For enterprises looking to shift from hardware investments to services and beyond, a change in technology and data infrastructure could be key. One approach is a focus on the “New IT,” a term coined by Lenovo, that features five elements: client, edge and cloud, network, and intelligence to meet business goals.
“The mission of the New IT is to enable and empower the intelligent transformation of various industries like manufacturing, transportation, finance, education and so on,” says Dr. Yong Rui, chief technology officer and senior vice president of Lenovo Group.
Both the current state and potential of AI promise to offer great strides in efficiency and smart technology for companies looking to accelerate transformation. For example, at Lenovo an AI-enabled production scheduling system outperformed an experienced worker during every test and even increased production efficiency. And although there’s still plenty of room for growth and improvement, AI-powered technologies are enabling better business outcomes. “AI is witnessing rapid development and is profoundly transforming how a company operates, as well as how its products, services and solutions are made and sold,” says Dr. Rui.
Beyond improving internal business solutions, using technologies like AI, edge and cloud computing, and 5G can also help businesses improve sustainability and meet environmental, social and governance (ESG) goals. For example, these emerging technologies, Dr. Rui says, can help reduce energy consumption and carbon emissions by optimizing energy usage and developing new methods to cool data centers.
Dr. Rui emphasizes the importance of being forward-looking with these technologies. From global supply chains to services and solutions to manufacturing, the technologies New IT features can be applied to a wide range of use cases today and are also continuously expanding.
“Sometimes we think the technology is far off in the future, but what we can do today is we can plan for the future,” says Dr. Rui. “We can think about how to develop this technology today so that we are prepared for the future.”
This episode of Business Lab is produced in partnership with Lenovo.
Laurel Ruma: From MIT Technology Review, I’m Laurel Ruma and this is Business Lab, the show that helps business leaders make sense of new technologies coming out of the lab and into the marketplace.
Our topic today is building the future with technology to solve big challenges like sustainability and digital transformation. Enterprises are looking to cutting-edge technology adoption to help evolve their own businesses as well as offer the best products and services to customers. How can AI and other technologies help with these lofty goals now?
Two words for you: shaping tomorrow.
My guest is Dr. Yong Rui. Dr. Rui is the chief technology officer and senior vice president of Lenovo Group. He is also a member of the Lenovo Executive Committee. Dr. Rui is a world-renowned technologist and Fellow of ACM, the American Association for the Advancement of Science, IEEE, the International Association for Pattern Recognition and the Society of Photographic Instrumentation Engineers.
This episode of Business Lab is sponsored by Lenovo.
Welcome, Dr. Rui.
Dr. Yong Rui: Great to see you Laurel.
Laurel: Lovely to have you here. So we should start by discussing Lenovo’s own digital transformation, with a shift from being a hardware company to a services and beyond company. Part of that shift is a concept Lenovo calls “New IT.” What is New IT, and how is it helping Lenovo with its own transformation?
Yong: Laurel, that’s a great question. Before talking about the concept of New IT, probably let me say a few words about the traditional IT, which people are very familiar with. Traditional IT features devices, servers, data centers, and on-prem applications. But the New IT concept proposed by Lenovo is made up of five elements: client, edge, cloud, network and intelligence. The mission of the New IT is to enable and empower the intelligent transformation of various industries like manufacturing, transportation, finance, education, and so on. As you’ve just mentioned, Lenovo is shifting from being a hardware-oriented company to an innovation-driven service-led company. This change of course is huge and the New IT technology architecture is the very engine or cornerstone to make that happen. And over the past few years we’ve made a lot of progress.
So let me talk about those five elements. On the client side, Lenovo launched its cloud PC product which is equipped with shared cloud and device computing power and storage. It is a hybrid workspace solution, specifically built for small and medium-sized businesses. And in early 2022, Lenovo created its edge computing business unit, consisting of a full range edge hardware portfolio, software computing platform called Lenovo Edge Cloud. Future learning technology enabled edge AI, and end-to-end edge solutions. That’s on the edge side.
And on cloud side, also in 2022, Lenovo established its hybrid cloud business unit focusing on technologies and products on cloud-native, AIOps and multi-cloud management. Then comes the network after client, edge, and cloud. For network, Lenovo is building core competencies on 5G and cloud network convergence technologies. We developed hardware, software, and solutions on 5G cloudified base station, 5G core network and vehicle load coordination systems. And of course, finally, the fifth and last key element of this New IT architecture is intelligence or AI. By developing and leveraging AI technologies, Lenovo is smartifying its line of products.
