Software and networking companies are likely to disrupt the industrial and commercial automation and control markets as more intelligence gets pushed to the edges of the network.

The industrial controls and factory automation market is over $150 billion in 2013 according to Markets & Markets.  The vast majority of this market consists of either stand-alone automation components or systems integrated into a centralized control or automation software system (e.g., SCADA in the utility sector).  According to a 2014 PWC survey only 24% of companies said they had deployed an extensive network of connected sensor devices.  Incumbents in this arena are racing just to get their devices connected to the internet and communicating data back to the cloud.  If this were a baseball game, we’d be in the second or third inning.

At a time when incumbent industrial automation companies are struggling with connectivity, IT-oriented companies such as Cisco and others are accelerating their push to bring intelligence to the edge of the network. Fourteen months ago, Cisco introduced the term “fog computing”.   Today when you enter that into a Google search, you get over 1.2 million results.  “Fog computing” refers to embedding computing power throughout a network instead of in a central “cloud”.  The idea is to reduce the costs of data transmission and increase the speed of analytics in a world of billions of connected “things”.

Bringing intelligence to the edge of an industrial network via concepts like “fog” opens significant new opportunities for customers.  With computing power placed closer to the end devices, customers can utilize the huge amount of real-time data that comes off machines.  Instead of doing post-event analytics, machine learning embedded at the edge can improve the performance of machines in real time allowing them to get better over time.  Video is one of the fastest growing types of data.  Edge intelligence allows analytics to occur closer to the devices themselves reducing the cost of transmission of high cost data traffic.  As these edge devices become more connected, the power of the internet of things can really emerge.  The machines can communicate with one another in real time.  Machine learning algorithms enable those networks to become more effective working together than as stand-alone reporting devices back to a centralized data store.

Who will own the intelligence at the edge?

Requirements:

• Network computing expertise:

Creating an automation system with the right intelligence placed in the right places requires understanding of networked computing hardware and how best to design such a system given the computing and data communication requirements.  How much memory is necessary? What is the most efficient network system for communication? What information needs to be centrally analyzed and what can stay at the edge? All of these questions and many more computing-oriented questions are critical to build the right solution.

• Analytics software expertise

Machine learning/deep learning and other forms of advanced analytics will further enhance the power of the intelligence at the edge. Writing self-learning analytical engines to embed out in the network so that all data doesn’t have to flow centrally is critical in this vision. Most, if not all, of these algorithms will be originally constructed using historical data.  But over-time embedding those into edge devices will improve the speed and actionability of the decisions.

• Device-embedded software

One of the most powerful impacts of connected devices is the ability to do over-the-air updates to devices in the field.  This requires embedded software that is designed for upgradability in the field.  Most of the embedded software in today’s industrial automation is not upgradable at all…and certainly not with simple pushes of OTA releases.

• Domain expertise

Industrial automation equipment is purpose built with deep knowledge of the industry and specific solution for which it is intended.  This domain expertise is a critical differentiator for industrial technology firms.  In many cases the little things make significant differences.  And many of these things are not seen until you are deep into user testing.

Contenders:

• Network computing vendors (e.g., Cisco)

These players bring strengths in software and network computing.  They are already trying to integrate deeper domain expertise into their devices through partnerships and hiring of key industry personnel.  They will never have the level of depth that a focused industrial player will bring.  They can potentially build more and more of the cross-device intelligence into their devices enabling them to capture larger share of the total value available from the device manufacturers.

• Industrial automation equipment vendors

These players obviously bring deep domain expertise as well as large installed bases of equipment.  In some cases, they also have their own automation control software products (e.g., Emerson’s PlantWeb or Johnson Controls and Siemens Building Automation Systems).  Industrial customers are loathe to replace automation equipment because of the risks of downtime or safety issues.  However, these companies typically do not have the level of software or networking expertise of their new rivals in this marketplace.

• New entrants with an “edge intelligence” focus

We see the emergence of new entrants into this market who will blend domain expertise, software, analytics and networking expertise to capture value from the industrial automation vendors.  They won’t seek to eliminate the Emerson’s of the world.  Instead, they will offer value added capabilities by bringing greater intelligence to the system.  These companies will bring analytical software and connectivity to the edge of the network to allow various OEM devices to work better together.  They will use Moore’s Law to their advantage to embed more intelligence into standard network computing capabilities to enhance the productivity and safety of industrial and commercial processes.

Conclusion:

We are in the very early days of this battle.  Industrial automation is a very attractive market: large, growing and highly profitable.  To date, incumbents have held the upper hand given their installed base and track records.  However, IOT and, in particular, technologies like “fog computing” or “intelligence at the edge” will open this market to new players with new strategies to deliver ever increased productivity and safety with new technologies.