Hello, ladies and gentlemen. Thank you for standing by for GDS Holdings Limited's Q3 2020 earnings conference call. At this time, all participants are in a listen-only mode. After management's prepared remarks, there will be a question and answer session. Today's conference call is being recorded. I would now turn the call over to your host, Ms. Laura Chen, Head of Investor Relations for the company. Please go ahead, Laura.
Thank you. Hello, everyone. Welcome to the 3Q '20 earnings conference call of GDS Holdings Limited. The company's results were issued via Newswire services earlier today, and are posted online. A summary presentation, which we will refer to during this conference call, can be viewed and downloaded from our IR website at investorsgdsservices.com. Leading today's call is Mr. William Huang, GDS Founder, Chairman, and CEO, who will provide an overview of our business strategy and performance. Mr. Dan Newman, GDS CFO, will then review the financial and operating results. Ms. Jamie Khoo, our COO, is also available to answer questions. Before we continue, please note that today's discussion will contain forward-looking statements made under the Safe Harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Forward-looking statements involve inherent risks and uncertainties. As such, the company's results may be materially different from those expressed today.
Further information regarding these and other risks and uncertainties is included in the company's prospectus as filed with the U.S. SEC. The company does not assume any obligation to update any forward-looking statements, except as required under applicable law. Please also note that today's earnings press release and this conference call include discussions of unaudited GAAP financial information, as well as unaudited non-GAAP financial measures. GDS press release contains a reconciliation of the unaudited non-GAAP measures to the unaudited most directly comparable GAAP measures. I will now turn the call over to GDS Founder, Chairman, and CEO, Mr. William Huang. Please go ahead, William.
Hello, everyone. This is William. Thank you for joining us today's call. Two weeks ago, our shares started trading on the Hong Kong Stock Exchange, opening a new chapter for GDS. I'm pleased to say that we achieved all of our objectives for the Hong Kong IPO and the secondary listing. We attracted a high level of demand from new investors, including strong participation from China. We significantly strengthened our equity base with over $1.8 billion U.S. of net proceeds, adding to our competitive advantage in terms of financing. We established a liquid market in our shares in Hong Kong to complement trading on Nasdaq. Last but not least, we raised our profile and the reputation among customers, government, partners, and investors, which help us to attract more business and secure more resources, further strengthening our market-leading position. Meanwhile, our business continued to perform strongly.
During the Q3 of 2020, we added nearly 24,000 sq m or 46 MW of new customer commitments. During the first nine months of 2020, our organic net add was over 72,500 sq m, more than we did in the whole of last year. In order to maintain our sales momentum, we added over 100,000 sq m to our secured development pipeline, including greenfield land for a major new edge-of-town campus near Shenzhen. We grew revenue by 43% and adjusted EBITDA by 48% year-on-year, and our EBITDA margin remained at 47%. At the beginning of this year, we targeted 100,000 sq m of annual net add, consisting of 80,000 sq m organic, plus 20,000 sq m from the Beijing 10, 11, 12 acquisition, which was pending closing.
Today, we are clearly on track to achieve around 98,000 sq m of organic sales, plus 27,000 sq m from M&A, including nearly 8,000 sq m from a new acquisition in Shanghai during the current quarter. This gives us over 120,000 sq m net add for 2020, representing 20% outperformance. Looking forward to FY 2021, we believe that the current level of organic sales is sustainable. Furthermore, with the potential acquisition of Beijing 14, which we announced in September, we already have visibility for another 2,000 sq m of M&A next year. As you can see from our disclosure, there is a consistent pattern of 3-5 hyperscale orders every quarter. Each of these orders represents a major development by one or other of our top customers. These developments are often planned in multiple phases to be delivered over several years. This gives us a strong foundation for our future sales.
Turning to slide seven. Over the past few years, we have made a lot of progress in developing and meaningful relationships with fast-growing new customers. We had exciting breakthroughs, which we announced the last quarter with ByteDance and PDD. We are actively pursuing multiple hyperscale opportunities with these two names. China's internet sector continues to produce very large companies, which emerge very quickly. Typically, they begin by using the major public clouds. However, as they develop, they also start to deploy their own private clouds, and they require also data center services. This is where we see incremental opportunity. Given the extensive presence of cloud service providers in our data centers and the multi-market platform, we have a unique value proposition in hybrid cloud. We are also starting to see financial institutions deploy hybrid clouds. We have a well-established customer base of over 230 financial institutions.
