Super Micro Computer, Inc. (SMCI) Earnings Call Transcript & Summary

Ladies and gentlemen, introducing Founder, Chairman, President and CEO, Charles Liang.

Thank you. Thank you. Thank you. Good day, everyone. Thank you for this opportunity. I'm Charles Liang, Founder and CEO of Supermicro. Thank you so much for coming today. What a wonderful and amazing world today. AI changing our world so much, making it so convenient, so fancy, so accurate and so fast. From autonomous driving to store self-checkout to advancement in health care, ChatGPT, now we are entering a new world of fast growth and massive adoption for AI infrastructure. Let's brain storm for the next hour together now to see what we can do to better our business, our life and our environment. As you may know, Supermicro has experienced unprecedented business growth with AI investment in years. And we are just now starting to see the early signs of the tremendous benefits that generative AI training and inventing are bringing to our world. I guess some of you know that we were the first company to bring optimized GPU platform to the market with NVIDIA back in 2014.

Yes.

Thank you. Since then, we have been one of the leading provider of AI system, offering the most optimized solution to many of the largest AI factory in the world today. Powering embedded AI, we are changing our life for the better. A better life leads to a happier mind. Happier mind is very important. A happier mind is needed for us to do right things. Speaking of the right things now and here today, bearing a great gift for the IT industry, a gift so great that data centers all over the world will celebrate like it's their birthday. I'm here to prove that green computing can be free with big bonus. Yes, free with big bonus. Many of you may be wondering, can this be true? I will spoil your surprise and answer with resounding yes, with a really big bonus, tens of millions of dollars or even much more. Before going to the technical details, I'm happy and proud to announce some update on my personal mission of reforesting the Sahara Desert with Green Earth Foundation, sharing the same objective as Supermicro's winning computing initiative. It's my way to give back to our one and only Mother Earth. We started to propagate the most drought-resistant tree species in the Sahara Desert earlier this year as part of the Great Green Wall Project, targeting a total of 10 million trees this year and 15 million trees a year later. The foundation's Silicon Valley headquarter is also making good progress on mindful propagation of this drought-resistant tree species. On behalf of all our foundation, I welcome more of you to join this meaningful action to make our Mother Earth greener. With that said, Supermicro had ingrained green computing DNA into the design of every solution we offer. For decades, we have been the leader in designing and supplying the most optimal and efficient building product of today's server storage infrastructure. We had just launched our new optimized X14 Intel Xeon 6 Sierra Forest-based and soon, Granite Rapids-based based platforms. X14 AMD Gen 5 Turin-based solution is also ready for early deployment now. This latest high-performance chip provides better performance for [ WAN ], but at the same cost -- at a cost of writing TDP. Together with our partners, we are on a mission to build a more sustainable data center infrastructure that address the rising need for more compute and storage in today's world, especially when it came to generative AI training and inferencing. The mission here is simple. Can we improve data center efficiency, decreasing TCO or provide higher performance with your new data center infrastructure? Supporting the native AI development can be challenged through the traditional AR pooling method and especially they are expensive and environmentally taxing. My new concepts, direct liquid cooling, DLC, have been around for more than 30 years. Unlike air cooling through air condition, requiring a substantial amount of electricity to operate. DLC, liquid cooling, can use room temperature. There's a room temperature water to provide optimal cooling for server at a much lower cost and with a smaller environmental impact. Our goal is to make DLC quickly become a midstream solution for any data center and AI factory that focus on increasing efficiency and reducing OpEx. Here is what Supermicro DLC, liquid cooling solution, can do. Assuming a midsized LLM data center for GPT-4, which requires roughly 8,000 H100 GPUs or 1,000 HGX systems. Initial capital cost for an optimized DLC environment and cluster can save you up to 30%. Although 30%, that is saving. For the total cost, due to the reduction of data center space, UPS scale, generator can be smaller. Air condition scale can be much smaller, server racks other components. This can make the initial facility and system hardware acquisition cost about the same or lower compared with air cool -- traditional air cooler solution. Thus, making green computing with DLC essentially free. Basically free. You don't want to pay extra for liquid cooling. Due to lower energy consumption by less air condition, DLC saving on OpEx can be up to 40%, achieving up to $60 million over 5 years, depending on location. This is the bonus. Bonus. Free. Saving $60 million may not be a really big deal and motivation for some big corporate. But through liquid cooling, we can reduce huge amount of CO2 emissions and make our planet win doing this. Giving me a happy mind all day, every day. You can do the same, if you like, and please. Additionally, DLC computing will reduce power requirements, save for the previous mentioned typical LLM data center. Air cool will need roughly 50-megawatt of power. With DLC cooling, total, you only need 10 megawatts to power the whole data center. For city and countries where power capacity is limited, this will open the door for a new class of data centers. In the past, most of the customers were not interested in DLC because they assume, and it was true that, number one, DLC has a very long lead time, you people know, roughly 4 months to 12 months even, even in last few years. Number two, DLC was much more expensive, you people know that. DLC was not very reliable and is difficult for maintenance. Number four, DLC was unproven for data center market. Indeed, in the last 30 years, the DLC market share, liquid cooling market share, grew from 0% to about less than 1% in the last 30 years. Well, that's why Supermicro is here with you to revolutionize the industry. Together with our partner, including NVIDIA, especially, we have spent a day and night for the last 3 years addressing these customer concern. Finally, we now have improved on all of them and shipping DLC cooling racks in volume production now. First, we have the scale. We promise to deliver DLC cooling racks or data center scale solution in 2 to 4 weeks instead of traditionally DLC that you need to wait for 4 to 12 months. Second, we can build DLC solutions at a lower cost than air-cooled solutions as we proved earlier. Number three, our DLC solution is super reliable, with potentially high performance and much solid uptime. In most of the cases, maintenance cost is equal or even less than air-cooled data centers. And number four, we are targeting 15% of the global new data center development. We'll be using our DLC solutions in the next 12 months and hopefully up to 30% in the following year. Isn't this amazing?

