Thank you very much for waiting. We would like to begin SoftBank Corp. earnings results presentation for the nine months ended December 31, 2024. We would like to introduce today's attendees. SoftBank Corp. President and CEO, Miyakawa. Board Director, Executive Vice President, and CFO, Fujihara. Today's presentation will be broadcast over the internet. President and CEO Miyakawa will give an overview of SoftBank consolidated financial results and business overview.
This is Miyakawa of SoftBank. Thank you so much for coming over. I would like to now explain the results for the third quarter of the fiscal year. Revenue was JPY 4,811.5 billion , up 7%. All segments increased their revenues. Enterprise, Distribution, financial segments continued their double-digit growth. Operating income was JPY 821.9 billion, an increase of 12%. Operating income also increased in all segments. Net income increased 7% to JPY 436.6 billion.
This is a summary of the consolidated financial results. As you can see, both revenues and income increased. As for the progress, we revised upward in November. As you can see, our business performance has been strong, and we believe that we will be able to achieve our goal. Here is the segment progress. All businesses are progressing well, but Financial business in particular is performing even stronger at 130%. I will now explain the Consumer business. Revenue was JPY 2,181 billion, up 3%. Mobile sales also continued to grow. This is the change in mobile sales compared to the previous year. As you can see, there has been a steady recovery since last year. We have been pursuing the Beyond Carrier strategy of expanding into areas other than telecommunications.
As shown in the overall results, we have entered a period of harvesting areas other than telecommunications, and diversification of our revenue sources is beginning to produce results. Since our core business, mobile, is once again back on a growth trajectory, I believe we have been able to transform our corporate structure to be stronger than before. The telecommunications industry has lost its vitality since the price reductions four years ago. But since each company has been undergoing a similar transformation, I urge investors to pay close attention to the future of the telecommunications industry. Operating income was JPY 440.5 billion, up 4%. Smartphone subscribers exceeded 1 million and increased by 4%. I would like to show this one once again, even though I said I am not going to do it last time.
The balance of transactions for the conversion to the SoftBank brand continued to be positive in Q3 as it was in Q2. The positive trend is now firmly established, and we expect it to remain strong for the full year. In addition to expanding our customer base, we will pursue improvement in ARPU. Next, I will explain Enterprise business. Revenue was JPY 673.6 billion, up 10%. In solutions, cloud and security performed well, increasing revenue by 27%. Operating income was JPY 140.4 billion, up 9%. Next, I will explain Media & EC business. Revenue was JPY 1,252.3 billion, up 4%. The highly profitable media domain has recovered after COVID, which has contributed to our performance. As for operating income, it was 33% increase. There was JPY 43.2 billion of one-time factors. Even excluding this, operating income was up 70%. Next, I will explain Financial business.
Revenue was JPY 203.6 billion, up 19%. Operating income was JPY 26 billion, as PayPay's profitability is now in place, showing a significant improvement. I will give you an overview of PayPay's situation. Consolidated revenue was JPY 183.7 billion, up 18%. GMV also grew steadily, increasing 23% to JPY 11.3 trillion. Consolidated EBITDA was JPY 35.3 billion, positive for the second consecutive year. Operating income has also remained stable, with three consecutive quarters in the black. We are pleased to announce a reorganization aimed at strengthening our Financial business.
Last December, PayPay announced that it would acquire the shares of PayPay Bank and make it a subsidiary of LY Corporation. Currently, LY Corporation owns 47% of PayPay Bank. Upon approval from the relevant authorities, PayPay will acquire the shares and make the company a subsidiary. After conversion of the preferred shares, PayPay will have 75% of the voting rights.
Next, PayPay announced today that it will acquire shares of PayPay Securities from us and make it a subsidiary of PayPay. Currently, PayPay holds 35% of the shares of PayPay Securities. With the transfer of PayPay Securities shares by SoftBank and only PayPay's underwriting of PayPay Securities capital increase, PayPay will hold 75% of PayPay Securities and make it a subsidiary. The purpose of this transaction is to strengthen banking and securities services under PayPay's leadership and accelerate integration of PayPay Group's various services, as well as to further enhance financial services. In the beginning, as shown on the left, the capital structure was complex, and there were multiple brands in overlapping areas, and it was difficult to understand, and some shareholders and investors pointed out that it would be better to reorganize. This reorganization will centralize the main financial services under PayPay.