As they evolve their IT architecture from the mainframe to server, we anticipate significant incremental opportunity. Over the past five quarters, we have stepped up our construction program from 84,000 square meter to 135,000 square meter of active developments. At the same time, our pre-commitment rate have remained solidly over 60%. We have talked repeatedly about why a secure development pipeline is a critical success factor. During 3Q20, we added over 30,000 square meter to our area held for future developments. During the Q4, we have added another 70,000 square meter. We now have a total of over 400,000 square meter secured, and we will continue to add aggressively over the next few quarters. It really is that important. Turning to slide 11.
Following the success of our edge-of-town strategy in Beijing and Shanghai, we made a very significant move to set up a new edge-of-town campus in the Greater Bay Area. We have acquired from the local government a greenfield site in Huidong, around 24 km from the edge of Shenzhen. It is one of the key areas identified for data center development in the new infrastructure plan recently published by Guangdong Province. Once fully developed, the site will yield a net floor area of approximately 72,000 sq m according to the initial design. We established our edge of town campus in Langfang, near Beijing, in Q2 '19. In less than 18 months, we have initiated nine data center projects and secured over 67,000 sq m of customer commitments. In Q3 '20, we initiated our Langfang 9 data center, which contributed a 25 MW new order.
We also added to our land bank with the acquisition of Langfang Land Side Street, and further land acquisitions are in progress. On the edge of Shanghai, we have two locations, Kunshan and Changshu. During Q3 2020, we leased three adjacent buildings in Kunshan with a total developable net floor area of 16,000 sq m. The first of these buildings, which we call the Kunshan 4, is undergoing conversion. Turning to the slide 14. We announced two M&A deals during Q3 2020, which added to our capacity in the urban area of Beijing. Today, we are announcing a new acquisition in urban Shanghai, which we call the Shanghai 19. It has total developable net floor area of 12,800 sq m, of which 7,900 sq m is complete and fully committed. The total acquisition and development cost, including cost to complete, is RMB 778 million. It's an attractive acquisition with good terms.
We will continue to pursue this kind of the opportunity, which adds to our presence in key downtown locations. With that, I will hand over to Dan for the financial and operating review. Thank you.
Thank you, William. Starting on slide 18, where we strip out the contribution from equipment sales and the effect of FX changes. In Q3 '20, our service revenue grew by 14.1%. Underlying adjusted gross profit, which we previously called net operating income, grew by 12.7%, and underlying adjusted EBITDA grew by 12.5% quarter-on-quarter. The Beijing 10, 11, and 12 acquisition closed on June 5th, 2020. Excluding the contributions from this acquisition in both the second and Q3s, our service revenue grew by around 8%, and our adjusted gross profit grew by around 7% quarter-on-quarter. Turning to slide 19. Service revenue growth is driven mainly by delivery of the committed backlog. Net additional area utilized during Q3 '20 was 16,589 sq m, continuing the recovery we saw in Q2 '20. We expect similar levels of move-in in the Q4. MSR was RMB 2,519 per sq m per month.
Without Beijing 10,11,12, it would have been slightly higher at 2,533 RMB. Turning to slide 20. Now, adjusted EBITDA margin remained around 47%, and we're still benefiting to a small extent from government concessions. Turning to slide 23. Up to the end of September, our CapEx year to date stood at 6.6 billion RMB. In Q4 '20, we expect another 3.4 billion RMB of CapEx, including acquisition consideration for Beijing 9, Beijing 13, and the new Shanghai 19 deal, and payment for the Langfang and Huidong land. This is what gets us to our 10 billion RMB CapEx guidance. Up to the end of September, we had paid cumulatively 640 million RMB of CapEx for managed BTS build to suit projects that we intend transferring to GIC. During Q4 '20, we expect to transfer the first project by way of selling 90% of the equity of the project company.