Yes.

From 0% to less than 1%, took us 30 years. And now from less than 1% to 15%, 30% in 1 or 2 years. [Foreign Language] Now to make good friends who are honest and capable and wise. I'm telling you that DLC is the way or the future as your good friend to be successfully targeted. So many good news. Maybe we can take a break now. So with everyone excited, let me share a little bit more. Now we had just established that DLC liquid cooling solutions are a more efficient cooling authority to air condition. But it's no surprise that they require more complex integration into data center infrastructure. As a typical air-cooled data center where I can maximize our with 20 or 30 kilowatts. Supermicro DLC solution can be 80 to 100 kilowatt per rack right now. This allows new data center development to create more powerful AI system to support the latest AI application and software in the fraction of the space. Currently, we are able to ship 1,000 racks for our industry-based DLC solution provides. As of right now, our U.S.A campus is shipping up to 50 DLC racks daily to serve some of the other AI leaders. What is that logo? One of our great partner, X, is currently deploying the largest and highest performance AI LLM cluster in the world. Although DLC is still at less than 1% total market share last month, now it's different because we start shipping a few weeks ago. Although DLC is still less than 1%, right? Our goal is to help the industry to reach 30% DLC adoption in the next year or 2. From 1% to 30% take 1 or 2 years. This will create another big hockey stick growth curve. It costs the same or less than comparable air condition solution. It's also a big bonus from a TCO saving. And it reduce CO2 emissions. It will preserve billion of trees. I love that. We have more than 10,000 engineering meetings over the past year on this new DLC technology, especially appreciate our hard-working engineers and great partner. I'm so excited that DLC is finally in [ modem ] production today. DLC liquid cooling can be free with big bonus. That's the green computing partner together. For the love of AI, we are fully invested in DLC liquid cooling. AI is so powerful and changing our life. I'm sure we still have many questions about it. Will AI someday control us? One more powerful AI chips are coming? How many powerful platform will be coming soon? Or even, what's the imagined future of AI? I know but I know only some. Fortunately, we are very lucky again to invite the AI genius, our common friend. Our common friend is very busy. NVIDIA Founder and CEO, Jensen, to share his great vision with us. Jensen.

