GDS Holdings Ltd (GDS) 2017 Q1 法說會逐字稿

Hello, ladies and gentlemen. Thank you for standing by for the GDS Holdings Limited's first-quarter 2017 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 is being recorded.

I will 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, and welcome to the first-quarter 2017 earnings conference call of GDS Holdings Limited. The Company's results were issued via Newswire's 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 investors.gdsservices.com.

Leading today's call is Mr. William Huang, GDS Founder, Chairman and Chief Executive Officer, who will provide an overview of the business. Mr. Dan Newman, GDS Chief Financial Officer, will then review the financial and operating results.

Before we continue, please note that today's discussion will contain forward-looking statements made under the safe harbor provisions of the US 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 the views expressed today.

Further information regarding these and other risks and uncertainties is included in the Company's prospectus, as filed with the US Securities and Exchange Commission. The Company does not assume any obligation to update any forward-looking statements, except as required and under applicable law.

Please also note that GDS 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 Chief Executive Officer, William Huang. Please go ahead, William.

Hello, everyone. Thank you for joining today's call. In the first quarter of 2017, we continued to make meaningful progress across all aspects of our business and have further strengthened our market leadership position.

As you can see on slide 3, we grow total revenue by over 60% and adjusted EBITDA by over 130% year on year. We invested over RMB50 million of CapEx to develop the capacity required by our customers. We raised nearly RMB80 million of debt to ensure that each new project is fully funded.

We maintained our strong sales growth momentum, signing up customers for over 7,000 square meters net of new commitment worth over RMB35 million in terms of annual recurring revenue. This includes full re-commitment of the area released by our churn customer.

I want to highlight that it is a rare case for any company to be able to relocate such a large amount of area in such a short period. This further proves GDS's high ability to execute and deliver.

As we -- as the strong market demands, as well. As the result, our total area committed grew to over 68,000 square meters, 85% higher than one year ago. We ended the quarter with commitment rates of 90% for area in service and 38% for area under construction. We continued delivering the backlog, increasing our area utilized to nearly 38,000 square meters, 58% higher than one year ago.

On the resource side, we enhanced our supply pipeline with the Shenzhen acquisition we announced last quarter, of a 10,000-square-meter project under construction. Construction is ahead of schedule and the customer will be moving in Q3, faster than anticipated.

In addition, I'm pleased to announce today that we have secured a new data center project in Beijing. It will add over 4,000 square meters of capacity when completed in the first half of 2018. This is a big win for us. Supply in Beijing is quite restricted. Our approval reflected the government's recognized of our unique industrial leadership and deep experience. Clearly, both of these new projects demonstrate our sourcing capability.

Turning to the slide 4, our impressive sales achievements give us high visibility to future growth. Quarter after quarter, we continue to execute against our plan, delivering our backlog to customers and driving revenue and operating income growth.

I will leave it to Dan to elaborate on these numbers.

Let's moving on the slide 5. Before I talk about sales wins and the resource development, I want to update you on what we see happening in the market. As we have said before, we believe China represents the biggest data center opportunity in the world and we believe GDS is better positioned to capitalize on this growth.

In our view, the data center market in China is growing at over 25% annually, with cloud and internet platform service providers driving around 70% of new demand, and the cloud being key. So to understand the data center opportunity we must first understand how the cloud is developing in China.

Mr. Joe Tsai, Executive Vice Chairman of Alibaba, stated that they see cloud as a $30 billion market opportunity in China, assuming 20% of IT spending migrates to cloud. In 2016, the China market was worth around $1.5 billion. It is still early days, but the cloud adoption has clearly taken off.

Alibaba and Tencent reported that their cloud customers and the revenues increased by two to three times over the past year. Alibaba forecast said the market will grow by four times over the next two years.