The proceeds will revert a small part of the CapEx, while the assets and liabilities of the project company will be deconsolidated. Looking at our financing position on Slide 24. Pro forma for the cash proceeds of the Hong Kong IPO, we have RMB 18.7 billion of cash on our balance sheet, and our net debt to EBITDA ratio is 1.2 times. Given our ongoing levels of organic and inorganic CapEx, and assuming the Beijing 14 acquisition closes in Q1 2021, this ratio will go back up to around four times within a couple of quarters. As in the past, we will allocate and reserve cash to capitalize new investments in our business plan, and will then leverage those investments to ensure an efficient overall cost of capital.
Given William's comments about the sales outlook and the importance of adding to our land bank, and our ability to keep on generating attractive acquisitions like our recent deals for Beijing 13, Beijing 14, and Shanghai 19, we will not be short of opportunities. While we are not changing our financing approach, we will strive to enhance our debt capital structure. We have an aggressive plan to refinance a substantial portion of our existing onshore RMB-denominated project debt to achieve longer tenors and lower cost. We have multiple refinancings going on right now with 10-15-year terms, back-ended repayment profiles, and all-in costs of below 5% based on the current loan prime rate. Finishing on slide 26 with our revised guidance. For the full year of 2020, we are raising the bottom end of our original guidance for both revenue and adjusted EBITDA and keeping the high end unchanged.
We now expect total revenue to be in the range of RMB 5.7 billion-5.75 billion. Adjusted EBITDA to be in the range of RMB 2.66 billion-2.67 billion. The updated estimates imply an increase of 38.3%-39.5% year-on-year in total revenues, and 45.8%-46.4% year-on-year in adjusted EBITDA. Our CapEx guidance of RMB 10 billion for the full year remains unchanged. We'd now like to open the call to questions. Operator?
Thank you. Ladies and gentlemen, we will now begin the question and answer session. If you wish to ask a question, please press star one on your telephone and wait for your name to be announced. If you wish to cancel your request, please press the pound or hash key. For the benefit of all participants on today's call, please limit yourself to one question. Your first question on queue comes from the line of Jon Atkin from RBC. Please ask your question, Jon. Your line is now open.
Thank you very much. I'm interested on the commercial front, what are you seeing in terms of customer demand on the hyperscale side? Are they wanting to maybe think about building more of their own capacity going forward? If so, does that affect you at all? Are the internet companies that up till now have been pursuing a cloud-first strategy, is there any trend that you're seeing where they're beginning to shift their IT onto their own equipment and perhaps signing more frequent deals with data center companies such as yours? Thank you.
Jon, this is William. The first question is the customer build by themselves.
Yeah.
Far we didn't see any change right now. I think especially in a Tier 1 market, which we focus, we didn't see any change in this. The second question is?
Internet companies.
Yeah
data center companies to host their own private clouds.
Yeah. There's very significant changes. I see a lot of the larger internet company, their IT became more complicated. They used to use the cloud when they grew up, during their grow-up period. Now it looks like more and more they use the hybrid cloud strategy. That's why we are very excited for the future demand on this portion.
Thank you. Maybe just a brief one for Dan, as we think about the drivers for 2021, what you're seeing in terms of your development pipeline, what you're seeing in terms of your late-stage sales pipeline. What are the factors to keep in mind as we think about top line growth, and then what the EBITDA contribution might be for 2021?
Thank you, Jon. William made some comments today to set initial expectations for our sales organically and potentially inorganically next year. I think this is higher than what we've said previously. If you're talking about revenue and EBITDA next year, of course that lags the sales by at least one year to two years. What we can look at there as a kind of indicator is how much new capacity is coming into service, because the capacity which we have under construction is 60%-70% pre-leased, pre-committed. As new capacity comes into service, that means that there's potential to deliver significant incremental backlog. In 2020, new capacity coming into service is being pushed to the Q4. You'll see there's a significant amount of new capacity coming to service in the Q4 of this year. We laid out the delivery schedule in the earnings presentation.