Hi, everybody. Now what?

There's -- AI is changing [Foreign Language] because of you. What's new today?

I have to admit just now, when I was coming to your keynote in the car, I fell asleep.

It's too hot.

So right now, I'm a little bit groggy. So if I say nonsense things, please, let me apologize first.

No problem.

Well, let's see. Charles, we've gone back a very long ways. And what are we doing?

I needed some water.

Okay. Okay.

I needed to speed up my energy.

Yes. They said I was on this side and you keep going on my side. This is what happens when we don't practice.

You don't need to, and you don't have time for it.

You don't need -- and so -- what were we saying? This is a very important time because we have a new agent computing coming. There are 2 things that are happening at the same time. The first is accelerated computing. Accelerated computing has arrived at a time -- oh, green computing.

Yes.

[Foreign Language] green computer.

Yes.

Okay. [Foreign Language] green computing [Foreign Language]

[Foreign Language]

I think when you say green computing, you mean energy-efficient computing, right?

Yes.

NVIDIA is energy-efficient computing.

Yes. We have some NVIDIA.

Look, green computing and green computing. All right. So accelerated computing's time has come because for a very long time, the amount of data processing has been increasing exponentially, and yet CPU scaling has slowed for many, many years. So we've been -- we have now an enormous amount of waste, wasted energy and wasted cost trapped inside the data centers. So when we accelerate the data centers, the saving is incredible because it has been so long of waste trapped. And so now we can release the waste and use that energy for a new purpose. Number one, accelerate every application, accelerate every data center, these amazing servers here, right?

So many new products.

So many new products. You have 220 new products. Unbelievable. Did he tell you that already? No wonder...

No. I've been working a bit high with it.

I came to announce Supermicro's products. And so that's the first thing. The second thing is because the energy efficiency and performance efficiency and the cost efficiency is so incredibly great with accelerated computing, a new way of doing computing has emerged, and it's called generative AI. Generative AI is an incredible thing. People say, generative AI, inference, it's related, not the same. Inference, recognizing cats, dogs, speech, inference. Generation, text generation, image generation, video generation, that's what we call a generative AI. The pressure of generative AI to -- not the pressure, but the transition to generative AI will affect every single data center in the world. We have $1 trillion worth of data centers in a world that's established; $3 trillion, probably by 2030, in another 6 years, we have to modernize all of them with these amazing systems. That's the reason why the demand is so great because all of these data centers has to be modernized. And Charles and the Supermicro team is ready to take your orders.

Isn't?

I'm your best sales guy.

Thank you.

I work on commission.

No commission. We will buy more chips from you.

Don't buy more chips. [Foreign Language]

[Foreign Language] Jensen, Supermicro is now shipping data center -- liquid cooling DLC rack in volume production now to lower the power consumption so you can manufacture more AI chips.

Yes. Yes.

Thousands of [Foreign Language]

I have many American colleagues. They don't understand my Chinese. I have many Chinese colleagues, they don't understand my Chinese. [Foreign Language]

We are shipping up to 1,000 racks per month now.

1,000.

Racks.

Multiplied by ASP?

Yes.

You're going to be a gigantic company.

Thank you. That's why I need more chips.

Did you guys all do the math? Millions times thousands times 52.

No, no, no. You charge me $2 million, more than $2 million per rack. [Foreign Language]

Are we allowed to do this on TV? Are we on TV?

I guess in the web.

Is this? [Foreign Language]

So yes, shipping about 1,000 racks now.

That's incredible. Now there's 600,000 parts. This is probably more than 600,000 parts. How many pounds?

[Foreign Language] I don't know. [Foreign Language]

I think it's 3,000 pounds, more than 3,000 pounds.

Roughly.

Yes. This is incredible. [Foreign Language] Incredible.

Yes. Our goal this year is it to ship more than 10,000 racks, actually.

