Two years ago GE made some big bets on the internet of things — what it calls the industrial internet. It opened an office in San Ramon, Calif. and started articulating a vision of connected sensors sending data to either a box on-premise or a cloud. Analyzing that data through GE’s algorithms would help GE and its customers determine maintenance schedules, when to operate at times that use the least energy, and a variety of other scenarios designed to improve efficiency and save money.
On Wednesday GE said that it has brought in $290 million so far this year from products built using this industrial internet philosophy and booked an anticipated $400 million in revenue. That may not seem like much for a company that had sales of $147.36 billion last year, but this is a two-year-old effort inside GE. GE also expanded the line of product offerings it has in its Predictivity line from 10 to 24, announced Intel, Cisco and AT&T as its latest partners and detailed its platform for building out the industrial internet, called Predix. Think of it as Amazon Web Services for the industrial internet.
The announcement basics
- Predix is a platform for industrial applications. Applications can be built for any system or machine — from jet engines to MRI scanners — and be remotely managed while connected to the internet. So far there are four components to the platform, for the sensors themselves, analytics, management of the connected devices, and a vague one called Predix Experience.
- Next year GE plans to offer a developer program that lets third parties integrate Predix platform technologies into their own services.
- Of the new partners, AT&T will handle connectivity via cellular, wireline and perhaps even Wi-Fi management techniques courtesy of AT&T’s Wayport division.
- Cisco and GE will continue an existing business relationship to “include collaboration in industries that may include oil and gas, transportation, healthcare, and power generation.”
- GE says it will work with Intel to “embed virtualization and cloud-based, standardized interfaces within the GE Predix platform.”
What it all means for GE and the internet of things
With this announcement it becomes clear that GE is following the web-oriented model of building out a platform, taking care of the hard problems associated with industrial data, analytics and hosting data in the cloud on behalf of its clients. To be clear, GE has its own servers, its own database technologies and other aspects of this platform, but this “big tent” approach will help it deliver a wider array of functionality as well as a scale that it might need as more customers get on board.
AWS as a partner provides the baseline compute, while Pivotal will provide both the data analytics and PaaS-like structure — kind of like a Heroku for building on Predix. Developers will use Pivotal’s tools to build applications that can help take advantage of GE’s Predix platform. So while there are 24 products right now for GE customers, there might one day be hundreds more made by GE customers, or perhaps outside developers, built using Pivotal’s tools.
As for connectivity and AT&T, I asked Bill Ruh (pictured above), vice president of GE’s software and analytics center, about AT&T’s pricing models, because it’s still unclear if clients will want to buy connectivity separately or as part of some overall Predictivity product. It’s also unclear if AT&T has pricing plans in place that treat cellular, wireline and other services as one monolithic option. When you have 100 sensors on every car in a fleet, you don’t want to think individually about the optimal way to connect each.
Ruh isn’t sure yet what clients want (they probably aren’t either) but, whatever it is, he’s prepared to give it to them. I’m hoping he’ll talk more about the challenges of ubiquitous connectivity when he speaks at our Mobilize event next week.
The bigger surprise was the inclusion of Intel, which Ruh described as the way GE planned to embed smarts both on the back-end servers but also in edge devices where ARM and other chip vendors have played for years. “We’ve used Intel in our machines for a long time,” Ruh said. “Quark is a great product and we saw that, and realized that is an opportunity to leapfrog to the next generation. The world is changing and everything will need to have intelligence in it, even at substandard-size.”
But given that Intel has also announced a software platform for the internet of things, as well as its own hardware-based gateway devices for handling client data, it’s also a measure of overlap occurring in this market as the giants (and a few startups) try to build out the definitive platforms for the industrial internet of things.
A messy market and early days
“The market is messy in these areas,” Ruh said. “It’s not a mature market for us but that’s the benefit of having $2-$3 billion to invest in 2-3 years. We can validate products and make a lot of mistakes because the end is unclear. The challenge is to be able to have the flexibility to move between solutions and protect our customers from the changes in the market. Our customers don’t have the investment and the tools to do it themselves, and the outcome is still a long way off.”
Ruh is aware that the pieces of GE’s platform are likely to change and that partners and competitors can come in and commoditize parts of that market. But he’s also aware that GE right now has something they don’t — data. When you are trying to figure out the optimal RPM at which to run an engine in a given situation, GE can tell you, based on their knowledge of building that machine or merely from servicing it. The Predix platform and Predictivity products are simply a way for GE to get even better data while offering the real advantages of the internet of things to its clients.
And much like a critical mass of apps can help a platform, or a large number of developers who can build on your IaaS or SaaS product can make a large difference to your success, data will build the barrier to entry for others who want to do the same thing.
This is a big opportunity, which is why GE is going big to keep complexity under the covers and give clients the reality of responsive businesses that can adapt to their environmental or business goals. So if it has to move from Intel chips to ARM later, or change its Hadoop distribution down the road, Ruh is fine with that, as long as the data keeps making his platform smarter than anyone else’s.
“We want to deliver outcomes and tech compatibility, and hide the vagaries of the market as it works itself out,” Ruh said. “This is not something you can do on a nickel and be successful at it.”
Updated at 8:42 am to correct the spelling of San Ramon.