You see there's over 50,000 sq m of new capacity coming into service in the H1 of next year. Typically, after new data centers come into service, not much happens for the Q1. It takes a couple of quarters before the move-in starts to build up. I think that this quantum of new capacity coming into service will increase what we call the backlog related to area in service. That will flow through to a further step up in the quarterly net addition to area utilized, which drives revenue. If you talk about EBITDA, then it becomes a matter of the MSR trends and the margin, which I'm sure some of your colleagues are going to ask me about in a moment, but maybe I'll answer now. I think the MSR is going to continue to decline very slightly as it has been doing.
Over a period of time, it seems to average out around 1% per quarter, and we've discussed the factors behind that. It's not indicative of what's happening to returns. In fact, it may be contrary to what's happening to returns. EBITDA margin has increased significantly this year. It's still slightly elevated due to the benefit we receive from government concessions. I think next year, the target to achieve maybe another one percentage point increase in EBITDA margin over and above what we achieved this year. I say target because we also have, in our plan, quite a number of corporate initiatives around smart data centers, around ESG, around renewable energy, branding, corporate communications, and so on. There will be some increased costs associated with that. In due course, there will also be increased return from that expenditure.
I'm just a little cautious on margins next year because we do plan to increase our capital expenditure.
Thank you very much.
Your next question in queue comes from the line of Colby Synesael from Cowen. Colby, please ask your question. Your line is now open.
Okay, great. Thank you. A few, if I may. $1.8 billion in proceeds from the Hong Kong listing, obviously quite a bit. Has this changed the company's view on market expansion? Are you intending or focusing on going into more markets in China? Perhaps with the new proceeds or just given where you're seeing demand go. Also, as part of that, you're interested in going outside of China with new builds. My second question, just quickly, you mentioned the opportunity to refinance debt. You have a variety of different refis in process. Can you give us an idea of what the total interest expense savings you think you might be able to achieve when all is said and done? Thank you.
Colby, I answer the first question. I think definitely, we raised $1.8 billion net proceeds. I think this is to show our view, our ambition in next five years. I think we see a tremendous demand inside China and also overseas like Southeast Asia and Hong Kong. I think even in China, we believe there's more new market will come in, become more important, except our previous four key markets. I think we do have the plan to expand our business, to still strengthen our business, get more market share in the current Tier 1 market. Also, we do have the very solid plan to move to some new important market in China. In the meanwhile, as we talked during our roadshow Hong Kong IPO, we do see a lot of the solid demand from our install base. It's in Southeast Asia expansion.
We have activated development those business plan. That's why I think we are quite excited at this moment to look forward to the next couple of years.
William, do you think that we could see you going outside of China in 2021?
It's too early to say. We will see. We activate development plan and evaluate all the demand and evaluating all the business partners in Southeast Asia. I think maybe in the near future, maybe the beginning of the next year, we will tell you guys what exactly the plan is. I don't want to say it's too early to say, but I can tell you we activate to develop the plan right now.
Hi, Colby. If we just isolate out the refinancing part of what we're doing, I would say that the cash interest saving is several hundred million RMB per annum. In terms of how it gets reflected in our accounts, when we refinance, in some cases, we have to write off an amortized portion of front-end fees that we incurred on the initial financing. It may not become immediately apparent in the H1 of next year that we have reduced the effective interest rate. If you give us a few quarters, I think it will become apparent later next year. In addition to refinancing, there's the incremental financing, which I believe we'll be able to do on these better terms. This is a significant improvement on what we've been doing in the past. It's a continuous improvement.
It's not that long ago that we were doing project financing at an all-in cost of around 7%. The facilities we've got in execution right now all work out around 5% or even less.
Got it. Thank you very much.
Your next question comes from the line of Gokul Hariharan from J.P. Morgan. Gokul, please ask your question.
Yeah, hi. Thanks for taking my questions. First of all, now that we are developing a bigger and bigger edge of town portfolio in at least the three main locations, could we talk a little bit about how is the demand looking at the city center, the older data centers, versus what you are developing at edge of town? Are we seeing any kind of differences in terms of demand, be it power density or other requirements? Second is, I think 2020 seems to be the year where a lot more of the CapEx is being spent on inorganic. As we think about the next couple of years, how should we think about organic versus inorganic? Do you feel that inorganic is going to be a bigger portion of the capital outlay, as we think about the next couple of years and some of the opportunities that you've talked about?
I think the first one is demand for the downtown data centers.