Charles, this is the thing that's really amazing. People think that we're building GPUs. GPU is a chip. There are 72 chips in here. And then there are 600,000 other parts. It's -- 72 chips probably weighs 1 pound. This is 3,000 -- 2,999 other pounds. So the amount of technology that's inside one of these racks is really quite extraordinary. This is a technology marvel. The most complex, most advanced computer the world has ever made.

Yes. It's simply the best in our world now.

Yes, absolutely incredible. And the software that it takes to run this...

All ready.

Is unbelievable. Yes. Unbelievable, isn't that right?

Yes.

And so I think that people now are starting to realize that when we say GPU server, of course, the brain is the GPU. The system is much, much more complex than that. And Supermicro does amazing engineering.

Thank you.

[Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language] Okay. There's no Americans here.

This year, we are going to ship, hopefully. Next year...

When we're together, sometimes we speak Taiwanese, sometimes we speak Mandarin. And then when we disagree, we speak English.

We try to make a DLC market share from 1% to 15% this year. Save lots of power for your chips.

Yes, yes. The energy efficiency is so much better. The cost to the data center is cheaper.

Cheaper.

That's right. People don't realize this. Liquid cooled systems eliminates an enormous amount of cost in the data center so that you can use that waste, capture that waste and put it into computing. In the future, computing throughput is revenues because it's token generation, and token generation is dollars per million tokens. Just like energy, dollars per kilowatt hour. We have now invented a new commodity. This is a very important idea for all of you. This is a new commodity. It has value. And the faster you can generate it, the higher throughput, the greater utilization, the higher your revenues. It is absolutely true. And it's directly measurable. That's why this is a factory, not a data center. That's why this is a factory, not a file server. It's not a retrieval of files. It's not used for exchanging e-mails. This is directly generating revenues for factories. That's why we call it AI factories.

And so powerful and only $3 million. [Foreign Language]

Okay. So $3 million, and you can generate who knows how much revenue per year, right?

$3 million, 3,000 and every year, I have how many months? 12?

The return on large language model generation, token generation is going to be very, very good.

Yes. Very huge.

And the reason for that is because the token embeds intelligence, and the intelligence could be used in so many different industries. And so the future is very important. It's time to start up, time to start up. Throughput, utilization, all matter. So reliability has revenue implication. Throughput has revenue implication. Startup has revenue implications. That's why it's so important that we integrate the whole system into a rack scale, get all the software working, connect it to all the networking so that -- and we build all of our own data centers. We build our own supercomputers, so that we know when you install this, when you install Super Micro in your factories, the start-up time will be extremely fast. Your utilization will be extremely high, and your throughput will be extremely high because your revenues depends on it. Factory output, is measured by all of those factors, very complicated.

Yes. Yes. And all of those are direct, NVIDIA software license all certified [indiscernible].

I love the sound of that.

[indiscernible] parking the cable and it can run...

Yes. And it runs. That's right. And all of the NVIDIA NIMs, all of the large language models, it just runs on all these systems.

Yes.

[Foreign Language]

[Foreign Language] Last year, AU air cooled.

[Foreign Language]

[Foreign Language] We are shipping thousands...

Still very handsome [Foreign Language]

[Foreign Language] Yes. Yes.

[Foreign Language]

[Foreign Language] in everything. [Foreign Language]

[Foreign Language] very beautiful. Charles said that this is everything. Everything in here is NVIDIA, for all the American citizens there.

From Super Brain to edge AI. Everything. All NVIDIA software.

All green computing.

All green computing.

All green computer.

All green. All the component support.

That's fantastic. [Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language] H100, H200, B100, for you, liquid cooling. Shipping in volume amount. And this one, your B200, who are ready.

Beautiful.

For your chip.

Beautiful, beautiful.

This will be how many times faster than it needs.

So we have -- for Blackwell, Blackwell has air cooled, liquid cooled X86, Grace, NVLink 8, NVLink 2, NVLink 36, NVLink 72. So many different configurations. So that depending on the type of utilization type of use case you have, the type of data center that you have, Charles is ready to serve you immediately, right?

