Advantest Talks Semi

The Metaverse Matrix: Transforming Semiconductors with Digital Twins

Season 3 Episode 3

Join us on "Advantest Talks Semi" as we explore the transformative impact of the industrial metaverse with Don Ong and his guest, Sanjeev Kumar.

Sanjeev Kumar is a Silicon Valley entrepreneur and advisor specializing in enterprise AI, data management, tech-bio, and industrial automation. With over seven years of investment experience, he has worked closely with early and mid-stage startups through VC firms and accelerators. Previously, Sanjeev held leadership roles at Informatica, BEA Systems, EMC, and Oracle, focusing on machine learning, data engineering, and enterprise software platforms. He also served as Managing Director of India R&D and GM for multiple products at BEA Systems and Informatica. Sanjeev holds an MS in Computer Science from Rutgers University and a BS from BITS, Pilani, India.

In this episode, we dive into how the industrial metaverse is reshaping semiconductor manufacturing, from digital twins optimizing production efficiency to real-time analytics driving smarter decision-making. Sanjeev shares insights on how AI, IoT, and advanced simulations are converging to revolutionize supply chains, predictive maintenance, and factory automation. We also discuss the role of edge computing and distributed security models in enabling seamless data exchange between physical and virtual environments.

From early digital modeling techniques to today’s persistent and immersive industrial metaverse, this conversation provides a comprehensive look at the evolution of these technologies and their future impact on semiconductor testing, design, and manufacturing.

Thanks for tuning in to "Advantest Talks Semi"!

If you enjoyed this episode, we'd love to hear from you! Please take a moment to leave a rating on Apple Podcast. Your feedback helps us improve and reach new listeners.

Don't forget to subscribe and share with your friends. We appreciate your support!

Don Ong (02:19.394)

Hello and welcome back to Advantest Talks Semi. I'm your host, Don Ong, and today we're diving into the industrial metaverse. If you ever wondered what happens when semiconductor manufacturing converges with advanced digital simulations, this episode is for you. We'll explore how virtual replicas of our physical processes can help us optimize production, test complex systems, and troubleshoot issues long before we make real-world changes.

Joining us is Sanjeev Kumar, a venture partner at Fusion Fund, a Silicon Valley-based venture capital firm that invests in cutting-edge technologies. Sanjeev's focus on innovation at the intersection of AI in manufacturing, IoT, and digital transformation. The co-op dealers of the industrial metaverse. Sanjeev, welcome to Advantest Talks Semi.

Sanjeev Kumar (03:21.048)

Thank you, Don. Happy to be here.

Don Ong (03:26.222)

Sanjeev, you've been at the forefront of digital transformation in an industrial setting. Can you tell us about your professional journey and how you became interested in the industrial metaverse?

Sanjeev Kumar (03:40.878)

Sure. My professional journey really started back at graduate school in my PhD program, which was based on data management and databases. And that's been the theme of my career for the last 35 years.

Recently, I was head of engineering for an industrial IoT software platform company called FogHorn Systems. And that's where I got into the thick of digital twins, edge computing, fog computing, and all the building blocks that are used in the industrial metaverse today.

And during the last two to three years, I've been evaluating, advising, and investing companies that are building whole systems or parts of the industrial metaverse, if you will. So that's my journey so far. And I've been in Silicon Valley and Bengaluru, India for almost all of my career of the last 35 years.

Don Ong (05:03.79)

Sanjeev, the term industrial metaverse, is gaining a lot of traction, but what does it mean in practice and why is this concept particularly relevant for manufacturing, logistics, and energy?

Sanjeev Kumar (05:16.526)

The term metaverse originated out of a science fiction novel from the early 90s. That was a place where people could immerse themselves, a virtual avatar of yourself could immerse yourself in a 3D version of the real world. This was centered more around social interaction and commerce and connecting with other people through this virtual world. And then sort of it evolved into gaming and so on. So, most of the games were based on a virtual experience that by wearing a set of goggles. You kind of took away the physical world completely and you were totally in the virtual world. More recently, this has become more important for the industrial world, which is where we've tried to model first physical model using computer-aided design to model the physical structure of things and then it evolved into product lifecycle management and then to the industrial metaverse where all of the processes in the physical world are being modeled in the virtual world. So, there's been an evolution across different decades, and we can look at some of these in detail.