Yeah. I think the very direct answer is the demand from the downtown, let's say edge of town, right?
More central.
More central. Still maintain very strongly, yeah. I think that's why we keep acquiring the edge of town data center, when we develop the edge of town campus. This strategy will continue. We will build our large scale, hyper scale campus in the edge of the town. In the meanwhile, we will still continue to build data center and acquire the data center resource in the urban town. The demand very clearly very strong.
Yeah, I just read-
Yeah, go ahead.
Sorry, go ahead. Oh, no. It's okay.
Go ahead. There is no difference in demand. That is all.
Difference is urban, I think.
Well, it's the same customers.
Yeah.
A typical architecture is the edge of town site is a hub, which will then be connected to a number of downtown sites. Downtown is the edge. I know we call it edge of town. In terms of the customers' architecture, downtown is the edge and the edge of town is the hub. Downtown, as you know, is really challenging. We have to be creative to create new resource and we've done a number of acquisitions, in fact, to give us large quantities of new supply in Beijing and now Shanghai. It would be easy to abandon downtown, because it is so challenging and time-consuming, but then we would be failing our customers. It's part of the value that we provide to our customers, that we have a total solution, both the edge of town and the downtown, and in all markets.
Your question about organic versus inorganic, and William made a comment, that currently his run rate for organic, it's going to lead us to around 95,000 sq m. I know William said 98,000 sq m.
Yeah
It wasn't Freudian. 95,000 is where the math comes out. William said that next year we can say already that that kind of level is sustainable, and we have visibility to that because we have a certain quantum of multi-phase orders which already give us high certainty of business next year and even the year after. The inorganic part is mainly driven by our desire to create more capacity downtown. In each market, there are a number of independent project developers. There always have been, there continue to be, and we watch them. The Shanghai 19 acquisition, which we're announcing today, is a case in point. It's a project that we've been aware of in Shanghai. We reached a stage of completion where we were able to move forward and make the acquisition.
As I say, we have been aware of this project and a number of others from inception. That's why we have a pipeline. That's why I believe that we can sustain M&A at the kind of current levels as well. Beijing 14 is not a done deal yet. We haven't entered into definitive sale and purchase agreement. If that goes ahead, as we plan, then it gives us 20,000 sq m of M&A for next year already and more than 12 months to do some more deals.
Got it. Understood. Thank you.
Once again, if you wish to ask a question, please press star one on your telephone and wait for your name to be announced. For the benefit of all participants on today's call, please limit yourself to one question. If you have additional questions, you can re-enter the queue. Your next question on queue comes from the line of Tina Yu from Goldman Sachs. Tina, please go ahead and ask your question.
Hi, management. Thank you for your time. If I may just ask one question, then it would be, I think, William, you mentioned in your presentation that you've been pursuing multiple hyperscale opportunities with ByteDance and Pinduoduo. I was just thinking if you could provide more colors on that. What location these projects may be, is it edge of town, downtown, or built to suit? The sort of expected timeline that these can be decided and come online and also maybe in three to five years, what kind of revenue contribution do you see from these two customers? Thank you.
I think number one, we start to build our business relationship with these two new logo. It starts from last quarter. I think given the time, we do expect build a more stronger relationship with them. I think according to their current position in China, our resource platform, our data center platform in China, which is quite unique and will create a lot of value for those kind of customer in the future. I think those customers, they have demand urban town plus edge of town hyperscale offices, right? I think we are well prepared for echo their future demand already. I think we are confident we will have developed more business with them. It's too early to talk to some account, because we are committed with those kind of account very closely.
I think I'm very confident we have new demand from them in the near future.
Thank you.
Your next question comes from the line of Frank Louthan from Raymond James. Frank, please go ahead and ask your question.
Great. Thank you. I wanted to just ask a more general question, talking about demand and how things may have changed a bit, and in particular, the digitization trends that you would've seen, say, a year ago. Now that the dust has settled a little bit, can you characterize what sort of the pace of those trends are with end users in China now versus what they were and how much that may have accelerated?
Yeah. Frank's asking difference we've seen in digitalization trends. You go ahead.