This growth is largely coming from the three internet giants in China, Alibaba, Tencent and Baidu. They see transformational opportunity in cloud and the related technologies. Each of them has put the cloud at the center of their strategic development. These three companies have a huge influence over the digital economy in China. They are able to leverage their dominant presence in different verticals to drive the development of their cloud platform in unique ways.

Additionally, there are other giant companies, such as Huawei, which are allocating more and more resource to public cloud development. And the global hyper-skilled players, such like Microsoft, AWS and IBM, which are significantly increasing their presence in the China cloud market.

Cloud service providers aggregate demand for data center capacity, but they have different kind of requirements, which they fulfill in different ways.

For their performance-sensitive data and applications, which need to be housed in edge data centers close to the end user in the tier-one market, they looked to outsourcing most of their requirements. Why do they outsource? The first reason is because their business is highly dynamic, and outsourcing allows them to accelerate their time to market.

Secondly, they want to focus their attention on the customer-facing products and services that they are critical -- that are critical for their success. And they recognize that sourcing, designing, building and operating sophisticated data centers is very complex.

Third, their business appreciates the flexibility of being able to access a secured supply of data center resource with short lead times when they need it, from a trusted provider. Demand is running ahead of supply in the key markets. Hence, a lot of new capacity has to be built. As the scale has increased in terms of space and power, it has become more and more difficult to secure suitable industrial buildings for long-term lease and to obtain sufficient power capacity.

Obviously, the market opportunity has attracted more carrier-neutral participants, but we don't see that -- we don't see them having a material impact on competition. They develop their assets on a case-by-case basis, and no competitor presents comparable to us across all the tier-one market. No competitor has the combination of expertise, track record and the comprehensive sets of sustainable advantages that we do.

We recognize several years again -- we recognized several years ago the strategic importance of capturing demand from cloud service providers.

First, because it represents a large part of the market opportunity and by serving these high-volume customers, we gain scale advantages. Secondly, because we believe that access to key cloud infrastructure platforms located in our data centers is a unique value proposition that will drive the growth of our FSI and large enterprise franchise and attract more cloud players, such as SaaS providers.

There are only a few high-scale cloud infrastructure players, so there are only a few opportunities to make this a winning strategy. How are we doing? We have had great success in capturing this demand, as shown on slide 6.

From almost nothing three years ago, cloud service providers now account for over 50% of our total area committed. In the last quarter, we obtained significant new commitments from three of the leading cloud service providers in China, including a follow-on order from one of the major global players.

Our top two customers are now each present in six to eight of our self-developed data centers and in four to five different markets. We estimate that we house the majority of their incremental cloud demand.

They consider us as partners in the development of their cloud business. As we grow in market leadership, we are actively taking steps to deepen relationships with key customers and formalize our partnerships in different ways.

So why do we win with cloud service providers? There are many reasons. First, we have the right kind of data centers, meaning for large-scale, high-power-density and high-efficiency facilities, and they are located in all the right place. So they map to where our cloud customers deploy their platforms.

Second, we are able to offer them certainty of future supply as a result of our secured expansion pipeline and the flexibility to take delivery when they need it. And third, as cloud service providers increasingly penetrated the FSI and the large enterprise verticals, they see value in collocating with many of the best names in China, who are our existing customers.

These attributes are not easily copied, and they distinguish GDS from all other carrier-neutral players. Besides the cloud service providers, in the last quarter, we added over 20 new FSI and large enterprise customers, further diversifying our customer base.

I would like to highlight a few customers. First, we signed a large, multi-site order from the leading online securities and fund trading company. Second, we obtained an order for the settlement and clearing platform being established by the Central Bank to support digital payment service providers. We already house the three leading e-payment platforms in China. We believe this new generation of FSI customers offer a long, long way for future expansion. And third, we just recently -- we won a bid with one of China's largest online travel providers.

Last, as we continue to build our ecosystem for both cloud and enterprise customers, we are in the process of obtaining the license and regulatory approval to provide a cross-connected service in China. We will update the market when more details become available.