PayPay will combine payment and financial services seamlessly and accelerate its growth under a simple capital relationship. Next, let me give you an update on our efforts to build a next-generation social infrastructure. We are working on building distributed AI data centers as a next-generation social infrastructure. Today, I would like to talk about progress in the sub-regional brain and the MEC or edge computing. In November last year, we announced AITRAS, an integrated solution for AI-RAN. AITRAS is a solution that embodies the concept of AI-RAN. AITRAS uses NVIDIA's Grace Hopper with Arm's Neoverse V2, which allows RAN and edge AI to run on the same platform.
The significance of providing this AI-RAN is, first, it enables autonomous improvement of network performance by having AI integrated into the base stations. Second, since the base stations have a capability of computing platform, they serve as super decentralized AI data centers. Third,
In addition to telecommunications revenue, new revenue sources can be generated by AI. Last but not the least, with its power-saving structure, it is expected to significantly reduce power consumption. Let me explain the second and third points in more detail today. We have built cell sites across Japan to create our mobile network. By turning them to AI-RAN enabled, we will evolve it into a network of decentralized AI data centers, as shown here. The AI computing platform implemented in the radio base stations will have excess computing resources during times when demand is low, as shown in the picture on the right. We will tap into these excess resources to be used for AI inference processing and services to create new revenue opportunities. Next, let me give you an update on the Brain Data Center.
In December last year, we decided to acquire the land and buildings of the Sharp Sakai Plant for JPY 100 billion to build an AI data center. We plan to deliver 150 MW by 2026, and then expand to 250 MW. As securing electric power up to 250 MW is a prerequisite, we will wait until we hear from response of power company by the end of this month before signing a final agreement. Let me talk about our idea on how the computing platform for the Sakai AI data center will be used. First, SoftBank will be responsible for providing the infrastructure such as land, buildings, and electricity. As a telecom carrier, we will also manage and operate the data center itself. On top of that, we will make investment by ourselves to install GPUs for development and delivery of homegrown AI and GPU clouds.
In addition to this, we expect third parties such as hyperscalers and the Stargate Project announced by our chairman, Masayoshi Son, the other day, to bring in their own GPUs to provide their services in the Japanese market. In such cases, we will lease the infrastructure to those parties and receive usage fees. The Sakai Data Center will serve as a model case for a new industrial cluster in Japan. The land and buildings we plan to purchase are vast, and there are other areas where other facilities can be run in addition to the data centers. For the remaining place, we want to create a testing ground where AI and existing industries merge. We want to make it a place to collaborate with all industries, including the primary ones, to address challenges such as labor shortage and productivity improvement, while continuing to verify new business models.
In the future, we hope that industrial clusters designed with the concept of such data centers will spread throughout Japan and serve as a foundation to support future industries. On February the 3rd, we made an announcement of our partnership with OpenAI. We will work on development and sales of Crystal intelligence, a cutting-edge AI capability customized for each company, by integrating its corporate systems and data.
We have agreed to establish SB OpenAI Japan, a joint venture with OpenAI, to drive the deployment of Crystal intelligence. Structure-wise, an intermediate holding company will be established, which is 51% held by SoftBank Corp. Then OpenAI will have equal stake with the holding company in SB OpenAI Japan. SB OpenAI Japan will become our consolidated subsidiary. This joint venture will exclusively sell Crystal intelligence to major Japanese enterprises. We aim to bring about corporate transformation in all industries in Japan.
SBG has already announced that it will pay $3 billion annually, or about JPY 450 billion, for OpenAI solution. Since SoftBank Corp. has struck a deal to use it on a pay-as-you-go basis, or volume-based, if you will, we want to make sure to take advantage of this opportunity. Now that the concept of Crystal intelligence, a driving force of corporate transformation, has emerged, I'd like to share my personal thoughts on the levels of AI.
This slide shows a summary of AI capability levels defined by a renowned AI scholar. AI levels are classified from level 1 to 5. Things like ChatGPT are said to be level 2, and Microsoft Copilot to level 3. Up to this level 3, AI plays the role of an assistant by giving advice to humans. Going forward, it will set to the stages of level 4 and level 5, which will transform companies.
I imagine that Crystal will initially be at this level 4 and evolve into level 5. First, we want to showcase its capability by fully utilizing it ourselves before delivering products to Japanese companies and Japanese customers with high confidence. This evolution of AI might be easier to understand if you compare it to autonomous driving. It is also presented to go through a journey from level 1 to level 5. There are different stages, starting with the one where systems like ADAS assist driving.
