NEC Corporation (6701) Earnings Call Transcript & Summary

Good morning. I'm Nishihara, the CTO. Cutting-edge technology development driving NEC's growth and the expansion of new business domains. So together with myself, Mr. Yamada, the AI Technology Office, will present. At the beginning, the impact of technology on society is expanding, which you are familiar with. With the AI industrial evolution progressing, AI is evolving from an automation tool into an autonomous actor in economic activities. There are also risks also the cyber space, new threats, fake information also. IT vendor, stable, secure AI provision is important. We need to have a robust infrastructure is important. So NEC security platform, we will be able to provide the social implementation of the AI security and platforms will be required. There are strong technologies at NEC. AI security platform is the area. AI, we start at the core, cotomi, large language model for the industrial. For security, we are able to -- for advanced expertise, we have the basic people and the platform technology network, IT systems and new things, we will be talking about optical computing as well. We have prepared objective benchmarks. In AI, biometric authentication, video recognition, analysis response to AI, cumulative PCT international patent applications, we are #1. Also, we have been accepted as prestigious international conferences and Chinese and U.S. companies. For the last many years, we have been working with a number of papers, Japanese companies for the last 20 years. NEC is the only one ranking in these prestigious area. Earlier, in terms of personnel, CISSP, professional qualification, internationally 560 personnel, we have the top class domestically platform technology as well. We are submarine cable for the next level, technology, we already have the basic research on this. Also earlier, optical computing, the basic research, nature magazine quoted us on that. So that's the future area. In any case, our strength, we will continue to provide and deploy as BlueStellar. We are implementing cutting-edge technology, both inside and outside the company. And then the knowledge gained, we would extract, analyze and standardize and promote the large scale to our customers broadly. Today, the BlueStellar, the cutting edge technologies contribution to existing business and creation our new growth business through BlueStellar. So first, regarding BlueStellar, contributing to existing business through BlueStellar. Mr. Yamada will explain. So Mr. Yamada, please.

Good morning, everyone. NEC, I'm in charge of AI. My name is Yamada. Today, in terms of technology, how BlueStellar is being utilized from the AI business, I would like to explain. First, the progress. NEC AI, we have individually, we don't have a specific disclose. This growth is amazing. NEC business also, we are growing with higher market growth, higher than market growth. So specifically, NEC, what are we doing is what we want to talk about/through BlueStellar, we contribute. Our AI technology is BlueStellar scenario, digital transformation, how can we provide to customer were based on scenario. It's going to be built in. That's how we provide the AI technology. We will provide through the AI agent, how we can transform business process is the perspective that we have for our technology. Also, supporting that is the AI platform technology. Why are we different from others? NEC is focusing our 2, how we can provide governance. The other customization, data pipeline to customization, how can we do that? These are the 3 points that I will be talking about with you. So starting with the agent, transforming business processes. May, our CEO, Mr. Morita talking about the AI agent will be strengthened bespoke at that time. This is this diagram from that presentation. Agent is a big process automation, not just as a tool. This is how we can automate. We are releasing sequentially marketing, manufacturing, we will have each of those areas, including procurement. So marketing. In May, we announced already best move is the title. What we do, marketing, what kind of things do we need to do, AI can give suggestions. And when we implement what's the benefit we can do a simulation. Marketer can go back and forth and make the definition. It's already cutting over at customer side and 125% for advertising retail is the awareness rate that we have been able to achieve. Also yesterday, we announced the procurement negotiation agent as part of supply chain management. This area already for the last 10 years, we have technologically developed and have the proof of concept. So starting this month, we will deliver to a customer using AI, procurement negotiations happen automatically, but took a few days, negotiating could be done in a matter of minutes. And then the provider and the supplier demand, supplier can get a benefit on a win-win situation. Also number three, [ sorry ] is in English, NEC document automation for proposal is supporting sales. Sales proposal documents to customer, how we can generate that. AI can create it for you. Using this material and communication should be done with the customer. What are the areas that need to be focused, we can make recommendations. And today, there are 2 announcements to make. First, cotomi Act is the name of the project. This is a digital project general automation tool. Recently, there are various systems using browser as interface and provided in the form of service. By embedding the tools in the browser, how customers