Memory for industrial automation is becoming more important as connected equipment, analytics and artificial intelligence increase the volume and complexity of operational data. Sensors, vehicles, network infrastructure and cloud services all create information that must be stored and processed. For automation engineers and operations leaders, the practical question is how memory can support those workloads while meeting application needs for capacity, speed, power use and reliability.
A Samsung Electronics memory presentation from June 2019 described this shift through several connected themes: growing data volumes, demand for higher-density and higher-performance memory, application-oriented manufacturing, and quality requirements in automotive and other demanding markets. These themes offer a useful framework for understanding memory’s role in industrial and connected systems.
Why industrial data is changing memory requirements
Industrial data increasingly comes from interconnected sources rather than a single machine or control system. The presentation identified public cloud growth, smart cities, 5G-connected devices, autonomous vehicles and the Internet of Things as contributors to expanding data flows. In an industrial setting, comparable sources include network equipment, sensors, connected vehicles and applications that send information to edge or cloud infrastructure.
Not all of this information has the same structure or processing requirements. Structured data can be organized according to defined formats, while unstructured data may require different storage and analysis approaches. Big data analytics and machine learning can place additional demands on computing systems, where memory capacity and access speed influence how data is made available to processors and accelerators.
This makes memory selection part of system design, not simply a component-level purchasing decision. Engineers need to consider the workload, where processing takes place, and how much data must be held close to the processor or retained in storage.
Memory for industrial automation: capacity, speed and latency
The presentation links big data analytics with demand for memory that offers greater density and performance. It identifies DRAM and solid-state drives (SSDs) within data-processing environments, and notes the use of accelerators such as GPUs, FPGAs and ASICs alongside high-bandwidth memory (HBM). These technologies serve different roles, but together they illustrate why a system may need both substantial storage capacity and fast access to working data.
Latency—the time delay between a system input and its output—is another consideration. In data-intensive applications, faster response can depend on how efficiently information moves through the computing architecture. The appropriate design will vary by application: a workload that prioritizes high capacity may have different requirements from one focused on low latency or sustained processing performance.
For automation teams, practical design questions include:
- Which data must be processed locally, and which can be sent to a data center or cloud service?
- Does the workload prioritize capacity, response time, bandwidth or a combination of these?
- How will memory and storage choices fit the processor, accelerator and network architecture?
- What reliability and operating-life requirements apply to the installation?
Edge, cloud and application-specific workloads
Memory demand is shaped by where computing happens. Public cloud services support large-scale data processing, while edge computing brings some processing closer to devices and operations. The presentation also points to on-device AI, smart homes, IoT and in-memory databases as areas associated with evolving memory needs. In industrial deployments, the balance between local and centralized processing depends on the application and its data flow.
Application requirements also differ across sectors. The presentation groups automotive, industrial, networking, high-performance computing and AI among the markets requiring distinct approaches. For connected and autonomous vehicles, it describes memory solutions intended for automotive applications, including automotive DRAM and eStorage, with attention to reliability, temperature notification, data refresh, capacity and power consumption. These are application considerations rather than a single specification that applies to every vehicle or industrial system.
Manufacturing flexibility and quality management
Memory manufacturers must also manage production across products with different performance and longevity requirements. The presentation describes a two-track approach: leading process technology for performance-oriented products and proven process technology for applications where longevity is important. It also identifies DRAM and NAND conversion as a way to support production flexibility as demand changes.
In a factory, supply-chain management and process control affect the ability to align production with customer demand. The presentation associates smarter manufacturing with data-based operations, efficient logistics, lead-time reduction, inventory management, demand fulfillment, automated operation and early yield improvement. It also emphasizes seamless quality control and intelligent planning. These are manufacturing priorities that connect operational data with production and supply-chain decisions.
Quality requirements are particularly relevant when memory is intended for automotive or other demanding applications. The presentation describes a progression of quality systems and requirements, including ISO 9001, TL 9000, QS 9000 and IATF 16949. It frames quality by design as predictive, preventive, collaborative and prescriptive. For technical decision-makers, the implication is to assess quality processes and application requirements alongside device performance and capacity.
Turning memory choices into system decisions
As industrial systems become more connected and data-intensive, memory architecture influences how information is stored, accessed and processed across devices, edge systems and cloud platforms. The right balance depends on the workload: density, bandwidth, latency, power consumption, reliability and operating-life expectations all matter, but their relative importance varies by application.
Engineering teams can use these requirements to guide discussions with system suppliers and memory vendors, while operations leaders can consider how data flows affect production, quality and supply-chain processes. Treating memory as part of a broader computing and manufacturing strategy helps align component choices with the actual needs of industrial automation.