Then going into conditional autonomous driving, where a human driver is still required. In the future, it will head to fully autonomous driving, where even a steering wheel is not needed. Likewise, AI also started with a simple AI like RPA that can be only perform rule-based tasks, as you know. Then it has evolved into AI as a human assistant like ChatGPT and Copilot.
In the future, it will become more autonomous AI agents. We expect that Crystal, like a fully autonomous vehicle without a steering wheel, will make autonomous business operations a reality. Since the announcement of Crystal, we have received various questions about our stance on AI. So let me give you our views once again. What we are aiming for is being the best at utilizing AI and becoming the next-generation social infrastructure. We are in the midst of a major overhaul of corporate structure, yet our commitment remains unchanged. To that end, on the software side, we will continue to take a multimodal approach to generative AI, such as creating our own Japanese LLM, partnering with Microsoft, and providing Gemini to our customers. These initiatives will continue. On the hardware side, we are building the most suitable infrastructure for AI.
This partnership with OpenAI will allow us to utilize Crystal, OpenAI's latest and most advanced model, more advantageously than anybody else in the world. We would like to actively work on this, taking it as a great opportunity to help turn around Japanese AI revolution, which has fallen behind. Finally, I will talk about ESG. We have been selected for DJSI World Index for the third consecutive year. We were once again the only telecommunications carrier in Japan selected. We will continue to work to ensure that we are selected again in the future. In November last year, we were able to win the Nikkei SDGs Management Grand Prize for the second consecutive year. We have also been selected for the newly established Prime Seat, which is like Hall of Fame. We have received multiple other top-level awards.
We will continue to make efforts for sustainability and contribute to society through our business. In closing, let me summarize what I have discussed today. First, all segments achieved increased revenue and profit, and we have made progress of over 85% in both operating income and net income against our upwardly revised earnings forecast. Second, we will consolidate our financial services, including banking and securities, under PayPay to accelerate growth. Third, we are making progress in our efforts to build next-generation social infrastructure, such as AI-RAN, Sakai AI Data Center, and SB OpenAI Japan. Last but not least, I talked about recognitions received for our efforts in ESG. We will continue to strive towards our philosophy of information revolution, happiness for everyone. Thank you very much. We will now open up to a Q&A session.
Questions will be taken first from the audience on the floor, then from the Zoom press, followed by Zoom analysts, institutional investors. If you wish to ask a question via Zoom, please access Zoom via the previously announced procedure and press raise your hand button, and please wait. Once you have accessed Zoom, please turn off the live webcast to prevent howling. We would like to receive questions from as many people as possible, so please limit your questions to two per person. If you have been nominated, please state your company name and your name.
I'm Yoshi from NHK. I have a question about initiative with OpenAI. What is the meaning and the purpose of doing the business with OpenAI? You already are working on LLM with the SB Intuitions. Is this the extension of those initiatives as well? Especially, what would be the role of SoftBank? This is my first question.
The day after tomorrow, on the 12th, you can hear from the SBG's earnings presentation. But for us, we would like to try interface with various AI and how AI would change the society and the world, and we are the one to actually practice it. That's what we believe as our positioning. Therefore, the joint venture with OpenAI, the background of this joint venture establishment with OpenAI was something that was natural for us to happen. Gemini and others also have great characteristics. We would like to leverage all these good points. For enterprise customers, we would like to motivate them to step forward and then change themselves, and also open their horizon to see foreign companies. So, those companies utilizing Microsoft or Google system, then we would like to suggest what the best for them.
We will not change our stance how to face to our current customers. But speaking of Crystal, as I said, it's an autonomous AI, so now it has entered to that level, which is different from what's been available so far. Therefore, we implement that first in Japan, and as Japan and for Japanese companies, looking forward to the better future as Japan as a whole. Thank you.
One more question. This may not relevant to the earnings presentation, but I have a question around the Fuji TV incident. SoftBank also utilized Mr. Nakai, and they are now suspending its commercials. What is the reason for that, suspension of its commercials, and also what are your thoughts on the further measurement?