Don Ong (08:10.694)

Thank you for that background, Sanjeev. The idea of an industrial metaverse can still feel abstract, especially to those who are more used to traditional fab operations. Before discussing specific applications, let's define the fundamentals. Many people still associate the metaverse with VR headsets and gaming. How do you distinguish the industrial metaverse from the consumer-facing one?

Sanjeev Kumar (08:36.174)

So metaverse clearly originated from the consumer world, where in gaming, this is where metaverse was the actual means of giving people an experience. So, the way it started in the industrial world was modeling of physical things, whether they were components or buildings or assemblies of different kinds. And this started in the early 80s with the computer-aided design. And this goes back to a company called AutoCAD in Silicon Valley. And that evolved into computer-aided manufacturing along with computer-aided design. There were companies like this is the early 90s, there were companies like Consilium, and so on. This then evolved into product lifecycle management, where it became more than just 2D and 3D modeling of the world. It was also about prediction and simulation of physical systems across different conditions. And this is where companies like Siemens and SAP and so on really started. And then, the 2010s really brought the era of digital twins, which is, you know, there were companies like GE that were popularizing the concept of digital twins, even through ads on television. There was a division of GE called Predix, which popularized this notion. And then finally we come to today, which is more the era of industrial that is more persistent, it's real-time, and it's immersive. And that sort of is how we've come to this world of digital twins and the industrial metaverse of today.

Don Ong (10:50.392)

So, the digital twin often comes up as a critical component for the industrial metaverse. So how do they tie into the industrial metaverse and why are they so essential?

Sanjeev Kumar (11:40.482)

The digital twin is the connection between the physical and the virtual world, if you will. It's a physics-based copy of machines, buildings, factories, or anything physical. And this has been, I mean, certain industries have been early adopters of digital twins, like automotive, aerospace, and manufacturing.

For example, there are prosthetic limbs that are made first virtually, which are designed to fit a particular person so that the interface is personalized to a person's condition. There are full-scale digital twins of a factory; before creating a new factory, one can first create it in the virtual world. You can optimize floor space. You can optimize the output it's going to generate in a certain time period. And then there are miniaturized or virtualized models of a city, which helps with traffic planning. It helps with carbon footprint. 

And sort of the governance of a city. And finally, you can have a full digital twin of an airplane, right down to all its assemblies, different systems, its assemblies of different systems, and down to the nuts and bolts, so that you can monitor its operational health on an ongoing basis in real time.

So digital twin becomes the context around which any analysis of the physical thing that the twin digital twin actually represents can be done.

Don Ong (14:49.434)

So IoT, AI, AR, VR, and real-time analytics are often mentioned together. So how do these technologies synergize to create a cohesive industrial metaverse environment?

Sanjeev Kumar (15:06.626)

So, you started with the digital twin of a physical component or assembly. And then you embed sensors in that assembly and gather data from the physical environment. And then that data is then decorated or integrated with the physical modeling of the digital twin to make it what's called a live digital twin. That is, now the digital twin represents what's going on in the physical environment on a real-time basis. You also decorate it with data coming off of sensors that it has. Then, if you model the process, let's say the assembly line process in the digital twin environment, you create what's called the digital thread, which is how the material is going from station to station. How is each station actually performing? How many mistakes or faults are happening there and so on. So, finally the visualization of a digital twin has become photo-realistic. It has become as close to the real world as possible.

This can be done through, so the sensor part that generates the data is where the Internet of Things really comes in. And then once you have the data, you can use machine learning and AI models to predict either its operational environment or predict its failures or even predict design changes. 

And the AR and VR part, the augmented reality part is where what you see in the virtual environment can augment. You can combine the physical environment and the virtual environment to see what exactly, to get a better understanding of exactly what's going on when things are not going as expected.

So, modeling of these physical processes makes the industrial metaverse environment more realistic. Once you're continuously evaluating, simulating, and predicting the behavior of a system, it enables you to collaborate with other people who bring different types of expertise together to solve problems when operationally things are not going as planned.

Don Ong (18:57.048)

So now that we understand the foundational concept, let's bring the industrial metaverse to life with tangible examples. So, Sanjeev, can you share some use cases where the industrial metaverse is making a meaningful impact in semiconductor manufacturing and testing?

Sanjeev Kumar (19:31.586)

I'll give you the example of a company I've been working with out of Germany called Ricoh. It's a company that, you know, it's wanting to create a digital twin or of a factory that isn't there. It's brand new. So, you can start from scratch in software. And, you know, optimize things and gradually move to the real production of parts and assembly. But what if you already have factories, which is mostly typically the case, where people already have their assembly lines that were planned on schematics on paper a long time ago? And how do you create a digital twin of assembly lines that are already there?

