial cast

Playing nice in the IBM IoT ecosystem

Bret Greenstein

While early IoT players built some of their rules engines and analytics on specific devices, in today's services-oriented IBM, everyone plays with everyone else. "Exactly as the internet evolved, none of the hardware is tied [anymore] to any of the middleware," Greenstein said. "The pattern in IoT is the same as the pattern in cloud; we're containerizing everything. As long as you have an OS that supports containerization, like Linux, Windows and QNX, we can basically package up our workloads and run them on a lot of places."

Through Node-Red, the IBM-developed, flow-based IoT development tool set, and its set of open RESTful APIs, Watson IoT can integrate with just about any hardware or any cloud, reaching any device or any database. "It becomes a very fluid ecosystem, in a good way," Greenstein added.

Today, the Watson IoT platform is available in public, private or on-premises deployment options in 175 countries. It occupies a "Leader" spot in IDC's MarketScape Worldwide IoT Platforms Vendor Assessment, along with GE Digital, Microsoft, AWS and PTC/ThingWorx. It also made Gartner's Magic Quadrant for industrial IoT platforms in May 2018. In November 2016, the company announced the launch of its Watson IoT Consulting Solutions practice, headquartered in a new $200 million facility in Munich, alongside eight other IBM IoT centers across Asia, Europe and the Americas.

Visual and auditory inspection

IBM also trades on its relatively long history in machine learning in the areas of visual and auditory perception. "IoT data is such a mess," Greenstein said, "filled with all kinds of data types that are not just numerical and well structured, but can include images, sound, video and vibration." Watson IoT can even be trained on the sounds of healthy and failing equipment, learning to interpret machine hums and noises like an experienced operating engineer. A more familiar application is visual inspection, using Watson machine vision to train on defects of different types.

Maribel Lopez

Maribel Lopez, founder and principal analyst at Lopez Research, noted that insurance companies have taken this use case airborne, using drone-mounted cameras to visually inspect rooftops for damage. Indeed, one of the offerings on Watson's extensive cloud catalog is IBM IoT for Insurance, which analyzes sensor and other data to prevent accidents before they happen, deter fraudulent claims and tailor policies to risk profiles. Other verticalized versions of its IBM IoT service include Watson IoT Automotive, Electronics, Manufacturing and Retail.

Gartner's Magic Quadrant reports "observed and verifiable industrial use cases" of the Watson IoT platform in "asset monitoring and predictive maintenance of in-field industrial assets, such as heavy industrial equipment, transportation and renewable energy generation."

Field service, facilities management, asset management

Predictive maintenance is high on the list of many IoT platform providers' use cases. So is manufacturing. IBM also has injected IoT into its long-standing Maximo asset and facilities management software. It's active in the automotive vertical, both with self-navigating cars and natural-language human interfaces. IBM is also working with such consumer manufacturing companies as Panasonic, Nokia and Ricoh Co. to optimize their product design, overlay voice controls and develop sensor-laden wearables for healthcare and worker safety.

Eric Goodness

"We're seeing IBM doing really well in IoT as it feeds into field services and facilities, building management systems," said Gartner Vice President Eric Goodness. "IBM is the No. 1 vendor in asset management and there's a natural strong integration between that and IoT."

Security is also one of IBM Watson's chosen "spheres of influence." IDC gives IBM special marks for its "multilayered security strategy," citing "security built into the architecture, security analytics dashboards, service for blockchain, threat intelligence and data anonymization."

IBM IoT tackles concerns over data ownership

Along with network security, another area of concern in IoT deployments is data ownership. Lopez explained, "People have said, 'I don't want Watson to train on all my data and then have Watson be able to sell this service to everyone else.'"

IBM's spokespeople address this concern upfront. While the IoT ecosystem is now very fluid in terms of running platform and hardware, it closely ties data to apps, Greenstein said. "Retail customers are not going to be turning their data over to Amazon, for example. Automakers can't turn over their IoT data and their car data to Apple and Google and Amazon because they're creating potential future competitors," he said. "Also, brands that we work with establish trust with their end users, and they can't break that by letting their data flow to a third party."

Greenstein admitted that this is a disadvantage to IoT development, because machine learning benefits from large and varied data sets. But IBM makes it clear that clients own their own data. "IBM relies on public and purchased data sources to get partway to a model. Our vision system out of the box can recognize dogs and cats and license plates, for example," Greenstein said. "Then we train it for the specifics of a specific client."

Professional services lead Watson -- and other IoT deployments

Gartner's Magic Quadrant report included some challenges reported by Watson customers, among them coordination between the innovating software teams and the deploying services arm, and between IoT-specific and non-IoT-specific cloud teams. Goodness also noted that IBM is "fairly rapidly" losing its early lead in market share growth.

"There's so much competition in this space," he said. "We're literally tracking hundreds of competitors that are offering IoT as a middleware stack or IoT-specific analytics. Then there are those organizations that come at it from the edge, from a pure centralized cloud platform perspective, and those that have a legacy installed base in manufacturing and utilities and other industrial enterprises that come at IoT from a legacy SCADA (supervisory control and data acquisition) or industrial automation and control perspective."

Goodness also said that the IBM IoT strategy has been slower to innovate in features than smaller companies. "IBM is being challenged, not only from an industrial perspective by the GEs, the Siemens, the Schneiders and ABBs, but there's a host of new, smaller providers that reside on AWS or Azure. And then you have AWS and Azure themselves."

Intense competition for IoT customers is mirrored by a great run on engineering and integration talent. Asked who they used to implement their IoT platform, he said, only 4% of 132 respondents in a recent Gartner survey went outside the platform developer and their chosen systems integration partner. According to Goodness, that shows that end users are putting a lot of their trust and faith in the platform developer to understand how to integrate the edge to the platform, and different IoT platforms together across large multinationals. "It's also a reflection of how immature the systems integrator pool of resources is when it comes to IoT," he said. "There just aren't a lot of [systems integrators] that have a lot of past performance to put on their resume."

In May 2018, IBM announced that it would be hiring 1,800 employees in France over the next two years for work in data science, cloud computing, IoT, AI, cognitive business, cybersecurity and blockchain technologies. Just as it enjoys an edge in its prior relationships with so much of global industry, IBM may have an advantage in its worldwide hiring ability.