In summary, our sales outlook is the strongest it has ever been. The market is growing faster than imagined, and the customers are looking to us to meet their requirements. This is a golden opportunity for us, and we are moving forward full speed ahead.

We have the strategic relationships, and we are rapidly adding new customers. Resource supply is the critical input for us to sustain the momentum.

As shown on slide 7, our area in service remains at around 61,000 square meters, which is almost sold out. With the Shenzhen 5 acquisition, our area under construction increased to just over 35,000 square meters at the quarter end. Out of this total, around 13,000 square meters is pre-committed, and around 21,000 square meters is available to feed our current sales efforts.

The new data center project in Beijing, we call it Beijing 3, will be included in area under construction in Q2. It is located next to our Beijing 1, which is already 96% committed. We are very confident about the marketability of Beijing 3. Including Beijing 3, we will have around 25,000 square meters of area under construction, which is not yet committed. 25,000 square meters is equivalent to about four quarters of new booking at our recent quarterly run rates.

We intend supplementing these resources in order to support sales in 2018. We have resource held for future development that we plan to activate and we have promising prospects in all our markets that we aim to secure in the next few quarters.

Our data center portfolio is shown on slide 8. Ten of those data centers are in service and five are under construction.

Before handing over to Dan for the financial review, I would like to highlight that the recently announcement appointment of Mr. Chang Sun as an Independent Director on our Board. Chang was formerly the Chairman of Warburg Pincus for Asia Pacific. He is one of the most renowned and successful investors in China over the past 20 years.

Real estate based business has been one of his focus areas. He helped to build highly successful business in the logistics, hotels and retailing sectors. He also founded the China Real Estate Developers and Investors Association. GDS is the first and only US public company directorship that he has accepted since moving on from Warburg Pincus. Chang brings a rich investment background and experience and we are delighted to have him join our Board.

With that, I will now hand the call over to Dan. Thank you.

Thank you, William. In this section, I will focus on four main areas. First revenue and operating leverage, secondly CapEx, thirdly funding and then fourthly I will make some comments about delivery of the backlog and quarterly expectations for 2017.

Starting with the first quarter 2017 P&L analysis on slide 10. On a GAAP basis, service revenue grew by 14.7% quarter-on-quarter to RMB343.7 million. However, this includes a termination fee of RMB44.1 million, arising from the previously reported churn of 1,225 square meters at our Shanghai campus. The termination took effect a few days after the start of the quarter and the fee is now cash in the bank.

In Q4 2016 we generated service revenue of RMB16 million from the churn area. The termination fee is therefore equivalent to nearly nine months' revenue. On a pro forma basis, excluding the termination fee entirely from Q1 2017, service revenue was RMB299.6 million, which is the same as for the prior quarter. In other words, even without the buffer of the termination fee, we were able to keep service revenue at the same level as before the churn event.

Turning to slide 11, I will show you how we can better track the underlying trends in our financial results. On a non-GAAP basis, adjusted NOI grew by 27.2% quarter-on-quarter to RMB179.4 million. After excluding the termination fee and equipment profit which is non-core, underlying adjusted NOI was 4.2% lower compared with the prior quarter.

The main reason for this is that we booked a full quarter of operating costs for the three data centers, totaling over 12,000 square meters of space, which came into service during Q4 2016. The contribution from these data centers will take several quarters to build up to breakeven.

On a non-GAAP basis, adjusted EBITDA grew by 34.7% quarter-on-quarter to RMB123.9 million. After excluding termination fee, equipment profit and the impact of foreign exchange changes, underlying adjusted EBITDA grew by 2.3% quarter-on-quarter. Underlying adjusted EBITDA margin was slightly higher, at 27.1% in Q1 2017, compared with 26.5% in Q4 2016.

Now let's go further into the key growth drivers on slide 12. In Q1 2017, there was an increase in area utilized of 816 square meters net of churn, or 2,041 square meters gross. Typically, the first quarter of the year in China is comparatively slow due to the New Year holiday.