use tools, that is learned autonomously. And as the organizational tool, this can be automatically deployed and executed. We have demonstration to show you today. Another topic is Command Center Agent. That is the name we use. This is an emergency call work that we will support in various countries, Command Center receives all the emergency calls, but the customers are in the state of panic when they make the calls. So how can we appropriately extract the information and take quick action. That will be the key. So from incomplete conversation, the correct situation understanding is necessary. And actions using the resources will be suggested. That is the agent service. And this is supported by the platform technology. I would like to talk about the governance issue and customization issue. First, on the governance issue, in what way can we use AI in a correct manner. There are various elements. So as a concept, we are providing end-to-end consulting service to be specific in what way should customers use AI and what other risks that need to be considered, and what kind of actions should be implemented and how that can be embedded in the organization. We will provide end-to-end consulting. Yesterday, we introduced 3 consulting menus. And at the same time, consulting alone is not sufficient by using cutting-edge technology, we need to protect them. I prepared 2 concrete projects. One is to anonymize the sensitive information, how to protect in order to avoid information leakage through a network, what do we need to do? That is anonymization. Another is the concept of guard rail. In addition to global standard guardrails, we need guardrails that matches Japanese culture and matches the rules of organizations. And in AI service, Cotomi, we incorporate them, and this is operated on the customer side. Another topic is cybersecurity, AI agent itself will autonomously operate the system. How to protect this is a very important point. Usually in cyber security, entry is detected and actions are taken and recoveries are made. But in AI, we are improving this process. And also in SiOC, we maximize the use of NEC's intelligence, how to collect intelligence, how to customize and deliver to customers and also the results of the analysis, looking from the customers' management perspective, what are the impacts? We do the simulation and present them. That is the kind of service we provide. Third point is customization. AI model, there are various developments, but so long as general model is used, it does not contribute to competitive strength of individual companies because peers also use the same. So the important point is the customization incorporate industry knowledge, the company's unique knowledge. That process is extremely important. We have track record of building LLM. So we have strength in customization. This is one example with Tohoku University. This is a medical ALM. And this customization service is the kind of service that we would like to continue to deliver to customers. We cannot achieve this by ourselves. We are working in collaboration with these companies. For example, we are collaborating with Cisco in the area of governance, with Google Cloud in the area of AI agent platforms, we are collaborating with Red Hat, the system-related AI model, we are collaborating in this area. So far was the explanation on the use of AI technology in the form of AI agent and what are the key technologies. So security, new customization we would like to talk about -- Nishihara will talk about these 2 areas by using the keyword harness and capsid knowledge.

I would like to talk about the future. It's not a way far into the future. But this is something that can be commercialized in the next midterm plan period, how can we achieve such a speed in commercialization that I would like to cover as well. AI security and platform. Let me start from AI. AI will continuously evolve. And how can we utilize and leverage the evolving AI. This is an easy example in Japan, how horses were utilized. Well, I would like to use this as a comparison. Horses until 5,000 BC, it was just a strong wild power in the prairie, but it was not contributing to human society. But then horse harness was developed. And thanks to the development of horse harness, they are now contributing to cultivating farmland and also other areas in the human society. Same applies to AI. This is a very intelligent tool. Horsepower is very strong. If you have a processor, you can copy. But different than humans, it's a different media. So [indiscernible], a leading authority in AI research says the interactions within a company that will be shut down and that its successor model will be deployed on another server through AI detection. It overwrote its own copy on to successor model. It was to deceive the operations to conceal the copy is a nightmare. AI ASI, once it's released and used them properly, there is a possibility that it may run out of control. From this stage, kind of highness we should have. NEC has begun research and development of technologies for this, what we need. So allowing humans to inspect and supervise AI. One is transparency. Humans can understand AI's decision-making and intent, then plus AI's future actions. Also, in terms of verification based on human requirements, AI verifies its own decision and guarantees that they are grounded in facts. The other AI behaves in accordance with human ethics, social norms and organizational goals. They can look at the target of the company. They can put that