Immediately. It doesn't mean they achieve. One hand, you got to achieve, second hand, we ship to customer.

Thank goodness, we only need 2 hands.

In 2 weeks, in 2 weeks.

That's incredible. And all of it, software compatible. This is really...

All board certified.

Literally, everything here is software compatible.

100%.

Yes. And software, as we know, is the most complex part of high-performance computing.

Thank you for those great opening. They are all ready to service our customers.

There are 3 very important software stacks that we have in our company that everything is built on top of. The first, of course, is CUDA. It's very famous. The second for all of the networking. Because networking is just not networking. Today -- networking today is a computing fabric. Networking today is a computing fabric. The [ winner ] were sending e-mail to each other.

[indiscernible] 400, 800 megahertz -- gigahertz.

Megahertz? This is not 1980s.

[Foreign Language]

[Foreign Language] megahertz. Kilohertz.

Gigahertz.

Gigahertz. Yes. 400 gigabits per second, 800 gigabits per second, and then, of course, next generation coming to 1,600. But the important thing is all of the software that we have that runs on the networking for distributed computing is on top of 2 software stacks. One is called DOCA for the Nick; nickel, for the fabric. And it enables us to distribute the workload across the network very, very efficiently because Ethernet was not designed for high performance computing.

You make our job easier, but still very busy because you have so many great product.

My job is to help give you job. We -- and because you do such a good job, it becomes -- gives me job.

Don't forget that you...

[Foreign Language]

[Foreign Language]

Yes, yes. [Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language]

This is an incredible, incredible system. In fact, these chips are all connected together using high-speed interconnect, the world's fastest SerDes. The SerDes is incredibly fast and very energy efficient. And so we can connect this Grace CPU to dual Blackwell GPUs. And that's very important because in the training stage, the memory system of Grace could be used for checkpoint restart. Checkpoint and restarting is very important for high utilization and high uptime. And so checkpoint restart could be stored in the system memory. That system memory is very low energy, very low power. And the link between Blackwell and Grace is very, very high. Second, during inference time, as you know, there's a concept called prompts. Context -- in context training. Prompting. That prompt memory, that context memory is right here. This is the memory, the thinking memory, the working memory of AI. And so this memory needs to be very high performance, very low energy. And so during training, we have good use for Grace CPU. During inference, we have excellent use for Grace CPU. And the interconnect is very high speed very low power.

Very optimal.

And so the benefit is because we compress so many in one system. If we save 20 watts, 50 watts on the interconnect, you multiply by the whole rack, then we can take the energy and use it for computing.

Yes.

So energy efficiency translates to higher performance.

Liquid cooling, too.

That's right. Green computing.

Jen-Hsun [Foreign Language]

[Foreign Language]

[Foreign Language]

[Foreign Language] I am a Super Micro employee.

[Foreign Language]

I'm a Super Micro employee.

[Foreign Language] Will AI control us?

Of course not. We have to -- the most important thing, of course, at the moment is we have to make AI work well. Right now, AI is, of course, working extremely well. And in many applications, AI has become good enough. To good enough to become useful. It has achieved the plateau of good enough, very useful. However, we wanted to be incredibly good. We want it to be very functional. Everything from guard railing, for fine tuning, skill learning, there are many different things that we still have to improve, okay? So we know that AI still has a long ways to go. That's job #1 is to advance the technology. At the same time, we have to advance safety technology. As you know, the planes that we all flew on to come here has autopilot, and autopilot is automatic technology. In order for planes to be safe, a great deal of technology had to be invented to keep the plane safe. Also, practices to monitor the planes, air traffic control, other planes, monitor the planes, pilots monitoring each other, many different ways to keep AI, keep autopilot safe. In the future, we'll do the same thing with AI. There will be AIs that watch AIs. There are people that watch AIs. There's right guardrails that keep AI guardrailed. And so there's going to be a whole lot of different technologies we need to create for safety, technology for safety. And then third, of course, we need to have good policies for safety, good practices and good policies for safety. Talking about it is very important so that we can all remind each other that we have to do good science, good engineering, good business practice, good policy practice, good industrial practice. All of those things has to advance.