So, Ricoh provides a way to do a scan of an existing assembly line and then create it. So, this is done through what's called a LiDAR scanner, which creates a point cloud of the components that you see. It could be any kind of assembly line. It could be for semiconductors. It could be for an auto assembly line or anything. What they do is they actually can generate the CAD assembly from the point cloud that completes parts of the assembly that are hidden, if you will. And this requires having a database of assembly parts that are specific to a certain, you know, manufacturing of certain things specific to a certain vertical. For example, this would be different for autos, it would be different for semiconductor chips, it would be different for, you know, jet engines, it would be different for airplanes, and so on. But it's deep learning models being utilized to convert the point cloud into a CAD assembly. 

And you essentially get down to each component level, you get a virtual representation that can be manipulated. Just like if you were playing SimCity, for example. You can slice and dice it at any granularity and manipulate things. You can change the assembly flow today to see whether you can optimize your factory floor space or if you have to, the typical example these days is auto assemblies going from gas-based cars to electric vehicles. So, the existing assembly lines need to be repurposed for producing electric cars. So that's the other main reason why automakers are trying to digitize or create a virtual representation of what they already have and then see how that can be modified, that can be reused, and what needs to be reused and what needs to be created anew. For example: do this in the virtual world than actually making the changes in the physical world where you don't, you could be going down the wrong path, for example. So, all this is done much better in a digital twin setting in the industrial metaverse, if you will.

Don Ong (24:06.168)

So that is really interesting. With supply chains growing increasingly complex, how can the industrial metaverse simulate logistical challenges and help manage disruptions?

Sanjeev Kumar (24:31.97)

There are logistics providers like DHL, semiconductor supply chain planners. They use digital twins to model the entire network of part suppliers. So, for example, if there are some changes in demand, these chains could be affected by climate changes or geopolitical shifts in any way or even design changes in the product. So how would a design change affect the suppliers as well as the monitoring of the operational environment? So, I call that upstream and downstream from the actual environment. The actual place where the assembly is being operated. And how do you handle that in such a way that you minimize the delays, minimize the risk, and also control the variables up and down the supply chain? Variables and dependencies, if you will.

Don Ong (26:33.37)

Great. So how are companies leveraging immersive environments to train workers more effectively, especially in high-risk industrial settings?

Sanjeev Kumar (26:44.246)

Yeah, so whenever there is operating of complex machinery or a complex environment is concerned, where injury can happen, or equipment can go bad or something like that. So, there are companies like Siemens and Caterpillar that are using AR and VR environments to train people before they set foot on the actual floor. So, it's helpful for the safety of the workers as well as for the upkeep of the machinery. Now, this is sort of a generalization of what we saw with the flight simulators of early 90s. Microsoft had this popular PC software that you could use, but then you could use it to train as a pilot. But the actual pilot academies had sophisticated airplane-simulating environments where they could simulate even an accident, and the downing of a plane. this is a generalization of that. So, anything, any environment that involves the operation of complex machinery. First, people learn to handle them in a virtual setting before they learn to do it in the actual setting. It's good for both the people as well as the machinery. 

Don Ong (31:10.554)

Thank you, Sanjeev. But these cases don't happen in a silo. They rely on a very robust technology stack from cloud and edge computing to IoT sensors. So, for the industrial metaverse to function seamlessly, interoperability standards are crucial. So, let's explore these building blocks and frameworks that help put them together. So, edge computing is very often mentioned alongside cloud infrastructure. How do they deliver the low latency, high reliability, and secure environment needed for industrial automation? 

Sanjeev Kumar (31:54.808)

Sure. So, two trends have been going on for a very long time. One is that the compute element, the computing elements are getting smaller, and they are showing up in places where they were traditionally not found. So, for example, you have programmable logic controllers, the small PLCs that can be found on assembly lines. You have Raspberry Pi-based systems that could be on top of an elevator, or it could be a small server running on a train, or it could be a micro data center running at the base of a cell phone tower, for example. 