The churn rate was 5.5% on a revenue basis, or 3.2% on an area utilized basis. Excluding the single customer churn event, the churn rate was 0.1% on a revenue basis. Retention rates are very high. This was really just a one-off event.

The average monthly service revenue, or MSR per square meter, was RMB2,708 in Q1 2017, compared with RMB2,797 in Q4 2016. The lower MSR per square meter is mainly due to the churn area having an above average MSR per square meter, as it was very high power density.

On the right-hand side of slide 13, we show a breakdown of our cost structure for Q1 2017. All of these numbers are on a pro forma basis, excluding the termination fee, equipment sale and cost and foreign exchange changes.

Starting at a data center level, utility cost, which is mainly a variable cost, was 23.8% of pro forma service revenue in Q1 2017, compared with 23.2% in Q4 2016. As mentioned, we had three new data centers entering service in the prior quarter, which will take a while to reach optimal power usage efficiency.

Other data center level costs, which are mainly fixed, were 31.5% of pro forma service revenue in Q1 2017, compared with 30.1% in Q4 2016. The increase was mainly due to booking a full quarter cost related to the three new data centers.

Underlying adjusted NOI margin was 44.7% in Q1 2017, compared with 46.7% in Q4 2016. Directionally, we do expect our NOI margin to trend up over time. To give you a better idea of the potential for margin improvement, we can point to our stabilized data centers, which for this purpose we define as data centers with a utilization rate of over 80%.

At the end of Q1 2017, we had 18,155 square meters of area utilized in self-developed data centers which were stabilized. The underlying adjusted NOI margin for these data centers was 56% in aggregate. At the same time, we had 12,859 square meters of area utilized in self-developed data centers which were still ramping up and, therefore, had not yet reached optimal levels of profitability.

Moving onto the corporate level, our SG&A, excluding D&A and stock based compensation, was RMB7.7 million lower in Q1 2017 and came out at 17.9% of pro forma service revenue, compared with 20.5% in Q4 2016. There were some year-end costs and professional fees which impacted the prior quarter.

Turning to our investment activities. As shown on slide 14, we paid CapEx of RMB380 million in Q1 2017, which was a step-up from the run rate over 2016. Replacement CapEx accounted for 3.3% of this total. We incurred CapEx mainly on our Chengdu 1 phase 3 data center, which is 100% pre-committed and due for delivery starting in mid-2017, and on our Beijing 2 data center, which after the Government imposed construction delay, we are completing as fast as possible. As William mentioned, we obtained a significant follow-on order for BJ2 from a global cloud player, which will roll out simultaneously in our Beijing and Shanghai data centers.

Shenzhen 4 phase 1 is close to completion. We expect to obtain significant new commitments for this data center in the next couple of months. In fact, the deals are far along in the contracting process.

Shenzhen 5, SZ5, the project which we announced in March, is proceeding ahead of schedule. The customer, which is pre-committed for 50%, wants to start moving in during Q3 2017. Based on the pricing for the pre-commitment, we expect the NOI yield for this project on a stabilized basis to be in the mid-teens.

Shanghai 4, SH4, is the fourth data center on our main Shanghai campus. We aim to bring the data center into service by the end of 2017. The previous phase, SH3, is now 89.3% committed. All of our major cloud customers are present on this campus, as well as over 100 FSI and large enterprise customers. SH4 is already substantially pre-allocated on a soft basis.

We have also shown here the new Beijing project, BJ3, where construction starts in the current quarter. It's well placed to serve the customers who are present in the adjacent BJ1 data center. With the addition of BJ3, our total area under construction is 39,315 square meters across seven sites, out of which 76.4% will enter service this year.