So perfect strategy. [Foreign Language] The more you buy, the more you safe.

The more you buy, the more you save. The more you buy, the more you safe.

Thank you, Jen-Hsun. [indiscernible]

Thank you. Thank you, everybody. Thank you. Thank you.

Thank you. Thank you. Thank you.

All right. Have a good show. Thank you.

So now we have talked about liquid cooling at a system level. Let's dive deeper into liquid cooling at a great scale. I would like to invite our senior [ PM Nelson Johnson NCW ] from our liquid cooling team to tell us more about it. So Nelson, can you please share with the audience the key benefit of our total patent play B2C rack-scale solution?

Absolutely, Charles. I'm very excited today because we just talked about midsized LLM data centers host 8,000 H100 GPUs or 1,000 of our HGX systems. Switching to Super Micro, direct to chip liquid cooling will be free. We mentioned that, right?

Yes.

And with a bonus. So when we're talking about compute and air cool deployments, we, Super Micro, makes it easy. The power demands of the generative AI workloads are tremendous. After all the new power saving technologies data centers are still running out of power. And we know that it can take years to make sure to build a new data center. So what's the upside here, right? By switching to direct liquid cooling from traditional air cool configurations, our customers are enjoying 33% power savings. And that's huge. All right, Charles, we mentioned this last year, right? We mentioned -- we talked about this last year, and I want to update the audience today. I have one more thing to share. I'm proud to share that we've delivered on our 2-week rapid rack delivery promise. This unprecedented speed and deployment ensures that the data centers can scale rapidly and meet the growing demand.

DLC on Super Micro. Instead of 4 months or 12 months, now this guy promise 2 weeks, in 2 weeks.

We promised together. We're in it together. So now I'm super excited to share with some more great news that now we're extending that same promise of 2 weeks to liquid cooling rack solutions delivered at scale. We are now already shipping a volume of over 50 liquid cooled racks per day. We're doing this. And we're not just dabbling in liquid cooling, we are dominating liquid cooling at Rack-Scale Solutions. Thank you.

Thank you, Nelson. [ Jen-Hsun ], what do you want to share? What's the good news?

Thank you. I am happy to share this news to show you all how Super Micro is leading the landscape of AI solution in 2024. First off, to date, Super Micro has shipped over 0.5 million NVIDIA GPUs. Now we actually did the math on that. This new computational magnitude eclipses the combined computation of today's world's top 20 supercomputers. Now allow me to demonstrate some of our cutting-edge products that was already glamorized right over here. So I'm going to actually start with this one here. This is actually built for versatility, a 32 GPU rack. Now it can support the NVIDIA H200s, like we have configured here, or you may choose to go with the AMD-MI300X or even Intel Gaudi. We make the choice available to you. In addition, we can deploy this in your data center, air-cooled environment, or if you're ready for that free and bonus with a liquid cold environment, this can do that, too. Now this, I'm about to show you, is what a deployment with 256 GPU cluster air-cooled looks like. With our latest most glamorous handsome super rack here, liquid cool density allows us to achieve deployment with half the compute racks giving you a brand new level to scale out your AI data center, supporting you all the way through to a new magnitude of deployment. Now we're not done yet. Complementing our entire solution is going to be our liquid cooling tower as well as our super cloud composer software. Our latest iteration is going to allow you in the one single panel to look at, monitor your entire data center from the chips and each one of these GPUs with sensors all the way down to our liquid cooling tower, all in one pane of glass. From -- this is Super Micro. We're advancing AI solutions. And the future of technology is here. Thank you.

One PO, one PO, home system, rack scale, DLC, water power management, all from one stop shopping. And delivery time can be 2 to 4 weeks. Good change. We changed the vehicle industry. [ CW ], please?

