GE Digital: Can Machine Learning Be the Key to Turning GE Around?

In 2011, General Electric (GE) made a big bet on machine learning and AI with the creation of what would become GE Digital. In a time of great flux at GE, will Digital survive? What will be its role be in the new GE?

At the end of his 16-year reign as CEO of General Electric (GE) in 2017, Jeff Immelt wrote that he was leaving GE as a “125-year-old start-up.” A year earlier, the company had invested $4 billion to strengthen its analytics and machine learning capabilities that was powering the historic conglomerate’s transformation from a manufacturer of jet engines, gas turbines, wind turbines and locomotives into a digital industrial giant [1]. Its clients recognized the value of the data coming from the GE industrial equipment to drive process improvements, and GE set out with the goal of automating data driven decision making to drive gains in gas turbine fuel efficiency, improve wind turbine output, decrease power plant downtimes and increase heavy service equipment intervals [2].

GE had an inherent head start in the Industry 4.0 transformation since they had been collecting data from sensor on their industrial equipment for decades. In 2012, they created the cloud-based Predix software platform to provide a means to collect sensor data to help machine operators and maintenance engineers improve machine efficiency, schedule maintenance checks and reduce downtime [3]. Using the data collected from these smart machines and domain knowledge, engineers create a “digital twin” of a machine (ex. gas turbine) or a system (ex. gas burning power plant) to model the state of a customer’s asset. Advanced machine learning algorithms allow these digital models of assets to continuously monitor, validate and update in real time, thus allowing for automatic detection of any deviations in system performance [4, 5]. In a gas fired powerplant, for example, the Predix system machine learning solutions give companies the ability to monitor the health and automatically detect anomalies in their gas turbines real-time to reduce downtime, improve turbine efficiency through nonlinear optimizations of combustion processes, and optimize the dispatch of energy to the grid based on supply/demand [4].

GE’s management has recognized that machine learning techniques are a key enabler to using their hardware more efficiently, and can be one of their key drivers of growth as they look to innovate in mature industrial industries [6]. In an attempt to continue to grow its expertise in the field of machine learning and create synergies between business, management has grown GE Digital’s through acquisition. In 2016, GE Digital purchased startups Meriduim (equipment management software) and ServiceMax (industrial field service worker management software) for a combined $1.4billion [7]. It also added startups such [8], Tamr [9] and IQP [10] to its portfolio to continue strengthen its competencies in artificial intelligence and finding effective ways to apply it to GE’s existing businesses. In July 2018, GE announced a partnership with Microsoft to deploy Predix on Microsoft’s Azure cloud infrastructure and to co-develop machine learning applications for digital industry applications [11].

However, while in 2011 GE Digital was one of the few players in the industrial digital field, the landscape has drastically changed. Industrial analytics solutions from industrial giants like such as Siemens, Honeywell and ABB and cloud-computing experts such as Google and Amazon pose real threats to GE Digital’s Predix platform and challenge for GE Digital place as a leader in digital industry [7].

Perhaps the largest challenge faced by GE Digital and its ability to continue to strengthen position in the digital industrial revolution is the strength of the company. With the ousting of John Flannery as CEO, an ardent supporter of GE’s digital transformation [12], and large budget cuts at GE Digital [13, 7], it is unclear what is in store for the arm of the conglomerate. Predix and its machine learning capabilities has demonstrated to be a key product in GE’s portfolio the past decade, and the company must continue to leverage its capabilities as the industry moves faster towards Industry 4.0. GE has the unique advantage of having a century of domain knowledge in the design and analysis of the industrial systems being monitored and analyzed with the artificial intelligence techniques. GE’s management team should thus continue to push the integration of machine learning earlier into the product development process. GE being able to create a focused Industry 4.0 product portfolio that integrates hardware, data capture, analysis, monitoring and adaptive optimization will be a key enabler fuel the growth of the new General Electric.

How does a struggling industrial giant continue to invest in and grow their machine learning competencies in times of restructuring? Should a company like GE be outsourcing the development of these tools to external experts and focus on hardware/system integration or continue to develop them in-house?

(Word Count: 751)