Well, the reason for suspending that commercial is based on what we have learned. Comprehensively, we made this decision. We would like to pay close attention and come to an appropriate decision. The reason for suspending his commercial is that because it touches the human rights things, then it is not something we can accept, therefore, we suspend it. We would like to pay close attention to what a third-party committee would decide, and then we also like to make our own decision as well.
Any other question from the floor?
Murai from Diamond. Again, question about your partnership with OpenAI. Sales of Crystal intelligence is one thing, and SoftBank will be the first customer to have Crystal intelligence. SoftBank Group said that they will pay JPY 450 billion. You mentioned that usage will be pay on volume, pay-as-you-go.
Of JPY 450 billion, how much will be borne by SoftBank Corp.? I assume details are not sorted out yet. Again, SoftBank Corp. has the biggest number of employees and the biggest user among SoftBank Group. I assume that majority of JPY 450 billion will be borne by SoftBank Corp. Talking about pay-for-use, are you going to pay the usage fee to SoftBank Group or OpenAI? I do not know how much would you pay, but if the fee is tremendous, then it is going to be a fee as opposed to investment. That fee could have an impact on the revenue. Again, I am curious about those fees and payment, et cetera.
I need to take some time to explain. I mentioned that Crystal intelligence is at level 4. It is different from traditional AI, and we are committed to solutions.
OpenAI is relatively small with about 1,000 employees. In fact, there is a competition over getting OpenAI's resources. We want to get advantageous position to use OpenAI resources more than others. I have had a lot of discussion with Masayoshi Son, and we also talked about maybe SoftBank Group companies to share burdens. Our operating income is still JPY 1 trillion. If we were to pay JPY 450 billion, or majority of JPY 450 billion, and we pay JPY 450 billion as dividend, and we get the net income of 5,000. Then if we were to pay JPY 450 billion for OpenAI solution, we might not be able to pay dividends. There is an upside around the joint venture, but no downside. That is something that we strive for or something we want to get from the joint venture.
We will never know until we try, but our arrangement is paying as pay-for-use. We need to make sure that we get the revenues in order for us to pay as you go. Nothing to lose for us, and there is only gain for us. Discussion between SBG and SoftBank Corp., which has been happening for a long time, and some people may say that. Well, actually, Masayoshi Son says that, "Do you want to live off a SoftBank Group?" But I am a CEO of SoftBank Corp. I want to make sure that transaction with OpenAI will bring us benefit, not a disadvantage at all, for our next year's earnings.
Again, please keep us updated. Pay for use, most of the JPY 450 billion will be borne or paid by SoftBank Corp., no? Whom to pay?
Well, it is still unclear in our agreement.
What is decided is, whatever will be paid by SoftBank Group will be borne by our consolidated subsidiary. In the meantime, development will be done for different companies in the group, and development fee will be paid to OpenAI. OpenAI will share the development fee, but that should not have a negative impact on us.
The second question is, I ask the question, how much would it be? Because in order for customers to get Crystal intelligence, they need to figure out how much is required. It is not going to be JPY 450 billion, but it is going to be lower than JPY 450 billion. I mean, usage fee for a customer who is going to use Crystal intelligence, how much would it be from customer's perspective?
In fact, pricing has not been determined at all. How much would it cost for development?
Whether finished product can be rolled out across the board or not, we are not sure. Maybe we should act as a guinea pig so that we can figure out how much impact the Crystal intelligence can have. Depending on how much we can expect to get, we will figure out how much cost would be most appropriate to spend. Maybe there is a group of 1,700 engineers or so in OpenAI, and again, like I said, there is a huge competition over the resources across the world. From OpenAI's perspective, of course, they want to start from enterprises that should definitely bring revenue for them, OpenAI. Even though we understand there are potential needs of Crystal intelligence, like local governments or small companies, but we want to start from big enterprises whereby we can expect revenue.
So when it is going to be deployed in SoftBank Corp.? Next year.
Next year, SoftBank is planned to deploy Crystal intelligence. When should we start delivering it to customers? As soon as possible. But again, in order for us to kickstart marketing ourselves, we are currently looking at reorganization and re-team buildings so that we can start selling Crystal intelligence as early as possible.
I am Ono of Yomiuri Newspapers. Today, there was an announcement about PayPay Securities. At the end of last year was about the PayPay Bank. You still have PayPay Insurance, but reorganization is going to continue or almost finished?