Laurel: So there’s certainly a lot there. How will New IT help companies with their digital transformation efforts?
Yong: Laurel, that’s correct. A lot of content there. Those five elements probably let me say more about how this New IT architecture can help Lenovo and other companies own digital transformation. First, Lenovo did not put forward the concept of New IT just for our own sake. It empowers Lenovo and also benefits various industries at large. Let me share one example. We all know product quality inspection used to be conducted by human workers, but it is actually a very tedious and time-consuming task. Also, after a few hours, human workers may get very tired and therefore can make mistakes. Fortunately, New IT is coming to help. And in fact, product quality inspection is a typical application scenario involving New IT, because it requires a high degree of collaboration between cloud, edge, and device to ensure it works well. On the cloud side, a master AI model is trained with public available data that’s on cloud side. And after being compressed, a smaller model is deployed to Edge servers at the factory.
When the model is applied to the device side, the inspection camera captures product images and recognizes defects. This is, of course, the ideal situation. In reality, there are two challenges. First, the AI model on edge may encounter new defects that it has never seen before during training. And second challenge, compared with non-defect examples the defect examples are very few, therefore it’s hard for the AI model to learn. And as we know for example, deep learning will need a large amount of examples to learn. But fortunately, Lenovo has developed a few short learning or small sample learning technologies where the pre-trained AI model will self-adapt to the new situation in the factory with very few training samples. Additionally, multiple edges can work together with the cloud to update the master model on the cloud side, based on the respective adapted edge models. So that’s a scenario and let me share a real case.
Lenovo has developed a New IT solution for a world-leading manufacturer of computer monitors. The system can identify more than 30 different types of screen defects. And what’s more, the system is self adaptive and can learn to identify and inspect new defects it has never seen before. After the system was put into operation, display defect inspection efficiency and accuracy increased by 30%, therefore greatly improving that company’s digital and intelligent transformation. This is, of course, just one example case. So far, Lenovo’s New IT technology and solutions have been deployed to almost a thousand companies in diverse industries.
Laurel: That is very impressive, because that “New IT,” made up with those five elements of client, edge, cloud, network and intelligence, can be seen infused through each layer of a product ecosystem. But how specifically is artificial intelligence helping build those new products, providing more data insights and also then transforming companies? Why should companies be excited about the current state of AI as well as what it promises?
Yong: Laurel, that’s another great question. I’m an AI person, and I always get really excited about it. So my understanding is with the development of deep learning and other technologies, AI is witnessing rapid development and is profoundly transforming how a company operates, as well as how its products, services and solutions are made and sold. So before I walk you through a manufacturing example, let’s first talk about the 2016 Go match between AlphaGo and Lee Sedol, the 18-time gold world champion from South Korea. We all remember that AlphaGo actually beat the human champion back in 2016, but what is the task for AlphaGo? It actually needs to find the best move to land a stone on the 19-by-19, which is 361 grid, not only for one move, but need to take into consideration for all the future moves. And it turned out the computation complexity of Go is much bigger than that of chess and it’s also much bigger than the number of atoms in the universe.
This actually reminds us how complex it is to find the best move in Go. So that’s Go, but let me come back to where we’re talking about.
So with that in mind, let’s turn our attention to manufacturing. In the manufacturing industry, a factory usually divides each customer’s order into a series of production tasks and then assigns them to specific production lines. This is called production, scheduling, and planning. And a dozen of complex factors need to be taken into consideration. For example, manpower, equipment, raw materials, production processes, and methods. So given these complex possible combinations, it’s actually very challenging to make a production scheduling plan that meets the multiple conditions and constraints, and at the same time, maximize the productivity through optimization of production resources available. So assigning a production task to a particular production line at a particular time is very much like the game of Go in which players need to find the best move out of the complex options we just described above, right?
Assigning a particular manufacturing order to a specific production line at a specific time, it’s very similar to find the best move in the game of Go, very similar. So if we understand the game between AlphaGo and Lee Sedol back in 2016, we can appreciate it is almost as complex as scheduling a production line. So to deal with this problem, we used deep learning and reinforcement learning to develop the Lenovo Advanced Production Scheduling system or LAPS. And we have developed and we have deployed LAPS to LCFC. And LCFC is the largest PC manufacturing facility of Lenovo. LCFC is the size of 42 standard soccer fields and has multiple campuses and dozens of production lines and receives thousands of customer orders every day. It’s really big and for every eight PCs sold in the world, one of them is built at LCFC. And LCFC produces more than 500 types of PC products from over 300,000 types of production materials.
