This week Reuters did a deep dive into problems with GE’s industrial IoT efforts. The company had taken a two-month “time out” to fix problems with its Predix software platform designed for the industrial IoT and has asked executives to cut costs. GE is also backing off selling the platform to all industries, and will instead focus on oil & gas, aviation and energy.
For more discussion on this story, listen to the latest IoT Podcast.
Since launching its industrial IoT effort five years ago, GE has spent billions selling the internet of things to investors, analysts and customers. The idea of adding sensors to its equipment, analyzing the data captured from those sensors and using that to generate business insights wasn’t new, but GE made a huge effort to commercialize it.
GE is learning that the industrial IoT isn’t a problem that can be tackled as a horizontal platform play. Five years after it began, GE is learning lessons that almost every industrial IoT platform I’ve spoken with is also learning. The industrial IoT doesn’t scale horizontally. Nor can a platform provider compete at every layer.
Get more IoT analysis like this in the weekly newsletter.
For example, Samsara, a startup that formed in 2015, aimed to build a wide-scale industrial IoT platform that started with generic sensors. It has since narrowed its focus to fleet monitoring and cold-chain assurance, which is how some of the earliest users of its product used it.
Helium is another company that had a similar vision. It makes sensors and sends data over a proprietary wireless network to a gateway or (originally) the Helium cloud. The original plan was that companies would buy the sensors and build out rules in the cloud. But that was too open-ended, and Helium now has decided to link data from its sensors to other provider’s clouds. It is also focusing on managing those sensors from a Dashboard setup.
In Samsara’s case, the company went from a broad focus to a few specific industries. In Helium’s case, it decided to stop building out a cloud where industrial customers could host their data so it could focus on management software and act as a connector between customers’ data and their cloud of choice.
GE’s decisions are similar.
A year ago it decided to stop building out its own cloud data centers and started signing partnerships so customers could run Predix in Amazon’s or Microsoft’s clouds. As part of the current refocusing, it appears to be narrowing its customer focus to industries where it already has customers using its equipment.
This makes sense. When I began covering the industrial and enterprise internet of things it quickly became clear that while IT could be generic, the data analytics elements and the real-world devices in the field that were being monitored would require specialized knowledge. What has happened is that at the computing (and even device management layer) larger cloud providers like Microsoft and Amazon are winning.
I think we’ll see connectivity also steer toward the big platforms, whether its cable providers offering LoRa networks, cellular companies, or a few of the specialized low power wide area network providers. The hardware will likely be dominated by bigger companies such as Intel, Dell, HPE and those with established IT credibility. We’ll also see industrial gateways from the likes of Advantech and Eurotech.
But it’s the many areas of specialization here the industrial IoT gets interesting. I think we’re going to see startups emerge around data analytics and AI such as Uptake, C3, SightMachine, Flutura, and Foghorn. Many of these will have specific areas of expertise such as C3 with energy and Flutura with oil and gas.
There’s also a role for specialization in building the overall IoT solution for companies. A hospital aiming to connect patient rooms needs a different set of knowledge around regulation and security than a company trying to create connected conference rooms needs. Today, those areas of specialization are ripe for startups.
Rick Bullota, who founded Thingworx which was sold to PTC, said that he anticipates a lot of M&A in the coming 12-24 months as more businesses embrace the insights and business transformation enabled by the industrial IoT. He also says that the buyers of industrial IoT products are coming more from the business side than the tech side, which means that the overall package of technology will become more important, rather than the individual pieces.
Get ready to hear more about end-to-end solutions.
Meanwhile, GE’s software and industrial IoT efforts are still impressive. While it planned on revenue of $15 billion by 2020, it is now scaling that down to $12 billion in part to clarify what is software versus connected hardware it sells to industrial clients. Back in 2015, GE’s Kate Johnson, then-CEO of Intelligent Platforms, told me GE reported $4 billion in software sales in 2014 and expected $7 billion in 2016. (Johnson left in July to work at Microsoft.)
So when it comes to the industrial IoT, the opportunity is big. It’s just taking a while to figure out how to attack the market. Step one is realizing that it’s silly to take on the big cloud and connectivity providers. Step two is quitting the platform dreams and focusing on a specific area of expertise.
Great insight Stacey, liked it so much I subscribed.
Early days for IIoT and I’m learning a lot, interesting from a ground floor perspective as I can see applications where it provides ROI but many that will not.
Early adopters may get stung as a disrupting vertical platform springs from nowhere.
My money in the short term is ‘plug and play’ cheap, robust packages that add value straight away, customisable, requires little or no upkeep that the customer doesn’t have to think about.
Regards
Andy
Hey Andy,
Not urgent.
A few questions in follow-up to your comment.
Please give examples of:
1. ‘plug and play’ packages
2. “cheap” yet “robust”
3. “cheap” yet “customisable”
4. platforms that come with “little or no upkeep that the customer doesn’t have to think about?”
I’m betting that you have more experience in this area than I do!
Educate.
Thanks!
Chris
Hi Chris!
I work for DATTUS (www.dattus.com). We check all of those boxes.
Chad
Hi Chris, for the commercial buildings vertical take a look at http://www.demandlogic.co.uk. Our company uses one hardware device installed in under 3 hours to connect to almost any existing commercial building. We model all of the Heating, Ventilation & Air-Conditioning equipment, plus all sensors and meters, into our virtual Asset Model and start polling live data within 24 hours. Presenting the data analysis back via the web through our collaboration portal. This rapid low-cost digital makeover means that the time-to-value is short, a matter of months rather than years.
If you have any questions please drop me a line.
Sonny
Stacey, useful insight on GE – very useful. It seems to be common for hardware companies trying to transition into the IoT market to think of the high volume of potential edge devices they can sell and then making an enterprise pitch with a 3+ year payback. Taking a different approach would mean a fundamental shift in the business model, which would be a high-risk proposition for shareholders. That hardware-first mindset is the starting point rather than asking what is the customer problem, is there existing data and can we put it to better use to deliver value. Once we proven value, then enrich the data set.
As a young start-up, Demand Logic, can afford to build up a hardware-light data-first service which focuses on quickly acquiring existing data points from equipment, sensors and meters. In some of the buildings we monitor there are between 180,000-200,000 existing data points under-utilised. In most offices over 50,000sq ft, there are 10,000+ data points. We can show value based on this information and then we go back and have a conversation about new use cases, for instance air-quality sensor monitoring for health, and occupancy sensing for space planning. This adds to an already rich digital picture rather than starting by selling new data, so the improvement programme is already paying for itself.
Hi, great reading.
I’m trying to get an assessment of how much does it cost (in a ball park) to “transfer” a manufacturing plant into “smart” plant (I know its differ and depends in different parameters)
Can you recommend of any studies, publications, case studies etc. so I can get a notion of how much does it cost and what are the costs?
p.s. I read Jasper’s publication. it’s good but very mobile smart platform oriented.
thx!