Overall, yes, we believe that it's almost finished. It's almost completed. As you said, insurance and others, internally, it's been discussed. Each one is not as big as other areas for reorganization. Therefore, in order for us to enhance our business, this is something we need to further discuss. Thank you.
I'm MJ from Bloomberg. In the presentation I heard about it, about data center project and the Stargate. How it's related? Can you elaborate on that? This is my first question.
The data center business that we are pursuing is that for Japanese companies on an inference base, so those Japanese companies can implement AI in an easier way. We have made a 10-year plan, and we are working on that. Stargate was not linked to it before. Sakai AI Data Center, when we started talking about it, and we also started discussion with Stargate as well. Right now, we have data centers being constructed at Tomakomai and Kitakyushu, and we have secured some sites to be able to secure enough power. It will be about a giga-size level building will be possible. SoftBank is concurrently doing the creation of domestic LLM.
When it comes to practical stage, then that would be inference, and it has to be much closer to actual enterprise. Therefore, we need to set up a data center so that the Japanese companies have better access to the data center. For learning, the vast investment in the U.S. and also in Japan, rather than Stargate, but OpenAI, when they deploy their business, the data held by Japanese companies, and also a company like us, special designated company, like 16 kinds of industries which own data and are not allowed to have data learned outside of Japan. There is a rule to that. We have also recently announced our initiative with Chugai Pharmaceutical, so the appropriate generative AI to be used. More and more Japanese companies are aware of what based generative AI or LLM should be used to embed their data.
Therefore, we, SoftBank, our SoftBank's original generative AI, LLM, and as for SoftBank Corp., we would like to also provide multimodal generative AI. In Japan, as for OpenAI facility, data center is also needed. Therefore, a facility and site and power to be secured, and also water for cooling. That can be managed by SoftBank current facilities, but the most expensive one, which is GPU. That GPU facility, when they were to do it in Japan, Stargate would be coming to us. We are give and take relationship. That's how we put together. As I mentioned, there's nothing for us to lose, means that data center's operation rate would go up immediately, so it has made us easier to do our operation. Thank you so much.
Right now, I can also expect there are many different types of businesses can be done with OpenAI in the future. Do you have any other areas that you are thinking of working together with OpenAI? Whether to do or not, the SoftBank Group decision. It'll be great if you could ask this question to the SBG at their earnings results presentations the day after tomorrow.
I mean. Next question, please. From the floor.
Sekiguchi from Keitai Watch. Different subject. Migration to SoftBank brand, which leads to improved ARPU and volume increase of Y!mobile, and volume increased of ahamo. Was that a reason why you decided to raise the volume for your brand? Again, my point is brand migration from Y!mobile to SoftBank. What's the current status of the brand migration from lower brand to higher brand, whether we win or lose against our competitors?
We posted increase both in revenue and income, so you can say whether we have won or not. Again, looking back, October, ahamo decided to raise volume for the brand. Since November, we have been able to catch up to ahamo.
But since then, the status is where, it was like September, before October, but there are a lot of talks about price reduction in the telecom sector. My biggest worry is our business partners or customers like small, medium-sized enterprises. I'm not sure whether the compensations have been risen or not. I don't want to raise prices easily, taking account of our customers' situation. I hope that industry will be healthier and less price competition driven. About brand migration, like a game of Go or Shogi, you may want to start from easier step, like starting from small and easy and go to more difficult, but better rewards. Pay-toku was a trigger to entice customers. I want to first make sure that we will not lose customers from our customer base.
Second, I want to make sure that as customers move to next level, then we as a company can go to next level. Thank you.
Next question from myself is about AI. SoftBank Corp. is responsible for infrastructure, you said, and for example, power, electricity, how are going to provide for? You have been utilizing solar energy since the great Tohoku earthquake, for example, and also you may have made investment overseas.
We secured 300 MW in Tomakomai and in Sakai, we want to get up to 250 MW. Several hundred Megawatt level of power is necessary to run AI or GPU. Traditional data center required 5, 10 MW, but now it's getting bigger and bigger in terms of power requirement. I have been always concerned about availability of electricity and power, and we try to also always look for where the power is available.
Of course, we don't want to cause power outage if or when we source electricity in certain area. Our data center that we are looking at is big in terms of power consumption. Renewable energy going forward is something that we continue to develop. We are open to a lot of opportunities whereby we can get power as much as possible. Going forward, maybe when a nuclear fusion, such technology, is practically available, then we might consider investing in such new technology, maybe small size. In fact, we already invested in such future energy technology, but we want to generate the power to the extent we would use by ourselves.