into the priority. Last, controllability is most important so that humans hold the final control function. These 4 pillars, if we can have the technology available, there is a rapid spreading AI that will become usable in human society such as horses became usable within human society. This is something we have to do. We have this specific for each phase. In generative AI, we prepared offerings. On the left-hand side, the blue are the products. It's Agentic AI era today, as Yamada-san said. So we are the AI Asian platforms. We have the platform. So all the necessary technologies are assembled. Next, ASI. We are preparing offerings aligned. So long as we have this, NEC want to introduce this and provide this to our customers. So we need to have a safe and secure situation, introducing this, giving competitiveness to companies. This should be looked at with data. One is the cyberspace digital data. This is what each company already possesses. They already have that. Manuals, the way they do things and then legal documents. Each company, the strength, there's other things. This is the non-explicit data on people and organization. Each individual organization people. This is culture, and this is the knowledge. This needs to connect to the AI model. Once it does, people can copy, which can become a power. The third is real-world on-site data, endpoint with customer. Each company, oral data with new data generated daily basis. For instance, the digital data and cyber space. You need to have is raw data for AI. AI model based on the values, we call it knowledge. So even if you just have raw data, it doesn't mean anything. What fits is top data product. Quite a long time ago, we spun out this product using this. We have this business. If we have it outside, it's a very big size that is a big business for us. All these technology assets that we have, AI-ready necessary data extracting is very good. This is the knowledge basis inside knowledge. Then AI model can be created. This part becomes very important for us. Next, the nonexclusive data area. 90% is nonexplicit. It's nonexplicit. So it's not available, not text and then is not obvious. So what are non-explicit? Middle, implicit means you can make it into text, but we have not done it. Same with yourselves, I'm sure. When you do things, when you make decision or when you make judgment, there are things you have to do. But you don't do it with any sense, you have some rules or grounds or -- but it's not put into text. Embodied, this is what the people can do with the way our [indiscernible] physical expertise. What is emotional and what they are thinking could be different. So that is the nonexplicit. This part also, we think we can put it into AI model. So we are preparing. As Mr. Yamada said, cotomi product was explained, there is a [ demulator ]. So please look at that. What you do office work, IT services, I'm sure you use it. In terms of the expense reporting, there are many things that IT systems are utilized [indiscernible] browser. It's quite complicated and you need expertise. Not a manual, but it's automatically AI detects and does it for you. Then 1 week later, it's like you have a great secretary assisted. So this is also a nonexplicit. If you want a step further, what we're trying to do, the way we think, the rules you have yourself could be made nonexplicit. GPT, when you start talking and do things as you go along, the repeated conversation is building. How did you make that decision or the workflow is also used in decision-making. And this is a better way of thinking. It's like you're working to a subordinate. This is used as a workflow store, digitalized so you can copy it. This knowledge can be shared within the company. This is an R&D phase today. We can put it into our product very, very soon. Next, real-world and on-site data. The real-world and on-site data, we have new data coming in every day. So in advance, you cannot learn it. So therefore, it's important to incrementally train AI with new data real time. This is R&D also. Also, the real world is diverse. So social infrastructure, office, retailing, important is the sensing, the information that sense, how could be safely put onto the network. After transfer to network and then to cloud and to in simulation technology, we have a lot of systems that we have experienced. All technologies are available that we can integrate. This is very useful as well. One example later in the demo, you can look at this as well. We have biometric authentication technology. This is the first of its kind. You don't have to stop by a stress free. So you can look at the retina both eyes all at one time. It's 6 kilos. So 3-meter from before, you can look at it. So in terms of authentication, so 100 million people can be done with this. So these sorts of intensive, something that we have to do for the integration. These are things that can be do a walk through authentication. This is also implementable. AI industry, in terms of the competitiveness, cyber, we want to have the digital into the AI ready data knowledge is how we would extract. Without this, there's some point also right-hand side. In terms of the nonexplicit data on people and organization to be extracted. This is what we are doing with authentication by customer, NEC can provide the secure data that we receive and then cut it out. This combination, there could be a lot of competitiveness and acceleration of knowledge creation. [indiscernible], doctor, there is the knowledge creation cycle that he has