Or it could be like what you find in electric cars. You have a server in electric cars, and you also have a mini server rack, even on airplanes now. So, the computing footprint is getting smaller, and the second is, the sensors are getting smaller as well. So not only can the compute go away from the giant data centers, to what's called edge computing or fog computing, but we also have sensors that are getting smaller and more sophisticated. And they can help gather data in real-time. So, these are things like temperature, air pressure, velocity, and so on.

So, both things are enabling computing to happen at places where it was not imaginable earlier. For example, the bottom inside of an oil well, for instance. Nobody thought of putting a computer inside of an oil well earlier, but now you have these edge servers that can be deployed. You have GPUs that are getting smaller. NVIDIA just came out with smaller powered GPUs that can be installed and incorporated into edge servers. So, this is what is facilitating edge computing. 

That is computing away from the large data centers that provide the core of cloud computing, if you will. And then you also have at the same time, there is the emergence of what I call distributed security or distributed trust models that can help provide a secure environment in which the data from the sensors can be gathered securely, analyzed locally, close to the source of the data, and then you only upload your results or analysis up to the cloud. 

So, you decrease the frequency of the interaction with the cloud. You can do more things. You can even run machine learning models locally. And you make fewer round trips and upload less amount of data to a cloud data center. Now, if you were not doing this computing in a distributed fashion, you would have to push all the data to the cloud, which is a very expensive proposition. And the fact that we can do this, we can push computing closer to the source of the data, provide a very elegant model for handling vast amounts of information, and focus only on the things that we should care about.

Don Ong (36:23.566)

So, for the industrial metaverse to flourish, we need common protocols that people can work together. So how are these standard bodies and the industrial players driving interoperability?

Sanjeev Kumar (36:38.028)

So as the technology stack around PLM, Product Lifecycle Management, as well as Digital Twins and then Industrial metaverse, has evolved over the last 25, 30 years, so has the underlying technology stack and the standards that go around that. So, this starts with network protocols, it starts with data formats, it has to do with the open APIs, and then finally with exchange or linearization and delinearization of machine learning models that go along with the data.

So, there have been a number of different companies that have been working together on this. So, there are standards that have emerged on the network protocols like MQTT, for instance, which is a protocol common between OT environments, operational technology environments, and IT environments. There are data format standards like based on JSON, which is the lingua franca of semi-structured data. And then there are standards around exchange of models in AI, which is emerging out of the AI space, if you will.

Don Ong (38:56.635)

So, I'd like to touch a little bit on security because this is always a big topic for us. So, bringing together OT, operational technology and IT creates cybersecurity challenges. So how do we secure these digital assets and protect our IP or intellectual property?

Sanjeev Kumar (38:58.691)

There are a number of areas of governance. There's security, safety, and governance, which are critical for any computing technology to be adopted on a wide scale. The issues around security range from the network layer to the data layer, to the software layer, the data plane, control plane, if you will. And finally, the notions of governance, which revolve around auditability and non-repudiation. 

Non-repudiation is that you can without any doubt say that such an operation happened, and that it was done by a particular person at a particular time. So, this transparency around governance is provided by logging and audit trails, which has been there in all the old technologies as well, but certain OT operational technology protocols like Modbus have suffered from security vulnerabilities. Normally, what is done is that such networks or subnetworks are fenced by modern, protocols, network protocols like MQTT to prevent unexpected intrusions, if you will.

Don Ong (41:51.514)

Okay, so although technology is sophisticated, ultimately people can make or break any transformation. Sanjeev, so let's discuss how companies can prepare their teams, shape the culture, and build critical skill sets to leverage the industrial metaverse fully. So, what challenges and risks do you see in implementing the industrial metaverse and digital twins?

Sanjeev Kumar (42:19.074)

There are three main challenges, the biggest one is “lack of connectivity”. There are plenty of operational environments that lack high-speed connectivity or any connectivity to begin with. And then once you have some connectivity, the question becomes how fast is it? Because the real-time aspect of the immersive nature of the industrial metaverse is enabled by a fast network, like a 5G or a 6G network. Thankfully, places that didn't have any internet connectivity earlier can now be connected through satellite internet, or places that had slow connectivity can get fast factory scale or wide area network, fast networks. Then the other aspect is that unless you have a very fast network, you cannot operate at the millisecond level. And this is what you need to do if you are trying to use a digital twin environment to look at operational problems in real-time. So, the more real-time the setting, the faster the network requirement. So that is the first challenge. 