As you can imagine, the scheduling of this factory is very complex. But the good thing is we built this LAPS system and since the LAPS system was put into use, the benefit has been significant. To illustrate its effectiveness, we held a machine versus human competition in the first few months of deployment of the LAPS system at LCFC. And if you recall, this competition is very similar to the competition of the AlphaGo and Lee Sedol, but in that case it’s a computer versus a human champion. In our case, it’s a computer versus an experienced scheduling worker. So the only difference in that, as I said, is how to have AlphaGo decide where to put the stone on the board. And in our case, the LAPS needs to decide where to put the production task.
So what we wanted to find out who will be the ultimate winner when it comes to scheduling, the LAPS system or the human worker in charge of the production scheduling? It actually may or may not surprise you, but our AI-enabled LAPS system outperformed the experienced production scheduling worker every single time. And the LCFC’s PC production volume went up by 19%, and the backlogs were down by 20%. And in addition to this increased production efficiency, the time spent on calculating the scheduling dropped significantly too. So compared with the six hours the human workers spent on production scheduling every day, it only took LAPS system several minutes to give us an answer. So that’s six hours versus a few minutes. This is just one of many examples how AI is transforming industries as we know it.
You just mentioned that the current state of AI and what it promises. So I talked about the current state, and I also want to say a few words about promises. So it is true that AI is seeing great growth and is being applied to products, services, and vertical solutions, making them smarter and more effective. But there is still a lot of work to be done before AI can unleash its bigger potential. Right now, AI is mostly data-driven and we need to feed massive amounts of data to the AI platform and model so that it can learn to recognize, say, what a cat is.
But the human baby does not learn to recognize a cat this way. The babies, they go out, see a cat or two, and then they know what a cat looks like. They don’t need a million training samples, they only need one or two. So that’s very different. So they don’t need big data to learn. Instead, small data is sufficient. So this is one of the big challenges that needs to be addressed for AI to reach the next level. So one possible route is to move from today’s data-driven AI to future hybrid AI where it integrates both the data-driven and knowledge-driven together. So that’s my thoughts about AI’s future.
Laurel: No, that’s amazing. It’s also really interesting to hear about AI in production in manufacturing and then this idea of learning AI, and like you said, a baby doesn’t learn what a cat is. And there’s clearly so much in between each of those ideas. I think particularly this concept is so complex and when we think about how AI is now changing and kind of evolving into this thing, we’re now calling the industrial metaverse, right? Which is a blend of those virtual and real-world capabilities. This isn’t science fiction anymore, this is actually happening. So what are some of those examples of extended reality or XR that we might see in those next few years?
Yong: That’s correct, Laurel. Actually, I think the reality is here, and with that said, the metaverse itself is still in its early stage of development. And different people have different definitions of what the metaverse is. From Lenovo’s perspective, the metaverse is a hybrid of physical and virtual worlds where people and objects connect and interact with each other. And the XR devices, including AR, VR, MR are the key human machine interface of and the portal to, the metaverse, with the combined and reinforced information from both words, the physical and the virtual. In the metaverse, we can provide users with more immersive and interactive experiences and solve industry challenges with higher efficiency and lower cost. So let me show you an example. So let’s take the electric power industry, for example. Historically, the inspection of power station equipment has been time-consuming and sometimes, as you can imagine, dangerous. Besides, human workers can make mistakes causing power outages and other accidents.
As such, these tasks can incur high cost for power companies. But now with the new metaverse technologies, we have the possibility to transform this industry into a safer and more efficient industry. The key is to build a metaverse that connects the virtual and the physical. And we actually thought about this a lot, and we concluded there are three ways to achieve this. The three ways are physical virtual mapping, physical virtual superimposition, and physical virtual interactivity. Again, let me use the electric power industry example to illustrate this. First, the physical virtual mapping means that we need to build a virtual version of the physical power station, which we refer to as the “meta-space.” Actually, two months ago we just finished our Tech World [conference]. I have a pretty detailed description about this meta-space. I probably won’t have time to go over that today with you, but for those in the audience who have interest, I would refer them to the Lenovo Tech World 2022 that happened in October.
They can have a more detailed scenario there. And then after this physical virtual mapping then comes the physical virtual superimposition, which means we overlay the digital information onto real objects through, for instance, AR glasses. This actually will significantly augment the capabilities of human workers, allowing them to check the status and identify more functions faster and perform maintenance tasks more efficiently. And thirdly, the human workers are not able to cover every corner of the power station, especially those hazardous areas that poses risks to health and life.
