GDS Holdings Ltd

HKEX:9698

Earnings Call Transcript 2018-Q3

[Operator]

Hello, ladies and gentlemen. Thank you for standing by for GDS Holdings Limited Third Quarter 2018 Conference Call. [Operator Instructions] Today's conference call 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.

[Laura Chen, executive]

Thank you, Joanna. Hello, everyone, and welcome to 3Q '18 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 investors.gds-services.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.

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 the views 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 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 CEO, William Huang. Go ahead, William.

[William Huang, executive]

Thank you, Laura. Hello, everyone. This is William. Thank you for joining us on today's call. I'm very pleased to report another quarter of outstanding performance with significant progress across our business.

Let's start. Starting with sales, which is the leading indicator of our future revenue and positive growth.

In the third quarter, we signed up customers for nearly 19,000 square meters of net additional area committed. All of this is organic. All of this is in Tier 1 markets. With this addition, we stayed on track to double the new business we signed last year. And at the same time, we delivered to our customers around 14,000 square meters of additional revenue-generating space. The utilization rate moved up to 68%. This drove another quarter in which our adjusted EBITDA grew by well over 20% quarter-on-quarter and 100% year-on-year.

Backing this up, we started the construction of 4 new projects and brought another 4 into service on time and within budget. We kept the sales and the capacity growth synchronized. The precommitment rate for area under construction was 45% based on signed contracts with significant more in the construction -- in the constructing process. And always, we ensured that all of our projects are fully financed with sufficient equity and long-term project debt.

Overall, this performance demonstrates the resilience of our business and our ability to keep on beating targets. We are once again raising our 2018 guidance for revenue and EBITDA.

Turning to Slide 4. A highlight of our sales achievement for this quarter was 3 significant new customers wins, namely, JD, Kingsoft and NetEase. This was a direct result of our targeting of larger-scale cloud and Internet customers who we believe are strategically important for our franchise because of their cloud platforms, valuable data and ecosystems.

On the enterprise side, we won our largest-ever order from a foreign financial institution, a top U.S. bank, which is moving fast to capitalize on the financial services liberalization in China.

Anyway, in this quarter -- new customers accounted for over 50% of our new business. This demonstrates that we keep diversifying our top customer base. At the same time, this does not mean our existing customers are pulling back. Based on what we see in our sales pipeline and what our top customers have shared with us, we don't see any letup in the current level of demand.

There is a lot of the data pointing to a slowdown in parts of the digital economy as a result of macroeconomic and geopolitical factors. How is that -- how is this affecting us? The short answer is that we don't see it. I think there are 3 main reasons to explain the apparent disconnect. First of all, our business is geared towards the growth of cloud platforms. And everyone knows public cloud in China is still at an early stage of development. The AI segment is just 15% of the U.S. but forecast to grow at nearly double the CAGR over the next 5 years. Alibaba, the market leader, has just reported 90% year-on-year growth -- revenue growth for its cloud business, which is similar to our growth rate. Furthermore, our largest customers still have a long way to go to migrate all their own IT onto their cloud platform. This internal demand is not reflected in -- to the -- reflected in the disclosed revenue figures.

The second reason is new technology, in particular AI. More and more applications have become AI-enabled. It's a major focus area for Chinese tech leaders. They expect to create a lot of economy -- economic value from AI over the next few years. From a technical perspective, AI requirements more data, more storage and more computation. Many AI applications are also highly latency-sensitive. This is increasing demand for data centers like ours in Tier 1 markets.

The third reason is that our largest-scale customers secure their supplies of data center resource based on 12 to 24 months planning time horizon. The capacity to fulfill their requirements does not exist. It has to be built. Therefore, our customers needed to contract with us now in order to ensure that they will be able to execute their business plans in 2019 and 2020. This is why we have a large backlog and high visibility for future growth.

In summary, we remain highly confident about the sales outlook based on our existing and the new customer relationships and the evolving technology.

Let's turning to Slide 6. I have talked about demand. Now let me talk about supply. There are many advantages to creating new resource supply in Tier 1 markets due to the pressure on real estate and the power. As scale increases, these challenges are only getting bigger.