Yeah. Frank, William asked me to respond. It's hard to isolate the difference that going through the COVID periods has made. Clearly, not just in China, but around the whole world, we were all required to adapt and adapt to new technology so that we could work from home and do everything from home, do IT from home. That's a structural shift. Our dialogue with our customers and new business is based on at least say their three-year forecast. What happened doesn't get reflected in some short-term change. Clearly, our new business has got a higher level, and I think that doesn't just reflect our success, but it reflects there's more opportunity in the market. Definitely now some of this is being driven by 5G. Particularly, say, IoT and smart city-type applications, both for cloud, large internet, and also for enterprise customers.
Do you want to add?
Frank, I think the trend is not only driven by the COVID, right? I think the whole logic is China, the digitalization is overworking to evolve. I think the cloud, as we mentioned again and again, cloud is still in the early stage. We will see in the next five years, cloud players still will be the major key driver to drive the data center demand. In the meanwhile, as I just mentioned, I think a lot of the new large Internet companies too introduced in China, like a Baidu, like a Bilibili, like a Pinduoduo, right? If you look at last five years, they are nothing, right? Now they became a capital company, more than capital company. I think they can still get back and penetrate into different segment, penetrate different vertical industry.
I think there's still a big space for the internet company to grow. I think and now a lot of internet companies start to use not just using a public cloud. They adopt to a hybrid cloud architecture. This will trigger a lot of cloud demand, plus a lot of data center colo or hyperscale data center demand. I think we are very confident. In the meanwhile, I just mentioned this. Like the traditional financial institution plus enterprise. They also have started to transfer their architecture from the traditional architecture to the cloud base, hybrid cloud base. This will create another wave of the demand for data center. In my view, there's three drivers, cloud, internet, enterprise. In the next five or even 10 years will be the key driver to drive the demand.
That's why we raised such big money that we try to catch up, echo this wave, right? I would add one more. As Dan mentioned, 5G is just an implement right now. We believe in up to two years, 5G will trigger more IoT stuff, will trigger more new applications, and also will be another potential key driver to drive the data center demand.
What are some of the key applications you think that come out of 5G? What are you seeing right now?
So far I think it's too early to talk about it, but you always see a lot the IoT stuff. We talked to a lot of the traditional industry. They all talk about the IoT stuff, and I think they are very clear 5G will drive the new application to implement to all the supply, all the value chain business, including the traditional manufacturing and traditional retail, traditional industry. I think the idea is not very clear now, but the market talk about a lot of the development right now.
All right, great. Thank you very much.
Your next question comes from the line of James Wang from UBS. James, your line is now open.
Good morning, management. Congratulations on a good result. First question from me is, I remember in the Q2 result, Dan mentioned there were a few locations that experienced some delays in activation of power supply. I'm just wondering whether these are resolved now. Maybe a broader question on this is, as the number of projects and the construction grows, is it getting more difficult to execute with the same level of precision as the past? There may be more slippage in capacity delivery over time? My second question is about the Huidong project. I'm just wondering whether power supply has already been secured for that project. What customers you have in mind for that location.
Also we're hearing, for example, there's potential overbuilding parts of Guangdong, so perhaps a bit more color on the demand and supply situation in that particular region as well. Thank you.
Okay. Thanks, James. The reference I made was actually just to a couple of sites. We can control, to a large degree, the construction, but the provision of the power infrastructure and the activation of power depend on supplier and sometimes there's some small delay. That was fixed, I think by the end of September. It may happen from time to time, but it adds up to a few months matters, particularly when we have delivery schedules for customers. Life can't be perfect, right? The degree to which this affects us is pretty small. The second comment, though, is way opposite. I think we have the largest data center construction program in the world. I try to benchmark it against some other very well-known large-cap global players. It looks like we have almost double the amount of capacity under construction. Operationally, execution is difficult.
There are lots of challenges, lots of complicating factors. I'd like to stress this as you brought it up, because I think generally maybe analysts and investors underestimate this when they talk about competition and people's plans. It's easy to say it. It's not easy to do it. We've been scaling up for 10 years. We started with three data center projects, and we went through 10 years of increments. Now we have around 20 projects, more or less permanently under construction. I think we've shown that we can handle this sufficiently and keep execution issues down to a minimum. The last question about Huidong. We bought the land from the government.