The second one is lack of interoperability, which is at the network level, this would be interoperability across, let's say, systems from Siemens versus let's say Honeywell or GE or different OT providers that provide both a hardware and software platform. So, if you're trying to operate across a supply chain, you need to be able to understand, you need to be able to call, you need to be able to have backend connectivity across legal boundaries of companies, you need to have data connectivity. Data standards that both the sites understand, as well as you need to understand the same domain language, the same semantics, if you will. 

The third aspect, the third challenge, is the lack of skills, which is, as manufacturing assembly lines and other operational environments digitize, you would need the people operating these environments to be more digitally savvy. Their savviness about IT needs to go up. So, for example, you could be manning a station on a semi-line and instead of just a simple console, you could have a full digital twin of the station that you can see. And you can manage multiple stations, the same person can do multiple stations, compared to before.  Simply because things are more automated now. So, you need the ability to learn operations that are adjacent to what you were doing earlier. This is because once things get digitized, you don't need that many people to babysit things.

As well as you need to understand what the software or the Metaverse product is telling you. Those are areas too, and also like when problems happen, how do you troubleshoot them, right? How do you make sense of what is being reported or how do you analyze the data that's being reported, for example? So, all of this requires new skills that existing people on the shop floor need to get trained on.

Don Ong (47:22.234)

But the industrial metaverse is not just about us. It's also connecting our suppliers, our partners, and our customers in virtual places. So how does this influence collaboration models and co-innovation efforts?

Sanjeev Kumar (47:41.088)

As I mentioned earlier, once you have multiple companies involved in a supply chain or in managing an operational environment, I'll give you an example of multiple companies monitoring the operational environment of a jet engine, of an airplane, for instance. So, these days, jet engine companies don't sell the engine to the airlines, they take on the operational responsibility of the jet engines directly. And this can only be done because you have enough numbers, the right type of sensors instrumented in, as well as monitoring of the jet engines happening in real-time. 

There's one company, there's an airline operating the plane, there's a plane manufacturer, and then plane manufacturer is then dependent on the jet engine manufacturer itself. So, you have a chain of suppliers for your assembly and then you have a chain of operators for your operational environment. So, if you were to do a design change, let's say on the plane or on the jet engine or anything like that, both the suppliers of providers of the parts that are changing as well as the operational environment that is changing would need to know what they need to do differently from before. And this requires each of the companies to be able to simulate the change virtually first. And that can only happen when the data formats, each of the companies, can understand. Firstly, they have common APIs to talk to as well, and the data that is exchanged can be understood on both sides in the same way. And so, anytime you have this connection of companies where the control flow and the data flows across legal boundaries, you need standards. You need these companies to collaborate on what each one is, what, and how each one is communicating to the other.

Don Ong (50:44.634)

Thank you, Sanjeev. So now we have covered everything from foundational concepts and real-world use cases to the technology stack, the human factors, and also organizational culture. So, let's look ahead, Sanjeev. So, what does the future pose for the industrial metaverse, especially in the semiconductor industry? And where do you see this industrial metaverse in the next five to 10 years? Are we looking at widespread adoption or is this due at its early infancy?

Sanjeev Kumar (51:21.708)

Yeah, think we're, I would say this is early infancy, but we are at early adoption, transitioning to early majority, if you will, of the customer base. And the way this is the digital twin and the industrial metaverse is evolving is that it is enabling new business models that were not feasible earlier.

So, the fact that you can have a live digital twin of a jet engine being monitored in real-time, enables the jet engine company to sell the jet engine as a service, if you will, rather than the physical product. So, they take on the responsibility of making sure that the engine operates optimally for as long as possible so that the customer is not liable for the downtime of the engine or its repair and upkeep. And that's taken up by the jet engine providers. The three major jet engine companies all do provide this option to customers, where you lease a jet engine, and you pay by the time utilized.

The airline pays by the time it is utilized. So, this is a new business model. Service, you can think of jet engine as a service, which wasn't feasible earlier. Now, this is generalizable to anything complicated, anything of significant complexity that is difficult for a customer to handle or troubleshoot or diagnose and upkeep, right? And the manufacturer, through the live digital twin, the manufacturer can take on the responsibility of making sure it works, and then the customer only pays by the time that it's been productive. So, this is one big example. Other changes in the industry are happening. 