The ability to deal with these challenges is one of the factors which sets GDS apart. As we initiate new projects, we consistently restock our resource pipeline. In additional to what you see currently under construction, we have another 100 square meters -- 100,000-square-meters-plus of capacity held for future development. Our ability to maintain continuous supply in all Tier 1 markets is a critical consideration for hyperscale customers. No other service provider in China comes close to us in terms of data center platform. This is why we don't see any significant change in the competitive intensity.

Going forward, we are evaluating campus-type developments on the edge of the Tier 1 markets to industrialize our capacity expansion, making it easier and more efficient for us to scale up. We are also actively considering entering 1 or 2 new Tier 1 markets. We expect to make announcements about such projects over the next couple of quarters.

We have a great track record of delivering resource to customers for their expansion. Although we don't talk about it very often, we also have a great track record of delivering exceptional operating performance. We were recently put to an extreme test. During September, Southern China was struck by the worst typhoon in a hundred years. As shown on Slide 8, we had 8, we had 8 self-developed data centers in the eye of the storm. Despite over 500 failures in the power grid and the damages to other critical infrastructure in the Shenzhen and Guangzhou regions, our data centers ran nonstop without any breach of SLA. This is the kind of achievement which our customers really recognize and appreciate.

With that, I will hand over to Dan for the financial and operating review.

[Daniel Newman, executive]

Thank you, William. Starting on Slide 13, where we strip out the contribution from equipment sales and the effect of FX changes. On a quarter-on-quarter basis, our service revenue grew by 20.2%. Our underlying adjusted NOI grew by 26.2%, and our underlying adjusted EBITDA grew by 29.4%. Our underlying adjusted EBITDA margin increased by 2.7 percentage points to 38%. Our reported adjusted EBITDA margin was 39.5%, a 3.1 percentage point increase quarter-over-quarter and an 8 percentage point increase year-over-year.

Turning to Slide 14. The main driver of revenue growth was the increase in area utilized with around 20,000 square meters added in the second quarter and a further 14,000 square meters in the third quarter. Our contracts with large-scale customers typically provide flexibility for how fast they move in. This gives us a bottom line in terms of delivery schedule, and upside if customer move-in accelerates. The outperformance in the second and third quarters reflects faster-than-expected move-in. This might surprise you considering reports from the tech supply chain, but it just goes to show that our experience does not necessarily correlate with what they are seeing.

Monthly service revenue or MSR per square meter in 3Q '18 declined slightly from the previous quarter. However, if we exclude the impact of the 3 Hebei projects, which came into service around midyear, the MSR actually increased. An MSR of RMB 2,759 per square meter, excluding Hebei, is 2.5% below the average for the preceding 12 quarters.

As we grow, our key objective is to sustain a return on investment, as measured by unlevered post-tax IRR, in the mid-teens. The MSR is a rough approximation for the average selling price, or ASP. But when it comes to measuring returns, we must also look at what is happening to development and operating costs. I'll come to this in a moment.

As shown on Slide 15, profit margins are on an upward trend. The growth drag at the adjusted NOI level, which we saw in the past few quarters, has now reversed. For this quarter, the margin expansion mainly came from the leverage on data center level fixed costs as a result of higher utilization. As shown on Slide 16, our data centers' area in service is now over 50% stabilized compared with 47% in 2Q '18. The increase in the stabilized proportion contributed to the margin improvement.

At the corporate level, on Slide 17, SG&A is 10.9% of service revenue. Going forward, we expect to achieve significant further leverage on our corporate overheads. Factoring in full delivery of the backlog, we can already see that the current level of SG&A is down to 6% of service revenue.

Turning to CapEx on Slide 18. In 3Q '18, our CapEx paid was around RMB 1.1 billion, including RMB 111 million related to acquisition consideration. For the year-to-date, our cumulative CapEx paid was around RMB 3 billion. Most of the costs of the Hong Kong real estate acquisition will appear in 4Q '18 CapEx.

As you can see from the breakdown of CapEx incurred, it will cost an estimated RMB 2.6 billion or USD 380 million to complete all the projects which we have currently in service and under construction. This would get us to a total of 191,000 square meters of fully fitted and equipped capacity.