I'm Hase from TBS. I have a broad question regarding new Trump regime, and especially about tariff policy has been taken. For SoftBank, what do you see about this tariff policy?
Directly, there is no impact to us, but I would like to refrain from making any comment on politics. We assume that there is no direct impact due to the tariff increase. But at the beginning of IPO, we had some struggle with the U.S.-China relationship, and we started with the headwind. Now we are having the tailwind thanks to Mr. Son's action. We would like to leverage this opportunity to achieve further growth. Mr. Son's and President Trump's, as we see on the media, or you also see that their relationship and also their chemical, like each other. Maybe, yeah, I would say so.
Any other questions from the floor?
Tayama from Toyo Keizai. I have two questions. First, about partnership with OpenAI. You mentioned that you will deploy Crystal next year. When will SB OpenAI Japan start operating? You mentioned in announcement that many people would move from SoftBank Corp. to that joint venture. How much would it be in terms of resource allocation?
We are in the final stage of discussion with regards to the establishment of the joint venture. I think within the second month or so, I think we can start SB OpenAI Japan. With regards to how we can deploy Crystal, engineers are discussing to figure out how best and how most effectively Crystal can be deployed. With regards to the 1,000 engineers, not only from SoftBank Corp., but also other group companies, like 300, and we want to have all 300 group companies to deploy Crystal by themself.
Especially engineers and sales force will be a majority of people sent to the joint venture.
Next question is about DeepSeek. It's said to be cheaper than traditional AI. What's your view on DeepSeek? To building AI data centers and implementing GPUs, it sounds like your strategy will not be affected or changed by someone like DeepSeek, so I want to confirm with you.
Yes, thank you. I tried to read the paper published by DeepSeek since it emerged. Again, here is my personal view. It's not verified, but anyway. Traditionally, when relearning is done, human-made data is used for fine-tuning. Looking at the paper by DeepSeek, it seems that they use data created by AI for reinforced training. Maybe synthetic data is used for reinforced training, it seems. I think it's good at logical inference, like mathematics. From cost perspective,
it's good because you may be able to develop AI at cheaper cost.
Someone like DeepSeek will not have an impact on your strategy? I want to confirm with you.
Well, I think again, like I said, we want to get whatever we can get if it's good. I don't think that could have any impact on our delivering long-term views.
Due to the interest of time, we would like to take question from the floor, just one more.
I am Matsuda from Nikkei Newspapers. I have two questions. One is about the mobile price. The discount discussion may come, so KDDI CEO also mentioned about it. Within the industry, I believe that it's now time to increase the pricing of mobile. Do you have any measures against that? Second question is about PayPay's reorganization.
KDDI also made a bank subsidiary last year and DOCOMO as well. The competition is getting severe. What area that PayPay has to reinforce?
Let me answer to your first question. As we operate mobile phone business, what costs the most is electricity cost as OpEx. For the past years, it is increasing more and more, and it is not even JPY 10 billion, it is higher. Right now, we are trying to absorb that cost, but we are seeing the cap. Also for the employees, the salaries, we are doing better this time, and we also need to well consider about it. Also to our customers and the customers' employees, including all. We need to support as an industry as a whole.
Not only discussing about discount, also suppressing the investment in 5G, while waiting for investment in 6G, I am very sad to see this situation. We should be in a more healthier way, and we should take an action towards increasing the price along with the increase of goods of cost. But there are other carriers as well. We are the only company which sees revenue and income increase. We are now going to take an action from our end, but as an industry, we should go back to a healthy situation. We were now just, it is a really cheap nation. Also the power of development and is also weaker now. We should put ourself back to what we used. Now the PayPay. Now PayPay turns positive. I have been often asked about the timing of IPO.
Of course, we are thinking about it, but not yet, because we believe PayPay has still room to improve. As for payment, PayPay has come to the very good position and a level. As I explained before, the first, second, third layer, and we would be growing one by one. We are still on the first layer. We would like to establish a solid foundation to be able to step up to the second and the third layers, where we believe that we will be grateful to have invested. Not looking at competitors, but just looking at PayPay themselves, it is only the first station. To go higher, PayPay has to-
Have completed what PayPay can do and wants to do. That is the parent company's perspective.