For example, there was the acquisition of Ansys, which is a simulation software company by Synopsys. This is also driven by the fact that the semiconductor industry is evolving from just pushing, keep pushing Moore's law to how the chiplets around big chips are assembled and located. So, there is this dimensionality or 3D dimensionality of semiconductor layout as well as design, which needs to be simulated first before it's done in the real world. Similarly, Siemens acquired Altair Engineering, which is again to improve their industrial metaverse product as well as services that they can provide to their customers. So, I think what we see here is that because of these technologies around digital twin and metaverse, we are likely to see more of X as a service business model rather than outright purchase of complicated assembly machinery if you will.

Don Ong (56:24.026)

Could policies, regulations or government support accelerate or slow the adoption of industrial metaverse initiatives?

Sanjeev Kumar (56:35.982)

If you look at any industry-wide initiative by the government, so for example, the CHIPS Act by the Biden administration, anything that requires a number of companies for either semiconductor fabrication or design and assembly can drive standardization across toolchains or the tools that are used to design and build complex systems. So, the government can mandate the adoption of certain standards, adoption of certain technologies, and how this work could be done, simply because they are funding things at an industry-wide scale and in a big deep pockets way.

And that's a unique role that is very hard for anybody else to, like private sector or anything like that to actually fill.

Don Ong (58:38.404)

So, we're leaning on here. So that's all great. With Digital Twin, we're collecting data, a lot of data with Industrial metaverse. We're sharing a lot of data across, and AI is a huge topic right now. Can we do something with this? Can we do AI with all the data that we're collecting with Digital Twin and Industrial metaverse?

Sanjeev Kumar (59:01.784)

Sure, all of the digitization, which is enabled by the computer elements and the sensors and so on, and the creation of the digital twins. Digital twins are there to provide context, to the virtual environment. The real benefit of the connection between the physical world and the virtual world is gathering data about the real world into the virtual world. And once you start gathering this data, you can start creating analytics analysis of this to give you better operational visibility into how things are going. 

The next step from there is to build some prediction models around your operations, as to how frequently something should be built, how frequently something should be serviced, or what the probability of something breaking down. So, all these are very typical questions that the data you gather over time can help answer. And what are the variables that affect your operational environment, and how can you model them. 

And then beyond a forward from prediction is prescriptive solutions, which is, can you make some design changes to your assembly so that it runs more efficiently or generates more output in the same time period? And so, these are prescriptive solutions, or should you change your process in some way? So, the manual intervention done by people goes down, for instance. So, all of this is enabled by, once you start gathering the data, you can start your evolution into becoming, truly becoming like an AI-based business, if you will.

Don Ong (01:02:24.868)

Great. So, what advice would you give organizations who are just starting their industrial man-averse journey? Which first steps or priorities would you recommend?

Sanjeev Kumar (01:02:36.566)

The journey for the industrial metaverse goes from being able to observe how things are happening, to analyzing how things are happening, to predicting how things are happening. And the best way to get started with this is twofold. One is you try to get as many compute elements as possible in your operational environment.

First, you get sensors, which then talk to the compute elements. So, you get sensors in your environment that can gather data that is critical to your machinery or operation of your machinery or your business process, it starts there. That's the beginning of your nervous system. The next step in the nervous system is the compute element, which should be looking inside the operational environment. And these are things like your programmable logic controllers, your edge servers, your micro data centers, and so on. So, these compute elements act as aggregation points of the sensor data, where some initial analysis can be done, or complete analysis can be done as well as running some prediction models to help you with making the operations more efficient. 

And then at the same time, you need to create a digital twin of your environment. And now that can be done through some sort of a scanner, depending on the resolution that you're trying to aim for.

And then the scanner gives you a point cloud from which you can create a CAD assembly. And then now you feed the sensor data that you're gathering into this digital twin. And then from there, you have a digital twin in a live situation. And that gets you started with not just observing but also analyzing how things are happening.

And initially, it's your expertise amongst your staff that helps you optimize things. And then you can codify those means to optimize through rules or AI models that get you started. And then you kind of use sort of some sort of modern display technologies to analyze the CAD assembly of your digital twin, which companies like Ricoh allow you to do. So that is how you create an environment that reflects their systems as well as their business processes.

Don Ong (01:05:49.891)

Thank you for joining us on Advantest Talk Semi. Today we have explored how the industrial metaverse can transform semiconductor manufacturing from predictive maintenance to supply chain optimization and beyond. A huge thank you to Sanjeev Kumar for his insight and expertise. Thank you, Sanjeev. And until next time, this is Don signing off. 

Sanjeev Kumar (01:07:36.566)

Thank you very much.