mentioned. So the nonexplicit data can be automatically integrated for the next step. This is a direction that they want to evolve. Next is on the security, IT system attack. For this, we have announced a product called SiOC, and that is demonstrated. So please have a look. AI agent, the attack to AI agent, this is done through partnerships on technology and using harness, the cognitive usage of the people or human perception. Let me talk about this topic. At WEF, the risk of this information has been listed as the highest short-term risk for 2 consecutive years at the time of disaster. And also at the election, this is considered as a very high risk. The checking of the generation of our content or understanding of the content and analysis of such, the tool to analyze was announced in August. As NEC has a patent is #1 in patenting on the image recognition, multimodal and digital technology is our strength. And those technologies are all included. As the next step, so now there's our content. What is the intention of this content? The narrative intelligence is another area we are doing R&D. I have introduced on the topic of AI. The point is various industries will transform using AI. And at that time, system integrator and service relationship will change. AI services generate new data and NEC understands fully the AI and convert them to new services and develop new service. And then another new data is generated. So it's not completely opening the integration, but we will circulate this cycle. And for our customers by customer, we leverage advanced trends as well. That is also an important point. On the platform technologies, I would like to cover this topic as well. Right now, the traffic is growing by 40% annually due to AI, we will need bigger traffic in the future. And you see it's one of the top 3 companies, and we succeeded for the first time in the world in the multicore transmission system, 12 core is possible. So we can increase the bandwidth nearly 12x and this was successful. And next is the repeater. If it is -- so it has to be reduced in size, and we also succeeded optical sign processing, that is a very difficult technology. We have been working on LSI, but we succeeded in the use of GPU and do this in software. So the speed is fast and cost is low. So towards next phase, the submarine cable or technology is already ready. Another is the quantum cryptography. The -- theoretically, it is considered impossible to crack, but NEC has been tackling this. We announced in July that we realized this on optical fiber. When there's a usual signals being transmitted in the same table, the optical transmission is also possible. And there's also Ethernet cable and there's a B2C, which is also secure. So that system is also provided. The optical technology, digital computing, AI content demand, what will become necessary, that is electric power. The power plant is to be placed at the location of data center and neural network and the matrix vector, we succeeded in doing this using optical technology and energy cost will be improved. And in there is a benchmark, the publication called Nature 4 are adopted by nature. And this is unprecedented in this industry for registration. This is quite significant. My message is that R&D and product, how can we link them quickly? That is the important point. As Yamada-san explained, various technologies were developed through R&D center. R&D and business unit collaboration is very strong internally. So now after 3 months of developing technology under R&D, we can commercialize. In the past, it took 1 to 2 years. Of course, it depends on the area, but AI and security is our strength and also the speed will become NEC's strength. On the growth business, let me introduce. This is the last topic on the intellectual property in the current midterm fund until FY '25 as intellectual property business are licensed, we have the several JPY 10 billion profit. And there are competitors but partners and we are to receive the license. But we also own licenses like design and other intellectual property. And by combining them, we are to create a new license. This is called the technology license. We have strength in technology, but deployment and mass production can be done by partner. And then expand globally. We already have success cases. So we intend to double through these measures. This is AI drug development businesses. Left-hand side is personalized cancer vaccine. This is in the clinical trial phase. We achieved extension of median recurrence-free status from 24 months to 30 months. We collaborate with the company called Transgene. And thanks to this, they succeeded in a capital increase. So the design technology of the vaccine, thanks to that, the valuation was raised. And also the basic research, we are collaborating with Taiho Pharmaceutical to develop shared cancer vaccine. We are analyzing whole genome to identify common cancer antigen candidates. Next, this is the advanced technology consulting services that we started from last year. The NEC's R&D staff will join the project on a hands-on basis and then add AI technology in the downstream as well, we provide DX strategy consultant and data scientists. So with this whole package, we provide solution to customers. With Sumitomo Rubber, we made an announcement the other day. And we will provide this expertise to other players within the same industry as well as other industries in the AI industrial revolution era, NEC will lead the creation of social value. Thank you very much.