That was the last question from the venue, and we would like to take question from Zoom participants. If you wish to ask us a question, please press raise your hand button. First, Mr. Sato from Nikkei Asia.
My name is Sato from Nikkei Asia. Can you hear me?
Yes.
I have two questions. If you could answer one by one, that would be great. First, about Crystal intelligence. My understanding is that since it is agent AI, it is important to nurture agent AI once it is implemented. I wonder if there will be a similar service offering in the future, what will be your strength against potential competitors?
One answer, rather than SoftBank Corp.'s strength, OpenAI's strength. That is what it matters.
So long as OpenAI's Crystal is excellent, and if there is another agent that exceeds the Crystal, OpenAI would struggle. But at the moment, I don't see anything that is superior than OpenAI. So we want to bet on OpenAI, to be honest. Not necessary strength per se, but before we started marketing iPhone, what we did was to have all employees to use iPhone and iPad and prohibit them from using paper. So we fully digitalized in our company, and let employees use iPhone to experience what iPhone could do for the corporate operations.
So likewise, Crystal, we want to start from using Crystal by ourselves. And see how the Crystal will change the operations before we approach to customers. I think that would be a source of our strength going forward. Thank you.
The next question is, maybe not to SoftBank Corp., but rather SoftBank Group.
Mr. Son's investment style, he would tend to get involved in companies that he invested in, and to roll out into different businesses by leveraging the resources of a business that he invested in, and he got heavily engaged first. Miyakawa-san, I wonder, there was some report, but I wonder the movement by Mr. Son, like Stargate or OpenAI, I wonder how much that would have impact on your decision or strategy about your AI in Japan.
I don't want to fight or have a quarrel with Mr. Son, so I don't want to answer straight to your question.
As much as possible, if you could share with us your view on Mr. Son, especially about his investment style and his investment in Stargate Project.
Mr. Son, like shown in his relationship with Mr. Trump or that announcement, once he wants to talk to somebody, he will take whatever it takes to get an appointment with the person to deliver the message by himself. I have been seeing him doing that so many times, and he is really good at negotiating and making deals. And even though he faced unexpected questions or some discussion, he could address them. While discussing with investors or engineers, he understands the essence of the communication and then look for investment opportunities. I think that's Mr. Son's style when it comes to making an investment or deciding investment. I have been working with him for several decades, and there are a lot of things that only Mr. Son can do.
While I have been looking at a lot of Japanese excellent business leaders, but Mr. Son is greater than many areas, and one of the things was exemplified in the recent deal.
Next, we would like to take questions from Zoom analysts and institutional investors. Masuno-san of Nomura Securities, please unmute and ask your questions, please.
I have two questions. Regarding the performance, I think it's pretty well, but looking at the next mid-long-term visions, and you developed your own LLM and autonomous AI has been also developed, and you are working really hard to establish the foundation of next generation AI, and where I have high expectation too. And at this time, the OpenAI scale is bigger. So you also just mentioned that it's up to OpenAI. So I have an impression that OpenAI was a little bit stronger than where you are, so being pushed.
It's not like that. OpenAI also told us that if there were opportunities, they would also like to work with us. We will continue our development of our LLM, and we are going to pursue a multi-model concept. What we found positive from the conversation between us, is that there are a lot of things that we can learn from OpenAI. What OpenAI has gone through, and they would tell us what is the better way to go, and Sam is a great business person. We would like to focus on what we have to do. Please don't feel sad.
The second question. You're going to secure 1,000 of engineers and a sales force, like your group companies, like ZOZO and LINE and other Yahoo!s and other engineers, and I think it's out of blue to them.
LINE also has 100 million users, and it means very strong customer base among your group companies, and it'll be very difficult to focus on what you have been doing. I'm a bit afraid of you putting all your efforts or allocating everything on this latest project.
Let's say a personal agent would exist on your mobile phone. That kind of world is not something about far away. In the future, there will be more things coming up. Regardless of what engine to be used, any companies can switch to the best AI to them. When I said it's up to OpenAI, it doesn't mean in a business perspective. We management people are to make right decisions. At this moment, we cannot find the great AI like OpenAI has.
Now we see a DeepSeek, but there may be much stronger AI coming up in the future. When it comes, then we should be able to make our own decision, and we'll be facing to the point where we need to make a decision. As I said, there is nothing we'll lose. That's what I feel.
I will be closely watching in the future too.
