We have been talking about artificial intelligence in the chemical industry for more than five years now. In the beginning, the conversations revolved mostly about whether it would have an impact. Today, we will talk about how.
In fact, nearly two-thirds of the industry agree with the significance of AI potential in improving operational efficiency. As per 71% of surveyed chemical companies agree that AI is a force for good – driving business efficiency and therefore creating positive outcomes for all.
And this is only the beginning. We expect to see 32% average annual growth in AI expenditures in the chemicals industry over the next 10 years, reaching nearly $17 billion in 2032. One rough proxy of where those investments will be made is a look at where the chemical industry is making venture investments in AI. Over the last six years there have been 64 deals -28 in R&D and engineering, 25 in manufacturing and production, and 11 in sales and marketing.
Most of these projects will use machine learning and other traditional AI technologies. At the moment, we expect only 5-10% will be in Generative AI. This will change as the current capabilities of GenAI are made more useful for a variety of creative applications and the precision needed is more consistently realized to allow it to scale exponentially (except for a few interesting exceptions, as noted below).
What can AI do for you?
Chemical and materials companies tell us that they are finding AI is helpful in several ways. It can:
Accelerate innovation. We have worked with a chemical company that is using AI to discover new compounds faster than ever before. Already, the new AI system has helped the company discover more sustainable molecules to replace existing materials. GenAI is playing an important supporting role here, in that while traditional AI is good at finding molecules, GenAI excels at sifting and summarizing research papers, turbocharging discovery.
Improve your supply chain. Companies are finding that AI is a helpful tool in improving supply chain planning. For instance, AI can be leveraged in supply chain planning to decrease lead time and improve fidelity. Its advanced predictive capabilities can enhance demand accuracy and reduce forecast bias, increasing fulfillment effectiveness and efficiency.
Streamline your processes. Most manufacturers have myriad data about what happens within the four walls of their plants. With AI, you can optimize across these data silos, shop floor to top floor, and then use GenAI to translate the math into specific instructions to lock in the optimized parameters.
Several AI-driven projects are already underway in the chemicals & materials industry. For example, we know a US-based global specialty chemicals manufacturer that is using GenAI to develop stronger, lighter and more durable polymers. One pilot project alone helped the company discover eight more sustainable molecules.
Another US-based specialty chemicals producer has used AI to create a manufacturing portal that acts as a central repository for equipment data, reliability and quality data, and ERP data. The upshot: engineers believe they will be able to make their industrial workflows 10 times more efficient.
At the same time, a leading global Germany-based chemicals player is using GenAI to reduce the adverse impact of chemical production on the environment. Another German firm is using GenAI to enhance process control in its chemical plants, which has reduced the number of accidents in its facilities. In Japan, a chemicals manufacturer working with an IT company are using natural language processing and GenAI to generate new ideas and hypotheses for new products and more efficient services.
All these opportunities will add up. If a typical chemical company were to pursue the full range of digital opportunities now available – simultaneously boosting revenue, decreasing cost of sales, and decreasing operating costs – we estimate it could create value amounting to anywhere from 4.3 to 6.4% of revenue with 15 to 25% improvement in Total Shareholder return.
Managing the Machine
As with any innovation, of course, technology is only part of the equation. Creating the right conditions for adoption is another key variable. Three ideas can make this process easier:
Data is everything. The return on investment for data collection and preparation is often difficult to calculate. However, one thing is clear: the value of the answers AI can provide will depend largely on the quantity of quality data on which the machine has been trained. Companies are often reluctant to pay to get this high-quality data, but targeted investments make sense. You may also have more data available than you realize: don’t overlook places where a lot of data may already be accruing, such as shop floor systems (process historians, SCADA, MES) or an ERP platform.
When in doubt, test. With a cutting-edge technology, coming up with a solid ROI isn’t easy. A practical approach is to launch a small minimum viable product (MVP) to test, modify, and evolve requirements before a broader scale deployment, saving time and resources. An MVP enables industry players to identify essential goals using AI, understand AI tool functionalities, and establish a continuous feedback mechanism for fine-tuning AI deployments.
Use what works. Unlike traditional software, AI systems learn over time. In the case of a monthly sales forecast, for instance, process automation can be used to compare the actual results with the forecasts. Over time, the system will automatically begin to realize that certain forecasts are always high while others are always low, and then recalibrate your models.
