From predictive to prescriptive – Two semiconductor CEOs walk into a coffee bar: Part 2. 

Revolutionizing semiconductor manufacturing with smart capabilities, advanced analytics, proactive processes and AI for secure, sustainable production.

The uncommon conversation over coffee that began in Part 1 continues…

CEO 2: Did you put out that fire? 

CEO 1: We did, fortunately. My team’s pretty smart. But what we haven’t solved yet is how to predict them to prevent them. 

CEO 2: But you do have some ideas? 

CEO 1: We think we’re onto something. 

CEO 2: So, there’s hope? 

CEO 1: We think so. Back to our digital twin for a minute. We talked about using it to compare runs, so you can see where the process is trending to help predict a problem before it occurs. 

CEO 2: That could be helpful (takes a sip).

CEO 1: Yes, but we must go beyond predicting failures to use the digital twin to prescribe the most effective preventive actions. 

CEO 2: Now that would be an improvement. 

CEO 1: With a digital twin you can think about your process in new ways – you can explore different options until you have modeled an optimized way based on the data….  

CEO 2: Collecting data is still key? 

CEO 1: Of course. (takes a sip) …but it’s still not real thinking. To have real thinking you need real-time data. Without it, you’re just working with past data and past performance. 

CEO 2: True. So, you need to connect all the devices, processes and systems in your fab to collect all the data from your current runs and feed it back into your digital twin for analysis? 

CEO 1: Yes, and the digital twin puts that real-time data into operational context. We can feed real-time manufacturing data back into virtual models to identify what are the most effective actions to prescribe. 

CEO 2: So, based on the data, you will know what to prescribe? 

CEO 1: Yes, we’ll know what actions we need to prescribe to prevent problems – like changes needed within our manufacturing processes.

CEO 2: OK. That sounds promising. What else? 

CEO 1: We can learn how to manage fab utilization rates better to drive profitability. Or how to make better capital expenditure decisions based on comparing historical and forecasted data.  

CEO 2: Will it equip my fab to evolve? 

CEO 1: Good question. Yes, it could, if you use a step-by-step process. With smart capabilities, you can optimize your manufacturing based on where your process is right now. You can also look at its current status and see how to make your next run better for higher yields. 

CEO 2: Ahh…continuous improvement. Our business can’t survive without it. 

CEO 1: With a digital twin, you can make knowledge-building runs virtually – without any real world production disruptions. Then you can weave your virtual learnings into future production. 

CEO 2: So it needs to think before it can act? 

CEO 1: Don’t we all?  But what makes it smart is the way we integrate simulation, real-time data, advanced reporting and analytics with execution control to go from prescriptive to preventive. 

CEO 2: What do you mean, exactly, when you say, “from prescriptive to preventive?” 

CEO 1: To be truly smart, the system needs to think for itself in real time to recognize developing situations to predict outcomes before they happen. Next, the system needs to become prescriptive, to automatically suggest the right improvements at the right time for the right results to take the actions needed to prevent problems – before they occur. 

CEO 2: Preventive is smart.  Especially when you prevent quality problems.  

CEO 1: Correct. The way I see it, when smart manufacturing is ready for full production, we can predict production issues early enough so that we can be prescriptive with real-time performance insights about how and where to make preventive adjustments. So negative events simply never occur. 

CEO 2: What kind of operational adjustments are you talking about?  

CEO 1: Adjustments to our MES, to adopt AI, and to become fully preventive to mitigate risk, to conduct preventive maintenance, achieve first-time right products, increase NPI yields and grow our business. 

CEO 2: That’s a lot. Welcome to the future. Before it happens.

CEO 1: It is a lot. But once you see how smart manufacturing brings together all the processes and systems, while optimizing future outcomes based upon the best historical models, best practices, and real-time data, you begin to realize just how smart it is. And how smart it will become as it evolves. 

CEO 2: So, it’s smart enough to keep learning? 

CEO 1: Absolutely. 

CEO 2: So, your virtual model could dynamically inform and optimize the production environment for both monitoring and predictive AI functions? 

CEO 1: Of course, and with AI, we’ll get smarter and faster across every aspect of our industry and our supply chain, from chip design and verification to manufacturing and automation, to enhance efficiency, to improve quality, and accelerate innovation. 

CEO 2: So, once you have your new optimized fab model, can you share it with your other fabs in other locations? 

CEO 1: Of course, I’d want to share the optimized model with all our manufacturing sites that are running the same processes. 

CEO 2: Sounds challenging. Who is going to bring all of these capabilities together in your fab? 

CEO 1: I’ve thought about it. We can’t just focus on the individual pieces. We need a holistic perspective. We need a vision. It’s time to talk to Siemens. They’re the only ones who have all the smart manufacturing pieces, the track record and the vision for the complete digital transformation we really need.  

CEO 2: Sounds like change is on the horizon. 

CEO 1: Not only change – transformation. 

CEO 2: I will not want to miss this… 

CEO 1: There’s a lot to talk about. Coffee next week? 

CEO 2: Sure. I’ve got to hear this. It’s on me. 

For more insights about smart manufacturing for semiconductors, visit www.siemens.com/us/en/industries/semiconductors/smart-manufacturing.html