In that case, human workers can send a physical robot to do the job in their place, they can plan a path for the robot in the virtual power station. Then the robot can move in the physical power station and perform the inspection task, including recognizing equipment readings, detecting abnormal heat and monitoring equipment status in the power station. And the third way, again, we call it the physical virtual interactivity. So those are the three ways we think that we connect the virtual and physical world. And of course, above, I used the power station inspection as an example to illustrate the metaverse, but these technologies really can create huge opportunities across many other industries.
Laurel: You can really imagine that example in healthcare just being absolutely industry changing, if you were able to. Yeah, that’s really quite astounding. And I think it’s a good distinction to really define for folks what the industrial metaverse could bring us with this ability, with the XR technologies, to do things that haven’t been done before with that nice blend between what is virtual and then that physical world. So speaking of that physical world, how will adoption of technologies like these that we’ve been talking about today help sustainability and enterprise social and governance or ESG goals? And what are Lenovo’s own sustainability goals? Because as you mentioned, those enormous factories creating a number of laptops, one out of every eight in the world, that’s quite a challenge for Lenovo as well.
Yong: Laurel, thank you for asking this question. I think for any technology innovation, we should be responsible too. And you just mentioned the huge factory LCFC, right? Imagine if we can save 5% of electricity power, that’s going to save a lot of carbon emission. That can reduce a lot of carbon emission. That’s definitely something we are very committed to and we are very seriously working on. So overall, Lenovo is committed to achieve a sustainable growth by helping decarbonize the global economy.
That’s one of humanity’s greatest challenges by contributing to societal development as well as business governance. So let me go through some of those aspects. Environmentally, Lenovo is fully committed to carbon footprint reduction in its operations. After exceeding our 2020 carbon emission reduction target, we have set a vision of achieving net zero by 2050. Lenovo has been focusing on CO2 emission reduction in many areas such as in production, which we just talked about, transportation and distribution processes, and also product packaging. That’s on the environmental side.
On the social side, socially, we believe the technology must be inclusive and accessible to all, especially those most in need of the technology. And in governance, Lenovo has made it a priority to comply with laws and regulations and uphold high ethical standards everywhere we do business, including data privacy, product quality and innovation. So let me share with you a couple of examples of how technology contributes to the ESG goals. The first example I want to talk about is the warm water cooling of data centers. Lenovo’s warm water cooling technology combined with our high performance computing cluster helps data center customers become more energy efficient. The technology can improve the overall power usage effectiveness, PUE, to below 1.1. The smaller number, the better. The industry normal is probably 1.3, 1.4, but with our technology we can reduce that to 1.1, reducing energy consumption and indirect carbon emissions by more than 42%.
And specifically, the old way to cool data center rooms is to blow the hot air away by using fans. As you can imagine, this is far from efficient for the current and future HPC [high performance computing] solutions. This is where the concept of warm water cooling comes in. Actually, let me emphasize this, this is warm water cooling. We don’t need to cool the water first, because if we cool the water first, we also need to consume energy. That’s not good. So we don’t cool the water first. We just have the room temperature water come in, and the room temperature water goes through our award-winning warm water cooling system and the room temperature water removes the data center heat cleanly and quietly. And after this, the room temperature water will reach about 60 degrees celsius. So it’s become room temperature from room temperature to a kind of much higher temperature water.
And because this water is in a pipe system, the heated water now can be reused to heat nearby facilities like swimming pools and office buildings. But that’s even better. So when the water comes in, it’s room temperature when it goes out from the data center, it’s being heated. And that heated water can be used in other facilities like in the office buildings. So that’s the first example, our Lenovo warm water cooling system for data centers. The second example I want to share with you is about our low-temperature solder technology. In 2017, Lenovo pioneered an innovative low-temperature solder technology. And last year, Lenovo shipped 14.2 million laptops manufactured with this low-temperature solder processes. In total, we have shipped, since 2017, about 15 million laptops. This has resulted in a total reduction of 10,000 metric tons of CO2 emissions. Lenovo is also working to expand the use of this technology and drive benefits that extend beyond the environment, including improved reliability, efficiency, and cost, all those areas, beyond just the environment.
Since last year, Lenovo has greatly extended this technology to more and more submodule vendors. Those are the suppliers to Lenovo. Not only do we use this technology for ourselves, but also to our suppliers. And they produce parts such as SSD, wireless modules, display panels, memory, and human interface device modules. And we not only give this to our suppliers, but also share this technology to industry openly, supporting low carbon footprint transformation. So that’s the second example I want to share with you.