Thank you, Charles. Our own water liquid cooling system not only improved thermal management, but also reduced the environmental impact compared to transitional chillers. By reducing the need for the treating, we redirect the energy from cooling to computing, enhancing performance and offering significant energy savings. Our DLC solutions efficiently utilize warm water up to 45 degree Celsius to dissipate heat up to 90% of the heat generated by a rack through liquid cooling. This automized thermal management, enabling enhanced hardware performance and the computation of throughput by reducing the need for treating again, we redirect the energy from cooling to computing, enhancing performance and offering significant energy savings of up to 40%. Super Micro DLC rack solution doubled the computing density compared to air cooling. This solution supports the highest TDP GPU and redirect densities while offering heat reduce for sustainability, aligned with green data center initiatives and the ESG goals.

[Foreign Language] Last 3 years. Thank you. Thank you a lot. Thank you.

Thank you, Charles.

So thank you. Thank you very much. Looks like going for DLC and everything to gain and nothing to those. Our DLC total solution, including system, rack [ PMB ], water tower, data center monitoring, management tool, a super cloud composer, these [indiscernible] and mature total solution, helping customers speed their data center readiness. Customers of Super Micro also provide onsite deployment, maintenance and service to make it easier for customers to go for DLC center quickly. The earlier data center customers go for DLC green computing, the most saving. Again, green computing with DLC can indeed be free with a big bonus. Thank you all.

Thank you. Thank you.

Before, so many companies hates go to liquid cooling, the liquid cooling too low, too expensive, not reliable, maintenance. Now NVIDIA service helped us so much. And we are able to have you every angle, every angle. So 2 to 4 weeks. We can go for DLC together. So we recently expansion in U.S., Taiwan and Mainland for city. Super Micro has very large server and rack production capacity to supply and service worldwide data center deployments. Our main Malaysia campus will further boost our already large capacity and supported even more new business opportunity. Currently, we had a capacity to design and deliver up to 5,000 racks per month. 5,000 per month only. [Foreign Language] Being an end-to-end total solution provider, we must give some that to the client side as well, and I'll switch gear to talk about our latest edge AI and telco solution. Here is [ Maury ] to share with us. [Foreign Language]

Thank you, Charles. Thank you. Thank you. Super Micro has focused on edge computing for more than 20 years. We leverage our building block solutions to extend our design know-how from data centers to the contact edge form factors, including the 2U [ Martino ] systems, and also for the telco virtualized RAN and 2U system support data center GPU and AI at a remote retail applications. Our new exporting hyper compact systems support AI inferencing by integrating up to 3 double GPUs, this optimized configuration has been adopted to accelerate AI. In the retail for virtualization services, factionless self-checkout and also the anti safety security features, Super Micro telco and IoT/Edge systems have brought great opportunities to bring the more edge AI to more places. We have our outdoor edge systems with IP65 rating, preventing water and dust challenged, running the white temperatures, lower minus 47 degrees, up to 47 degrees as well. And this, running at the high core count CPU with a multiple enterprise level GPU. Now let's accelerate AI inferencing together. Thank you.

Thank you. This all seems too short as we have still so many things to share. Let me summarize what we have gone through today and bring this to a close or [indiscernible]. 30 years DLC from 0% to last month, still less than 1%. And now we still see technology already from our engineering team and our partner. So we hope we can grow DLC market share from less than 1%, again to 15% in 1 year, hopefully 30% in 2 years. So it should be right, it should be right. Same cost both traditional air cooler. Big free TCO bonus. A data center can save up to $60 million, for example, and reduced CO2 emission. Preserve. This is able to preserve billion more trees for our planet. Super Micro is shipping 1,000 DLC racks monthly now, and we'll be up to 2,000 maybe soon. Customers need to order now because the lead time is getting longer. We have so many back order. Green computing with a DLC can be free, like what we just proved, which is a huge bonus other than tens of -- or hundreds of millions of dollars in the savings. The biggest bonus is your happy mind, knowing you are saving our Mother Earth, the only one. Let's go for green together. Thank you so much. Thank you. Appreciate it.