[1] J. R. Immelt, “How I Remade GE,” Harvard Business Review, September 2017.
[2] P. High, “The CEO Of GE Digital On What Is Next For The Industrial Icon,” 23 July 2018. [Online]. Available: [Accessed Nov 2018].
[3] L. Winig, “GE’s Big Bet on Data and Analytics,” MIT Sloan Management Review, Feb 2016.
[4] GE Power Digital Solutions, “GE Digital Twin: Analytic Engine for the Digital Power Plant,” 2016.
[5] R. Bean and T. H. Davenport, “How AI And Machine Learning Are Helping Drive The GE Digital Transformation,” 27 June 2017. [Online]. Available: [Accessed November 2018].
[6] E. Woyke, “General Electric Builds and AI Workforce,” MIT Technology Review, June 2017.
[7] S. Lohr, “G.E. Makes a Sharp ‘Pivot’ on Digital,” The New York Times, 19 April 2018. [Online]. Available: [Accessed Nov 2018].
[8] F. Lardinois, “GE Acquires to Deepen its Machine Learning Stack,” Techcrunch, 2016. [Online]. Available: [Accessed 2018 Nov].
[9] B. Darrow, “GE Saved Millions by Using This Data Startup’s Software,” Forbes, 17 May 2017. [Online]. Available: [Accessed Nov 2018].
[10] W. Ruh, “GE Digital’s latest IIoT acquisition makes Predix and its industrial applications more accessible,” GE Digital, 27 July 2017. [Online]. Available: [Accessed 2018 Nov].
[11] B. Ruh, “Accelerating Industrial IOT Adoption for Customers with Microsoft,” GE Digital, 16 July 2018. [Online]. Available: [Accessed Nov 2018].
[12] J. Flannery, “GE’s Long Digital Game,” GE Reports, 1 Aug 2018. [Online]. Available: [Accessed Nov 2018].
[13] D. Cimilluca, D. Mattioli and T. Gryta, “GE Puts Digital Assets on the Block,” The Wall Street Journal, 30 July 2018. [Online]. Available: [Accessed Nov 2018].


Juul Labs: Transforming Vaping through Open Innovation


When Chanel trades sewing machines for 3D printers

7 thoughts on “GE Digital: Can Machine Learning Be the Key to Turning GE Around?

  1. Innovative take on a widely covered company. I would strongly favour the development of a in-house tool in order to maintain full ownership and confidentiality regarding the programming parameters, avoiding copy-cats. This is not to say that external experts should be disregarded, on the contrary, I would actively pursue them to join the team. By bringing those experts in-house, you could benefit from their uniquely held knowledge and expertise, whilst at the same time maintaining their current structure.

  2. I think it is a legitimate question to ask GE when they are struggling to cut cost and increase their bottom line. In a lot of companies, development is always regarded as a cost center. However, as the article mentioned, more competitors are going into this field as well, it is a long-term investment to maintain the competitive edge to other companies. The tension here is real and out-sourcing doesnt seem like a good idea since the data could be sensitive to their clients. Moving forward i would put more effort into integrating the newly acquired companies and choose a few focus area to develop the technology.

  3. I would argue for partnering with enterprise-focused technology companies like Microsoft to help develop products and learn from the experience before trying to do it themselves.

  4. Thanks for the piece. I think this innovation at GE will be crucial to their future profitability. They need to move into technology to not be left behind – in an aging and competitive industrial landscape where GE has taken significant reputational damage recently they need to differentiate themselves with their business consumers. One significant way for GE to do this is to offer machine learning insights and technology-enabled platforms. Given GE’s unique position in the market they have – as you mentioned – access to highly valuable data. In order to keep this data proprietary and in-house GE should invest in machine-learning within their organization and not outsource development. With outsourcing I worry they will not create a differentiated product for consumers nor will they continue to improve upon it. Without constant improvement their formidable competitors, Google, Amazon and other industrial companies, will take market share.

  5. This is a very intriguing article. Although my initial conclusion was to outsource the capabilities to another company and have GE focus on their core business, as I re-read I started to think whether data collection and analysis shouldn’t be their differential. Let me give one example: suppose they contract IBM do develop an algorithm to collect and analyze data from GE machines around the world, what would hold them from selling this technology to a competitor?
    On a unrelated note, I struggle with the ownership of the data and privacy. Collecting that much data from clients might give GE an unfair advantage, for instance, in future bids. GE has huge responsibilities by collecting, storing and analyzing this data and they should be concern with the risks this brings.

  6. Two thoughts come to mind reading this article. First, I think it is a mistake to outsource machine learning for GE. As we move to a more digital future, these competencies will be table stakes to remain competitive as a firm. GE has a true competitive advantage in their data, so they should invest in machine learning to take advantage of this. Second, it is difficult to shift a firm from non-digital to digital and be as effective as digitally-native firms. I think of Walmart as an example of a large, non-digital firm that has been able to make this shift successfully. Walmart did this through a combination of two things: (1) heavy internal investment and (2) acquisitions. Walmart built out a large e-Commerce office in San Francisco to attract the best talent and invest in their e-Commerce capabilities. Additionally, Walmart bought digitally-native firms such as Jet and Bonobos to bring in expertise. GE has engaged in similar activities and it should continue to do so. It will take time and GE should be patient for investments to pay off.

  7. Thanks for the write up on GE! I agree with Dom above that outsourcing machine learning would be a terrible mistake for the company. Given the amount of disruption and restructuring the company is going through, it is more important than ever that GE focus on growth opportunities that can help right the ship. I believe GE should make a few big bets in machine learning and double down on devoting resources towards these new initiatives.

Leave a comment