With regard to financing, as shown on slide 15, during Q1 2017 we repaid a RMB199.6 million mezzanine loan, which had a high interest rate and obtained RMB532.8 million of new debt facilities. Our blended financing cost was 7.1% in Q1 2017, compared with 8% in Q4 2016. Excluding the convertible bond, the cost was 6.2% in Q1 2017, versus 7.3% for the prior quarter. Almost all of our loans are floating rate, linked to the PBOC rate.

At the end of the quarter, our gross debt was RMB4.47 billion and our net debt was RMB2.94 million. Net debt increased by RMB464 million since year-end 2016. The ratio of net debt to last quarter annualized pro forma adjusted EBITDA was 9.2 times. Fully diluted for conversion of the CB, the multiple was 6 times, so clearly conversion of the CB would make a significant difference.

Because we are in a heavy investment phase, consolidated net debt to EBITDA ratios may not give a good indication of our financing structure. On a pro forma basis, if we deduct the financing obligations related to data centers under construction, the net debt to EBITDA multiple falls from 9.2 times down to 7.5 times. If we further deduct the financing obligations of data centers which are still ramping up, the net debt to EBITDA for the stabilized data centers, which I referred to earlier, would be 2.7 times. The level of debt, which we put on each project, is designed to reach this kind of multiple, once stabilized.

As things stand, all of our data centers in service and under construction are fully-funded, including the two latest projects, SZ5 and BJ3, where we are in the final stages of securing the debt financing.

The fact that we had very strong customers and many long-term contracts definitely helps us to access the debt finance which we need. Assuming a continuation of our current funding approach, we believe that we have sufficient capital to take on around five further projects, subject to size and timing.

Turning to slide 16, I would like to make some comments to set appropriate expectations for the quarterly progression of our results this year. Our backlog grew to over 30,000 square meters at the end of Q1 2017. This provides high visibility to future revenue growth over a one to two-year time horizon. On a near-term basis, growth rates depend on how quickly customers move in, and hence how quickly you can start billing them for the committed resource.

Around 56% of the backlog relates to area at data centers which are currently in service. For this part of the backlog, the move-in process is smooth and ongoing. The balance of 44% relates to area at data centers which are still under construction. For this part, the move-in process can only begin in the third and fourth quarters of this year.

I mentioned on our last earnings call that we have a renewal this year for 4,365 square meters, which relates to a single customer at a single data center. At the time of renewal, the customer will change out its IT platform and we will undertake some refit.

As part of the deal for a multiyear renewal, we've agreed not to bill the customer during the change-out period, which is expected to last three to four months. The change-out will happen in phases over the next few quarters, but it will start to impact our revenue in the current quarter.

We took account of the move-in schedules and the temporary billing interruption, when we provided our FY17 revenue and adjusted EBITDA guidance, and we reaffirm those numbers.

In terms of quarterly progression, taking Q1 2017 pro forma as the base, we expect revenue in Q2 2017 to grow in the high single digits in percentage terms, quarter-on-quarter, and then at higher growth rates in the second half of the year. We expect adjusted EBITDA to follow a similar pattern.

With that, I will end the formal part of my presentation. We would now like to open the call to questions, so, Operator, please proceed.

(Operator Instructions). For the benefit of all callers participating in today's call, if you wish to ask a question to management in Chinese, please also immediately restate your question in English. (Operator Instructions). Our first question is coming from the line of Gokul Hariharan from JP Morgan. Please ask your question.

Yes, hi. Thanks, William and Dan. First, let me ask a quick question on the change in tone on interconnect or cross-connect. Could you talk a little bit about what has changed from a regulatory framework perspective, that is -- it feels like you are more confident on getting rights to do cross-connect, especially inter-data center cross-connect?

And could you also talk a little bit about -- on a pro forma basis, how much of an impact it could have if you could charge for cross-connect, just like some of the global center data guys are able to charge?

(Inaudible - microphone inaccessible)

Yes, hi, Gokul. The old telecom guy, William, has handed it over to me. There was a change in the -- what is referred to as the catalogue of telecom services in 2015, including --

The end of (multiple speakers).