And the third example is the inclusive product design. Lenovo has a product diversity office that reviews and evaluates product features from diversity, equity, and inclusion perspectives. And we do this regularly for all of our products. And if you’ve used our ThinkPad before, you probably can remember for our keyboard, the F and J keys, they have these raised lines to help those visually-impaired PC users to properly align their fingers on the keyboard. So for all the functions and features in our products, we want to think about what’s the diversity, what’s the equity, and what is the inclusion we need to think through so that our product is more inclusive. So those are the three examples I want to share with you on the ESG aspects.
Laurel: No, those are all very important and I’m glad you touched on the ecosystem that really is needed for all of us to work together and keep focused on those ESG goals. It’s not something that is done alone, is it?
Yong: Wonderful. Yes, we definitely need to work with all the partners, all the industry, to make this happen.
Laurel: So although we’ve been talking about a lot of, sort of futuristic technologies, some aren’t that far off. In the next few years, how does thinking about this kind of innovation, and what is coming, really help Lenovo and its customers with their own digital transformations now and then to plan for this new technology?
Yong: Yes, when it comes to Lenovo, I think other companies as well, digital transformation, we need to be forward-looking. Probably, it is not a wise idea to only pay attention to what’s happening right now, really we need to be forward-looking. Sometimes those technologies are far off in the future, but it can also change the way we’re thinking today. So let me expand a little more on this concept. We have just talked about some examples around AI and how AI “smartifies” our manufacturing processes, our global supply chain, our services and solutions, and of course our PCs and tablets and smartphones and data centers. But let me give another example here. I have to admit this a little bit technical, but bear with me.
I think heterogeneous computing will be a very important technology for the future. If we look back at the history of computing, say for the past 60 years, we can see that in the past decades computing workload has changed dramatically.
For example, modern workloads, such as video analytics and AI model training have very different computing patterns and resource requirements than the traditional workload. Previously, computing for traditional workload is mostly done by CPU, right? We all heard about CPU, right? And the traditional workload is mostly calculated, computed by CPU. But today, even the workload has changed, as I mentioned, in things like video analytics, AI model training. Sometimes it may be more efficient and more effective to use GPU, NPU, DPU in addition to CPU. And this is called heterogeneous computing. It used to be just CPU, so it’s homogeneous. But today, in addition to CPU, we have GPU, NPU, DPU, and other PUs. And this is called heterogeneous computing, to make computing more efficient and effective.
GPU is very good at parallel computing. NPU is good at machine learning and AI, and DPU offloads data transmission from CPU, so that CPU can concentrate on computing and DPU is going to move data around. As you can imagine, given all of these different processing units, we will need a scheduling system or platform or middle layer that can efficiently make use of the heterogeneous resources and computing capabilities. And at the same time, we want the heterogeneous hardware to be transparent to developers so that they can concentrate on the problems that they try to solve, not worrying about the different low-level processing units. So we really want to have this scheduling system or middle layer to make our developers’ life easier and more effective when they are developing their application and they’re solving their own problems. This heterogeneous computing is still in early stages of development, but it will be an important technology to accelerate digital transformation in the future.
It could be applied to a wide range of application scenarios, including scientific research, industrial simulation, digital twins, smart cities, financial analysis, vehicle scheduling, new drug discovery, energy consumption and conservation, and emission reduction. So really what I’m trying to say is sometimes we think technology is far off in the future, but what we can do today is we can plan for the future. We can think about how to develop this technology today so that we are prepared for the future. And that’s all I want to talk with you today, and thank you, Laurel, for giving me this great opportunity.
Laurel: Oh, you’re welcome. And that it’s a fantastic way to end. It’s been such a great conversation with you today on the Business Lab podcast, Dr. Ray, thank you very much.
Yong: Thank you.
Laurel: That was Dr. Yong Rui, Chief Technology Officer and Senior Vice President of Lenovo Group, who I spoke with from Cambridge, Massachusetts, the home of MIT and MIT Technology Review, overlooking the Charles River.
That’s it for this episode of Business Lab. I’m your host, Laurel Ruma. I’m the Global Director of Insights, the custom publishing division of MIT Technology Review. We were founded in 1899 at the Massachusetts Institute of Technology, and you can find us in print on the web and at events each year around the world. For more information about us and the show, please check out our website at technologyreview.com.
This show is available wherever you get your podcasts. If you enjoyed this episode, we hope you’ll take a moment to rate and review us. Business Lab is a production of MIT Technology Review. This episode was produced by Giro Studios. Thanks for listening.
Learn more about Lenovo’s Global CTO Study here.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editoria