That means that the deal was to acquire the land together with the government's commitment of the, call it carbon quota, the right to use the power capacity and to coordinate with the State Grid to provision the power supply. The land is located in an industrial park. It's actually called a data center park. It's an established location from that respect. Overbuild, oversupply. We've always heard that, right? Meanwhile our commitment rate is 95% and our pre-commitment rate is 60%-70%, which you won't find any data center company in the world which has the level of commitments and pre-commitments. A site like this is actually very close, we said, to the edge of Shenzhen, right? It will be first in queue. Very attractive to hyperscale cloud. We'll announce an anchor customer in one or two quarters.
Right. Thank you. Thanks, Dan.
Your next question in queue comes from the line of Edison Lee from Jefferies. Edison, please go ahead and ask your question.
Hi. Good morning, management. Thank you very much for taking my question. I guess the number 1 is about your comment on the growth in hybrid cloud demand. I suppose that the hybrid cloud demand is going to come more from enterprises rather than large-scale internet companies. Does it mean that your proportion of retail in the business is going to go up? I know that you guys don't want to try to distinguish between wholesale and retail. I'm just trying to get a sense of the breakdown of the demand coming from large internet giants and traditional enterprises, and how that's going to impact your pricing, and also your utilization rate going forward.
I think the demand profile, as I just mentioned a bit, and then the key demand still will be the cloud service provider in the short term. We do see the internet company, since they adopted their architecture to the hybrid cloud. I think the demand will be more fast, the growth more fast. In terms of the enterprise customer, as everybody know, the enterprise customer, their sales time is longer than the other internet company and cloud player by now. We do see the change of their architecture. We see some very solid demand from some of our large financial institutions. I think, number one, this is the trend. Number two, I think in the short term the demand still will be number one, cloud, number two, internet company, and number three is enterprise customer.
In terms of the new enterprise customer's demand development, I can tell you that four years ago, we had 350 customer names, right? Now we have almost 700 year-to-date, right? Almost 700. That means we have doubled our customer number. The number is mainly driven by enterprise customer. We never stopped to get a new customer in an enterprise area. We will still continue to get that. The good thing is they are now traditional enterprise. They start to use the hybrid cloud. We think we will get the benefit from next few years. Yeah, right.
Yeah. I think Edison, you're right. We use hybrid cloud in reference to large internet companies, which maybe sounds pretty unusual.
Yeah.
That's what we're seeing. It reflects the range of applications and data that they have, the range of their requirements. For example, video streaming might rely more on, say, CDN, for which they will use a public cloud. There could be other business areas, other different applications, which requires a large concentration of the servers. In which case they put that on a private cloud. Then there's traffic migrating between the two. I think it is right to talk about hybrid cloud as reference to opportunity with large internet customers. We haven't provided any targets. I hesitate to do so, on a three to five-year basis we do aim to increase the proportion of core retail, financial institution, and enterprise from where it is today. That's despite what we expect to be the continued very high volume growth from cloud internet.
Thanks, Dan and William. Can I ask two follow-up questions? Number one is, for these traditional enterprise customers, do they have more demand or do they want more to be in the core tier 1 cities or are they happy to be at edge of town locations? How does that impact your capacity expansion plan? Number two is that if you have more of these traditional enterprise customers on a retail basis, does it mean that your utilization rate will likely ramp up more slowly than before? You need to go out and find these enterprise customers and retail model.
Yeah, you're right on that last point. The enterprise deployments are mainly downtown.
Yeah.
There may be a limit on that. Yeah. 1,000 sq m plus order size, which could go to edge of town. We do see some order size like that, but it's not yet part of our customer mix range of town. I think, yeah, the enterprise business does require downtown.
Yeah.
The sales cycle is longer and it's more intensive with the increase in our sales resources to cover that segment. We also have a lot of follow-on sales potential with our established customer base and our, let's say, 700 odd customers is almost 250 financial institutions in that mix. That's been built up over more than 15 years. We were the first mover in providing business continuity and disaster recovery. A very high-quality customer base there, and there's a lot of potential, particularly as they start to change their IT architecture to make new sales to those long-established customers.
