Don't be all things to everyone - A lesson learned from GE in Industrial IoT
Nikunj Mehta
AI-Driven Decision Intelligence Expert | Protecting & Optimizing operations | 30+ international patents
(edit: Updated the list of layers to include application enablement)
Stacey Higginbotham has covered the Industrial IoT market challenges quite well in her thoughtful report. She is spot on that comprehensive Industrial IoT platform plays are doomed. Our own perspective at Falkonry is that Industrial IoT is going to be built as a stack, just like an IoT stack described by our friends at PubNub.
There are layers in the stack that can be horizontally designed and are applicable to many industries. For example:
- Industrial networking, e.g., Kepware Technologies, Softing
- Data historian, e.g., OSIsoft PI, Inductive Automation
- Dashboarding, e.g., Tableau Software, Software AG
- Predictive Analytics, e.g., Falkonry
- Workflow management, e.g., SAP EAM
- Application enablement, e.g., OSIsoft PI, PTC Thingworx
In each of these layers, a core design philosophy to empower industrial practitioners will lead to success. Also, an intention to limit focus to a core function is essential and enables customers to make the right choices for their needs in the various layers. The range of problems in the Industrial IoT is so large that unless a function with the narrowest (API) footprint is targeted, a company can easily get distracted and go down a rat's hole.
My VC 'friends' admonish us for having built a horizontal technology but because we concentrate on a repeatable need that is narrow in scope and high in value, we are able to coexist with giants who have many concerns and priorities across their platform offering. That repeatable need is predictive modeling on time series data and it needs to scale across domains and problems, which means that it should work directly for industrial practitioners without the need for data scientists. Many of our customers have taken long term subscriptions and will be great learning experiences not just for us but the entire Industrial IoT market.
I can only say that if you are going horizontal, then focus on as narrow a slice of the stack as you can serve well and hope that your slice was fat enough for you to be hugely profitable. If Falkonry's core thesis is right, and time will tell quite soon, then there is a huge payoff for such a horizontal technology. If not, there is very little money or customer's time wasted.
As David Packard learned early on, what kills a good start up is not starvation but indigestion. Turns out this applies to not just startups but even established companies such as General Electric (GE).
Digital Strategy, Transformation, and Enterprise Architecture
7 年Please come down to Houston, I'd like to have a chat with you.
Chief Product Officer at IoT Squared
7 年Being at the core of the General Electric (GE) #DigitalTransformation leading at GE O&G there are many lessons learnt, appropriate criticisms, and invaluable insights. GE's stumbles on the Predix platform could have one underestimate the impact of the Digital DNA now engrained throughout GE, its businesses and its expanded ecosystem. The Predix platform was just one technical part in GE's overall mission.
Hi Mike, I was always taught to listen first - that enables the understanding that can be developed into (usually) one of the solutions. However, 3-D printing is the answer; now what was the question .......... ;-) (Only joking)
Sometimes it helps to go back to basics: a user buys a car as a transportation solution (at least today!). The car incorporates thousands of components (technology enablers) delivered to the assembler as 'plugable' sub-systems. Work out if you're selling a car, a sub-system or a component and that will define your target market. Try to do all three and you clearly need to be bigger that GE.
The platform play is intensely difficult as pointed out in the previous posts. It was very clear regarding GE when the ads talked to the possibilities, for example saving 1% of locomotive fuel costs, with no comment on what the platform could actually do to achieve that. I contrast, Mtell brought a laser focus on "stopping machines breaking" - a singular massive problem costing up to 10% of the world-wide manufacturing market. The IIoT platform is worth little without the analytics that produces the value. The platform does not create the value. It is an enabler, not a real value creater...and in many cases, do you actually need that comprehensive platform to reach the value? I have said many times what's the problem to be solved? Do not start with the platform and expect to find problems...which was GE's tact. That's the answer; what's the question?