Artificial intelligence (AI) is growing at a rapid pace and becoming increasingly significant for use in pharmaceutical manufacturing.
Shutterstock AI
Within the past few years, there have been tremendous advances in using AI for process design and scale-up, advanced process control, smart monitoring and maintenance, visual inspection, and trend monitoring for deviations and Corrective and Preventive Action (CAPA).
As AI becomes more integrated into pharmaceutical manufacturing, the FDA recommends that execution-control software be located as close as possible to its associated equipment to avoid adverse effects on performance or security. For this use case, a local single-GPU system is ideal for deployment close to the workload, providing a compact and cost-effective platform for AI deployment.
Single-GPU computers are a cost-effective and space-conscious platform for AI workloads. As its name suggests, a single-GPU AI computer relies on one graphics processing unit to handle the intensive mathematical processing required for AI applications. These systems can be configured to support a range of AI workloads across the pharmaceutical manufacturing lifecycle, from data collection and machine learning to vision inspection, modelling, and inferencing.
This has significant practical applications. One example is the use of AI inference to monitor bioreactors. Inference refers to applying a trained model to new data to generate predictions, classifications, alerts, recommendations, or control signals. With bioreactors, AI inference can be used to continually monitor and evaluate multiple process variables including pH, dissolved oxygen, temperature, nutrient levels, and cell growth to determine whether the process is working as intended and adjust the parameters of the control.
With so many potential configurations and uses, single-GPU AI computers become even more versatile when integrated into mobile workstations built for the same role. These workstations are designed to handle the heat generated by moderate-wattage single-GPU systems while enabling deployment where the computing power is most valuable: directly within the pharmaceutical cleanroom.
Mobile workstations for AI
One of the biggest challenges in adopting local AI computers in a cleanroom is the significant amount of heat generated by AI computing systems. The workstation's high heat output will result in an increase in the room temperature, changes to the airflow in the room and difficulty in maintaining a stable humidity level. Even a moderately powered single-GPU AI workstation can use hundreds of watts.
Fortunately, advances in workstation design are beginning to address this constraint. Mobile workstations from companies like ARISTA Corporation are now being built specifically for AI processing and can now support computing loads of up to 500 watts while still meeting the strict demands of controlled cleanroom environments. These workstations employ smaller form factors, improved thermal management, and sealed or cleanroom-compatible enclosures.
“The workstation is flexible and can be deployed where it is needed and connected as required. The greater value, however, lies in its ability to be configured for different AI workloads across the entire manufacturing lifecycle,” said Paul Shu of ARISTA Corporation.
Wi-Fi is integrated into each workstation so that operators can communicate wirelessly throughout the entire facility, regardless of location.
A range of unit options is available, including varying touchscreen LCD display sizes from 18 to 24 inches and single, dual, or triple screen configurations. Multiple touchscreens support multi-tasking beyond anything else available today. Operators can see the SCADA System on one monitor and look at documentation such as inspection checklists and training materials on the other. The display of the screens is aligned to the natural line of sight of a typical operator, which helps to reduce eye strain and fatigue.
Each single-GPU computer is housed in an enclosure made of 316L stainless steel. This provides a high level of protection against chemicals and corrosive substances such as cleaning agents and solvents typically used in pharmaceutical environments. Three industrial waterproof keyboard options are available to accommodate different facility requirements.
Shu adds that additional customisations are available depending on customer needs.
The practicality of AI in pharmaceutical manufacturing is directly dependent upon how easily it can be integrated into day-to-day operations.
This is where deployable, purpose-built workstations provide a clear advantage. By bringing single-GPU AI computing power into pharmaceutical cleanrooms in every phase of the manufacturing process, mobile workstations make AI more practical, more accessible, and more effective for real-world manufacturing use.