Okay. Yeah, that is great. Thanks for the comments.
Your next question comes from the line of Hong Ji Lee from CICC. Hong Ji, please ask your question.
Thanks, management. I ask one question. Continuing on the power supply, as you asked, we have multiple pipeline connects construction program, especially in Longkou. How do you see the power supply over there? Will the power supply impact our delivery? In other urban area in edge town, except for the government hotel, they have other ways to guarantee our power supply over there, I think.
Sorry. Look, sorry.
The line is not clear.
The line is really unclear. I was checking with Laura. Could you repeat the question once more?
I think she said on power supply in Longkou.
Power supply in Longkou.
Will that affect our delivery next year?
Okay.
Yeah. I think Longkou is a city, has some limited power capacity, number one. Number two is that we are first mover in Longkou. We definitely secured the power capacity for the future development. I think this is our first-mover advantage. I think the power supply will not impact our delivery in the Longkou area in the next or couple of years. Is that your question?
Thanks.
Okay.
Once again, if you wish to ask a question, please press star one on your telephone. For the benefit of all participants on today's call, please limit yourself to one question. Your next question comes from the line of John Choi from Daiwa. John, please ask your question.
Management, thanks for taking my question. I have a quick question. I think initially, management, you guys mentioned about your ESG efforts. Like next year, you'll spend a bit more, and are you still able to extend on 1% EBITDA margin? I think, with the recent carbon neutral initiative by the government, what is our plan for on the renewable side, and how will that impact our costs in terms of developing and would that have any impact on our MSR? Just quickly on, Dan, as you mentioned about the acquisition inorganic. It seems to me that inorganic has been stepping up since the past 12 months. Is it because the projects are being more mature in town cities or the prices have been coming off? Is there any trend that you could share with us? Thank you.
Hi, John. Thanks for initiating. I think John's the most recent one to initiate on GDS report, if I may say. We'll publish our first ESG report next year. Until then, I don't want to come out and set any expectations because I think there's a lot of irresponsibility in this area where companies talk about targets without giving a timeline or any metrics, and it just looks to me like marketing. We've established a team. We have and we are enhancing our expertise in the power sector to try to manage this better. It requires ingenuity and creativity, to be able to increase the proportion of our data centers which use renewable energy. We work hard to do that and take it very seriously. You said the inorganic proportion seems to have stepped up.
We've done, I think with the Shanghai 19 acquisition, which we announced today. That was the 11th acquisition which we've done since, I think the Q2 of 2016 was the first M&A deal. It's been around two deals per annum. Some of those deals have been larger. For example, one of the deals in 2019 was Beijing 10, 11, and 12, so it's like three data centers. The deals that we've done in 2020, like Beijing 13, was 20,000 sq m. In fact, we hope soon to announce that we've been able to upscale that quite significantly. Beijing 14, which hopefully will move forward and sign definitive agreements and close next year, is also 20,000 sq m. I think we're still doing around two deals per annum and hope to maintain that if not more. Let me comment about acquisition multiples.
We're seeing two kinds of M&A opportunity in terms of stage of development. Most of our deals have been projects which are under construction, or at least when we sign the sale and purchase agreement, there is still a substantial amount of work to be done to complete those projects. In that case, we've been buying projects at single-digit multiples of estimated stabilized EBITDA, factoring in what we pay and the cost to complete. Shanghai 19, you can do the math from what we disclose, definitely fits into that category. There's been a couple where we've acquired data centers which were already complete. As I mentioned before, there's more competition for those kind of opportunities, including from financial investors. We will do those deals when we see very strong strategic rationale. Of course, the multiples are double digits now.
I'm interested to see when the REIT market in China develops, which won't be long now, in respect to data centers. I'm interested to see what kind of cap rate data center REITs will command. My expectation is it's going to be way below the level at which we've done our deals, even the more advanced or mature deals.
Great. Thank you.
As there are no further questions, I'd like now to turn the call back over to the company for closing remarks.
Thank you everyone once again for joining us today. If you have further questions, please feel free to contact GDS Investor Relations through the contact information on our website or through Fontaine Group Investor Relations. See you next quarter.
Thank you, presenters. This concludes this conference call. You may now disconnect your line. Thank you.