Yes, end of 2015. Included under basic services a category of -- I'll say in English, call it a telecom infrastructure service, which entitles the licensee to actually install and operate fiber for very specific and limited purposes. And we believe that cross-connect within a data center would fall within the remit of that license.

So far, only very few licenses for that particular category of service have been issued. We initiated the application process and we've received preliminary approval. I think it could take some quarters, so we don't want to be specific. In fact, for a new category of license it can be quite a protracted process before the license is finally issued.

In terms of the impact, cross-connect is already possible inside our data centers -- or between our data centers, but just not through our own fiber. So, the fact that we have the ecosystem of cloud and enterprise and FSI customers is already part of the value proposition. Obviously, it would facilitate our own product offering in the network area if we could manage it entirely ourselves.

I don't want to talk about it in revenue terms. I believe that this will become a major driver of the growth of our FSI and enterprise business, going forward.

Yes, Gokul, I want to add on some points. First of all, I'd say we are pretty focused on the -- to grow our colo business. In my view, it's a good base, because based on our skill of the hosting colo, our customer is, as we introduced it before, our hard customer, or in the future, they already have a lot of demand connect to each other.

As you'd see in the US, cross -- connected to cloud and then connected to multi-cloud and the cloud connect cloud and the customer connect to customer. It's getting more momentum, right. So this is what happened in China already. And GDS is in the best position, because our customer profile -- if you look at our past customer profile, why is the cloud platform -- we have the major cloud happening in our data center.

And with -- on the other hand, we have the financial institution and the large enterprise customer and internet platform. So this will drive a lot of demand of connect to each other. But so far, we want to develop very carefully, to fulfill the license issue. And also, we want to develop our product properly to provide the service for future.

But this will cause GDS, number one, is help GDS strengthen our stickiness of our customer. This is a major -- I mean, a benefit for the future. Second, potentially can be a new revenue stream. But we want -- we don't want -- it's still in a very early stage. Although we already start to prepare, but we -- as Dan mentioned, we don't want to talk too much right now, but once it's everything ready, we will introduce more detail.

Okay, great. Second question I had is, if we look at your potential ramp-up schedule in terms of pre-commitments, as well as new data centers coming online, both Shenzhen 5 as well as the newly secured Beijing 3, if you think about, let's say some time in 2018, do you still feel that your top two customers who have close to 50% area committed right now will be becoming bigger and bigger as opposed to (inaudible) your area committed when we get all these data centers pretty much committed? Or do you think that this mix can still remain in that 50%, 51% range?

And just asking the question in perspective that the top two internet companies who are pushing in the cloud space in China seems to be basically dominating the market. I think number one is really dominating but number two is also coming in, and three, four, five are still really far behind. So just wanted to understand how you see this shape up over the next couple of years.

Portion represented by the top two customers. Yes Gokul, it's only speculation, of course. I think at this stage the major impetus is coming from the cloud. And the service providers they have to roll out the platform before they can attract the demand. So I think we are in a phase where we will see the percentage represented by these cloud service providers increasing.

But then over time I think we'll see the cloud segment will diversify because there'll be many more SaaS providers. Maybe smaller individually in order size but larger in number. And then, of course, there will be the growth of the enterprise and FSI side because we already have mostly access nodes for the cloud platforms in China inside our data centers.

That's how I see it. I don't see it as being a two-horse race. We have valuable relationships with more than two hyper-scale cloud service providers. But the strength of our relationship with those two is very significant. And as William mentioned that we are taking steps to formalize the partnership to recognize all the ways in which we can be of mutual benefit to each other.

Understood. My last question on margin leverage. Dan, I think your proforma margin is hovering still around 26% to high 20s' range. Is it second half of this year when we really see that start to move up towards the 30% range that you've talked about before? Or is that still dependent on how quickly some of these new data centers come on line?

