With aging systems limiting efficiency, security, profitability and the ability to take advantage of AI, the manufacturing industry is in the midst of a data transformation. The stakes for change are high, driven by increasing demand for critical goods such as semiconductors, the growing understanding of (and concern over) supply chain vulnerabilities, and the potential for significantly more intelligent and streamlined operations.
Legacy technology poses significant challenges in manufacturingβnot least, siloed technology that limits organizationsβ ability to synthesize and make informed decisions from real-time data. Legacy systems are also more costly to maintain, harder (if not impossible) to scale and prone to operational inefficiencies and downtime.
Indeed, the 2024 Manufacturing Industry Outlook report, compiled by the Deloitte Research Center for Energy & Industrials, cites a study noting that manufacturers anticipate βthe industrial metaverseβ could lead to a 12 percent gain in labor productivity, with generative AI βexpected to hold immense potential in areas such as product design, aftermarket services, and supply chain management.β
Weβre seeing this across industries, with manufacturing becoming one of the largest data producers because more and more connectivity is occurring. This leads to better system optimization and the ability to add more AI, more enterprise reporting, better system scheduling and a slew of other productivity improvements. For example, in the pharmaceutical industry, lab instruments at the edge are now data creators. In a modernized environment at one pharma company, that data is streamed in real time to an analytics repository to automate regulatory reportingβreporting that was formerly performed manually based on data collected from disparate spreadsheets.
Whatβs the Delay?
But if a move to βsmart manufacturingβ will enable manufacturers to optimize operations and navigate a challenging labor market, whatβs the delay for some organizations?
To put it bluntly, it isnβt easy. Legacy systems are often incompatible with modern technologies, making integration into a single interface difficult and costly.
Perhaps a bigger hurdle is the fact that these systems are chugging away and taking care of the business at hand. They may not be pretty, nor aware of other systems in the organization, but they are deeply entrenched in daily operations. No one would argue that modernizing their capabilities and integrating them tightly into a cloud-native infrastructure make sense, but fears about the workflow disruptions that replacing them would cause are hard to overstate.
With all that said, the future is now, and manufacturers cannot continue to put off whatβs inevitable if they want to compete. Another study cited in Deloitteβs β2024 Manufacturing Industry Outlookβ report noted that βa striking 86 percent of surveyed manufacturing executives believe that smart factory solutions will be the primary drivers of competitiveness in the next five years.β
Manufacturers tied to legacy systems will benefit from the use of AI, but they wonβt reap its full potential. The industrial ecosystem requires rapid development and adoption cycles to maintain operational efficiency and productivity. It also requires systems that can communicate and work together seamlessly. Legacy systems just canβt do that.
Indeed, to leverage AI effectively, manufacturers must modernize their infrastructure to support secure, scalable and manageable edge computing. Manufacturers need to build and operate functions consistently from the cloud to the edge and from large-scale systems to small form factors. This consistency will simplify development, testing, deployment and management, making operations more efficient and scalable.
Fortunately, several factors have come together that will ease the transformation of decades-old systems. These include more powerful edge devices, enterprise-grade open source edge platforms and tools, shrinking and more purpose-built large language models (LLMs), and a growing understanding of the need to break down IT and OT silos to optimize the use of data.
In addition, the digital β or, more to the point, data β transformation of legacy systems does not have to be a rip-and-replace situation. Indeed, itβs important for manufacturing organizations to realize that they can start with basic MQTT to transport and transform data so they can begin to leverage it to learn and build models.
From there, manufacturers can update their legacy systems in a way that optimizes AI and edge computing through a strategic modernization approach.
The first step is to identify legacy systems running in the organization (in every nook,cranny and server closet) and detail what functions they serve and who uses them. These might include analog and soft controllers, sensors and drives, manufacturing execution systems (MES), and historians. Some will be candidates to modernize while others you may take the approach of pure data extraction.
A phased approach will enable manufacturers to chip away at monolithic and siloed applications and break down still-useful functions into more manageable components that can be migrated to the cloud or hosted on-premises to enable easier integration with AI and edge computing technologies. This microservices-based approach will pave the way toward hybrid deployments, providing manufacturing companies with a more flexible, scalable and process-oriented approach to modernization
Security will definitely be a complicating factor in this migration. With more and more interconnected devices, it will be important to implement advanced authentication and encryption methods to secure data transfer among legacy systems, AI platforms and edge devices. Manufacturers should know their limits in this area, and should seek to partner with security-conscious technology providers at every turn.
Wherever possible, manufacturers should also implement enterprise-supported open source solutions, which play a crucial role in modern manufacturing by fostering innovation and collaboration while providing security, governance, scalability and support. They enable manufacturers to adapt and customize software to meet specific needs, promoting a more flexible approach to technology adoption.
Finally, itβs important to remember the role that people play in this transformation. Organizations should invest in training programs to upskill employees on new technologies and processes, and to inform new hires about the legacy systems that are in the process of being modernized and/or that will stand as is for the time being. Manufacturers should also work to develop a culture that embraces ongoing technology and operational innovation.
In the manufacturing industry, the powerful combination of AI, edge computing and automation will very quickly become table stakes. By taking a careful, staged approachβwith an appropriate sense of urgencyβmanufacturers can balance their need to keep the trains running and to adapt and adopt technology that will maintain their ability to compete.
If we are going to focus on AI for this, then we need to have a focus on data transformation. We donβt have to wait for them to modernize everything to start. Many are starting with basic MQTT to transport and transform the data so that they can begin to leverage the data to learn and build models. I would suggest we add that in here.