Five years ago, the Working Group on Industrie 4.0 presented their report at the Hanover Messe Fair in Germany. This deceptively low-key event marked the start of the fourth industrial revolution. In the subsequent period, Industrie 4.0 or Industry 4.0 has gone from a governmental document for policy wonks to the default strategy of the global industrial sector.
As vendors and investors seek to capture a share of this growing market, we have witnessed some early failures and successes. GE’s decision to retreat from major parts of its digital business should serve as a stark reminder that we are in unchartered territories.
Within the Machine Learning based Predictive Maintenance category there has been a surge in venture capital-backed investments in startups. Given some of the optimistic analyst growth predictions, this is unsurprising.
As VCs continue to invest in Big Data and Machine Learning startups, some early investors are already cashing in. We have recently seen a handful of Machine Learning (ML) based Predictive Maintenance vendors have been acquired, including Precognize (Samson), Predikto (United Technologies), Mtell (Aspen Technology) and DataRPM (Progress).
In this article, we will evaluate what the impact of the acquisition of ML based Predictive Maintenance vendors on the industrial market.
Who is acquiring Machine Learning based Predictive Maintenance vendors?
To date, there are two categories of purchasers.
First, technology companies that wish to solidify their offerings in the growing Industrial IoT market view Predictive Maintenance as an entry point into large industrial producers.
Adding Predictive Maintenance to their portfolios is viewed as a competitive advantage.
Second, some Industrial OEMs are transitioning to a service model also referred to as Hardware as a Service or HaaS. These OEMs view digital transformation as an opportunity to lease their hardware to industrial plants and then receive ongoing recurring revenue by bundling maintenance services. At the core of this offering is the ability to analyze vast quantities of sensor generated Big Data in real time and extract operational insights.
OEMs that lack the expertise in Artificial Intelligence and Machine Learning are acquiring this competency by purchasing existing vendors.
How are industrial plants impacted by these transactions?
It depends on the acquirer type.
The first scenario is when a technology solution provider incorporates the Machine Learning solution into their own offering. The end customer of the Machine Learning solution may be forced to purchase the technology of the new parent company. There is also a risk that current customers that do not purchase technology from the new parent company will face disruption in their predictive maintenance activities. Either option would likely increase the overall cost of predictive maintenance for the industrial plant.
If an OEM purchases the solution, it could be incorporated into a piece of hardware and then only be available to the OEM’s customers.
Can industrial plants build their own solutions if further acquisitions occur?
There are almost no industrial plants with deep competencies in Machine Learning and the ability to develop and support an internally-built Predictive Maintenance solution. To release a software-based solution requires hiring software engineers with experience with the Software Development Lifecycle (SDLC). This is the easy part. Building a Machine Learning based application requires domain expertise in areas of rapid innovation that is increasingly hard to find in today’s tight labor market.
We recently wrote a blog article titled “Is It Better to Build or Buy Machine Learning Predictive Maintenance Solutions?” that examines this issue thoroughly.
Will Google and Amazon level the Predictive Maintenance playing field?
My colleague, Deddy Lavid (co-founder and CTO of Presenso) recently published an article in the Inside Big Data Blog on this topic. From his understanding of the Machine Learning marketplace, tech giants such as Google and Amazon are only providing ad-hoc ML tools that do not scale easily. They are not providing Predictive Maintenance solutions per se.
It is possible that a large technology company may develop an end-to-end Predictive Maintenance solution in the future, but this not yet the case.
Industry 4.0 is considered a revolution and is no different from earlier revolutions: new elites emerge at the expense of existing power centers.
As existing players use their accumulated resources to shrewdly ramp up their Machine Learning capabilities and acquire new market entrants, industrial plants will have fewer options for Machine Learning based Predictive Maintenance.