Next question is Mr. Kikuchi from SMBC Nikko Securities.
Thank you very much for taking my question. Again, Crystal, SB OpenAI Japan. I have two questions around that. Positive side is, Mr. Son mentioned at an announcement that doing in Japan and also around the globe, but the company is called SB OpenAI Japan. Is it limited to Japan and only you are responsible or you are doing business in the Japan? When it comes to global deployment in the future, will SoftBank play a key role? Then, who has the copyright of Crystal? SB OpenAI is an agent, sales agent, but that doesn't have patent or copyright. That's the first question. Is SoftBank OpenAI Japan or SB OpenAI Japan will be just a sales agent or what?
Thank you for the question. First, why it's called SoftBank AI, SB OpenAI Japan, why Japan is in the name?
I think it's good because the joint venture we established is to sell Crystal exclusively to Japanese enterprises. This is for the sake of Japanese companies. That's why the company name has Japan in there. About copyright will be held by OpenAI. In that sense, SB OpenAI Japan is a sales agent, but again, it has exclusive right to sell Crystal to Japanese companies. That's the key. Talking about exclusivity, currently, it's limited to Japan, but when it comes to a global launch, which we have been talking about for a long time, as a tool to go beyond Japan, AI is one of them, and whatever we create, whatever tool we create in Japan, we want to roll out globally. That's something that we have been talking about. When it comes to global launch, to be honest, maybe one vehicle is not enough.
To expand Crystal exponentially, SoftBank Group alone may not be able to have enough resources capacity to go global. So we will see how fast Crystal will be deployed and how much resources we could support. But since there is an OpenAI Japan, there could be OpenAI Germany or OpenAI Korea. From Son-san, as an investor's perspective, I think a global launch makes sense. But we didn't have enough time to finalize everything before signing contracts, so we'll see.
Second question about fee and cost. I'm sure there are a lot of uncertainties, but JPY 450 billion that has been talking about, forget about that for now. Miyakawa-san said that nothing to lose, but everything to gain. Does that mean that, for a single year, payment or JPY 450 billion doesn't have any impact on your financial performances? Can I believe you?
Yes, believe me. Trust me.
I am a business leader. I will not run a business in a way that something would have negative impact on the business.
So for a single year, can you confirm that JPY 450 billion thing will not have a negative impact on your business performance for a single year?
Right. Correct. Developing homegrown LLM, and we want to expand the parameter number to 1 trillion. For that, we have made a very proactive investment. In reality
From our financial capacity perspective, maybe we cannot afford number of GPUs as much as we wish. But, again, I will make sure that the initiative this time will not have negative impact on our mid, long-term plan or whatever commitment we have made in long midterm plan.
Next, Kazuki-san from Daiwa Securities. Please ask just one question. We apologize for that.
Now you are working on to build a 1 trillion parameter and recording the cost. So how are you thinking about allocating the cost to OpenAI and the GPU? And how much do you think is needed to secure these two?
So as for GPU, including what's running, we have 6,000 GPUs right now, and we have budgeted for 10,000 GPUs.
Other than that, a further purchase plan, we do not have any plan for the further purchase of GPU right now at more than 10,000 that we have as a plan. Now also for the OpenAI, there will be another investment. We have to see the cost efficiency, and then it will be determined internally. It could be the next medium and long-term business plan, management plan. For data center, how we invest in data center, at Sakai, there was already a land and buildings. We utilized the existing structure and make it into a data center. However, at Tomakomai in Hokkaido, we are now working on the land development work, and it is almost completed, and it is very large, and depending on the needs, it is going to be a container base. Depending on the needs, we are going to invest as needed.
Therefore, it does not have a big cash impact. Considering the target needs, right now, Tomakomai and Sakai would be sufficient for power supply. Yes, for 1 trillion parameter, I think will be sufficient. If many Japanese companies start using for inference, then it will require a big power. It used to take four to five years to build a data center until a year or two years ago. Right now it can be done on a container base. If we succeed in deploying a container-based data center, then we can invest as we go upstairs along with the need.
Thank you so much for your cooperation, Kazuki-san, to ask just one question. We conclude the Q&A session. We would like to close SoftBank Corp. earnings results presentation for the nine months ended December 31, 2024.
Today's presentation will be available on demand on our website later today. Thank you very much for your time today to attend the SoftBank Corp. earnings results presentation for the nine months ended December 31.