From my point of view -- I mean what's important for me to see is that each data center is hitting a high level of profitability and giving us a high return. And that's happening consistently. And the stabilized data centers, actually the margins are still going up. They're not all fully utilized yet. You're correct what you said, the leverage will be more significant in the third and fourth quarter of this year, and continue into next year, yes.

Okay, great. That's all I have.

Our next question is coming from the line of Jonathan Atkin from RBC Capital Markets. Please ask your question.

Thank you. So I was interested in your -- the top five customer slide. I just wanted to clarify are any of those top five global players or not? Slide 22.

Yes, one of them is, one of them is a global player, yes.

Right, okay, just wanted to clarify. Then I mean --

Number four if you wish. You're going to guess anyway.

Right. So I wanted to talk and get you to comment a little bit about the broader environment given the growth in the overall category. And what do you think is GDS's current share as you look at some of these spending trends among cloud service providers, internets and even FSIs?

And then maybe just my last question I'll get that right straightaway, you gave very useful commentary about how the revenue develops through the end of the year to hit your full year guidance. I was interested in the fact that you've got new capacity coming on line and so there's going to be some operating expense implication. Can you talk a little bit about the margin or OpEx development as we move through the year? And as we think about unit revenues should we keep them -- the MSR number that you talked about at around 2,700 is that going to be roughly the same? Thank you.

Yes, Jon you're first question was -- just make sure I answer it clearly, were you asking about what is our market share of the cloud business, right? Is that what you asked?

Yes, you look at your pipeline and -- so these customers are growing with companies other than GDS, so I was interested in your view as to competitive supply. And is competitive supply growing as rapidly? Is it happening among the telcos? Is it happening among other Chinese companies? Or are global data center companies also adding capacity in the market?

Okay, I'm not sure if I -- let me see if I understood correctly. Jon, if you look at it from a cloud service providers' point of view in each market they typically have two to three zones, which are separate from each other. And hence, it's quite common to see different service providers for each zone. So if we are doing well we should get around one-third to half the business, which we have done at least that.

Who else is getting the business? Typically, it will be the incumbent telecom operator in each region. And sometimes the incumbent or even one of the competitive telecom operators, they don't have data center resource suitable to satisfy this requirement. So we've actually benefited both from getting, let's say, one distinct piece plus in some cases, we are effectively working with a telecom carrier to fulfill their piece.

All right? Does that answer your question about the cloud market share or did I miss something there?

Yes. Maybe one other question. Is there any discernible supply coming on by non-Chinese players to meet data center demand within China?

Well, without betraying any customer confidences I can tell you that at least the top two global players are stepping up their presence in China.

I was referring more to data center construction. So are there other suppliers in the market that you're seeing differently than perhaps six months ago.

There's always -- so you mean, Jon asked me on data center supply, right?

Yes.

Jon, the market is still quite tight. And all of this demand needs to be satisfied with new builds. There is not that much new build going on. So if you're asking from the point of view is there a lot of supply that's going to materialize, the answer is no.

From our point of view, we don't worry too much anyway because we are not just competing at the asset level in a particular market. So I think we are fairly immune to that, because our relationship with the cloud service providers is a reflection of a lot of other attributes, which William enumerated.

Okay. So I see on slide 16 that you talk about ASPs being roughly similar in the backlog as what you currently have.

Yes.

So I guess my final question then just to be a little bit more specific is about OpEx or the margins and how that's developed.

Yes. Well, I think the expectation on the revenue per square meter you should assume it stays around the level that it's been in the last six or so quarters. It will fluctuate because it's affected by the timing of during the quarter when somebody moves in. It's affected by power usage levels. It's affected by the contract terms where sometimes the pricing builds up to the full billing amount over a period of time. But essentially, as I've said a few times, the average selling price in that contract backlog is almost exactly the same as the MSR that we're actually yielding for the capacity which is billable now, today.

And your question about OpEx, clearly, there is a lot of capacity that's going to come into service during this year. I mentioned the numbers. Three-quarters of the area under construction will come into service this year, and that's always a drag on the consolidated NOI and EBITDA margins.

So the underlying profitability is there, as I tried to demonstrate by referring to the stabilized data centers, but the trend is always going to be held back, because we're in a phase where there is actually more capacity under development than we even anticipated.

Thank you very much.

Our next question is coming from the line of Michael Hart from Guggenheim Securities. Please ask your question.

Hi. Thanks for taking the questions. I guess first to follow on some of the themes that Jonathan was asking about, it seems like as you look across your markets broadly, there's a lot of demand and limited supply, so that should translate to a pretty strong pricing. I was wondering if you could give us some more granular color on which markets you see that have particularly strong pricing trends, or if there are any markets you would call out where maybe you haven't seen as much strength in pricing as you might have seen elsewhere.

Michael, I'll go first. William will add.

Okay.

Thanks for your question. The China market is fairly concentrated, as indeed I believe the US and European markets are as well. It's really about Beijing, Shanghai, Shenzhen, to a lesser extent, Guangzhou and Chengdu, and that's where the cloud platforms are being located right now in this period of development.

And I would not distinguish the supply-demand or pricing situation in any one of those markets. All of the demand has to be satisfied with new builds. There isn't any significant difference in terms of the amount of supply relative to demand in any one of those markets, and pricing is pretty similar in each of those markets.

There have been times in the past where there maybe have been a little bit of divergence, but not material. It's pretty similar, at least in Beijing, Shanghai, Shenzhen, it's very similar.

Yes, I would like to say the market which we choose, which we target, is all the tier-one markets in China, and the demand is major driven -- the data center demand, majorly driven by this area.

So this demand is very strong in each core market. So that's why we are target for -- target to this key market. That's our strategy. So this is that we can price pretty similar during -- compared with this four different core markets.

But in China, if you compare with the tier-two cities and a lot of the tier-two, tier-three cities, a lot driven by the governments -- the data center built by the governments, the price, it's a huge difference. So our core market, we are sitting on the golden market, in our wording. So the demand is strong, and the supply is not -- it's more and more difficult.

Great. Thank you. That's really helpful. I guess the next thing, following on the issue of supply coming on, I know you mentioned that the demand really requires new builds, so you're seeing customers, they're willing to take commitments for construction that hasn't begun yet. Do you have a level of pre-commitment that you look at to have either firm contracts or soft pre-commitments before you begin construction? Or do you feel pretty confident that you can begin construction and demand will show up for that new supply?

Michael, we have the benefit of having established customer relationships with many of the customers who matter in China, and we are closely aligned in terms of our resource plan and their resource requirements. If anything, actually, the planning process with certain customers is getting more -- is getting closer and more frequent. So when we undertake a new project, I'd say we already have a very strong sense of the demand.

How that translates eventually into a contract, that can take some time. Of course, from external point of view, our disclosures are based on what's contracted. But internally, we for now use the word soft allocation, and these projects are soft-allocated long before you hear about them being contracted.

So when you look at our capacity that we have under construction, a lot of that is soft-allocated.

Yes, Michael, I would like to say that if you look at last year and the year before last year, every year, our new incremental order is from -- last year, it's almost 70% is from our existing install base. So that will tell -- translates, we already know a lot of our customers, their intention and their demand and their plan.

So this is -- as Dan mentioned, actually internal insight, we already know -- get some commit from our customers. Then we go to contract. So this year, we repeat again, so a lot of our new incremental orders, 80% almost this year, were from our existing customers.

Great. Thank you very much.

(Operator Instructions). As there are no further questions at this time, I'll now turn the call back to the Company for closing remarks.

Thank you 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 the Piacente Group investor relations.

Thank you.

This concludes our conference call. Thank you.

Thank you.

Ladies and gentlemen, that does conclude our conference for today. Thank you for participating. You may all disconnect.