For the projects under construction, the estimated CapEx per square meter is around RMB 60,000 or USD 8,800 per square meter at current exchange rates, which indicates how we have been lowering costs and maintaining our returns at target levels.

Turning to Slide 19. During 3Q '18, our net debt increased from RMB 7.2 billion to RMB 8.8 billion. Part of the increase was due to the capital leases for the SH9 and BJ6 properties and part was due to the settlement of CapEx payables. Our net debt to last quarter annualized adjusted EBITDA ratio decreased from 7.8x to 7.3x. However, in 4Q '18, with the payment for Hong Kong, we expect it to rise again. Our effective interest cost was 6.1%.

Most of our existing loans are project financed at the data center level. These facilities are structured in a conservative way, matching the debt service to the project cash flows over a 5-year period. Typically, once the data center is stabilized, the project debt to data center EBITDA multiple is around 3x. With a high level of precommitment in the form of 6- to 10-year contracts with investment-grade counterparties, the ability to service this debt over the project life cycle is assured. While projects are stabilized, our practice is to refinance on a longer-term basis. This means that we always have a lot of project finance going on.

In 3Q '18, we completed 4 project finance transactions for a total facility amount of RMB 792 million. Currently, we have another 5 project finances ongoing for a total facility amount of RMB 2.9 billion, out of which RMB 808 million is refinancing.

The China banking market is currently very liquid and very receptive to our business. The benchmark rates for determining our interest cost are stable. In fact, they have not changed for 3 years. This gives us a high degree of confidence in our ability to continue project financing and refinancing to the extent required.

Every project which we've announced to date is fully funded with equity and debt. This means that we are fully funded to the level of 191,000 square meters of move-in-ready space. In addition to this, we still have capital available for allocation to new projects, which we will announce in the future. We are confident in our ability to leverage this capital in the same way as we have always done. With the combination of available capital and leveraged capacity, we estimate that we could build out a further 50,000 square meters on top of the existing 191,000 square meters in service and under construction.

Turning to Slide 20. As at the end of 3Q '18, our backlog had increased again to over 61,000 square meters. We currently have around 100,000 square meters which is revenue-generating. The backlog implies that we can increase the revenue-generating space by 60% without signing any new customer contracts. Given the operating leverage, this should translate into more than 60% EBITDA growth from the level of 3Q '18 based purely on what is already contracted.

Finally, on Page 21. Our 3Q '18 results clearly show that we are tracking ahead of expectations in terms of revenue and adjusted EBITDA growth. The accelerated move-in brought forward new service revenue that we expected to commence later. Given where we are today, we are further raising our guidance for FY '18 revenue and adjusted EBITDA to not less than RMB 2.75 billion and RMB 1.02 billion, respectively. The new numbers imply year-on-year growth of over 70% for revenue and over 99% for adjusted EBITDA, in both cases exceeding the year-on-year growth rate we achieved in FY '17. At the same time, we're maintaining our FY '18 CapEx guidance of RMB 4 billion unchanged.

With that, I will end the formal part of my presentation, and we would now like to open the call to questions. Operator?

[Operator]

[Operator Instructions] Our first question comes from the line of Jon Atkin from RBC.

[Jonathan Atkin, analyst]

So I was interested -- I guess, a question for William or for Dan. As you look at potentially acquiring projects that are underway by smaller developers, what does the pipeline look for sort of small tuck-in M&A? And then a couple of questions that came to mind from your slide deck. Slide 16 gives information about the stabilized asset pool, and I wondered if you could tell us what the margins are for that set of stabilized assets. And then lastly, on Slide 18, you talked about your construction program, just under half of the area is precommitted. 3 of the projects that you list there are 100% precommitted. And as we think about the non-precommitted sites in Beijing and Shanghai, are those likely to fill with 1 single customer? Or would you potentially fill that with multiple logos?

[William Huang, executive]

Jon, this is William. I see what we see there in the market -- I mean, what happen is, and we see there is pretty a lot of -- there is a lot of the acquisition opportunity in the market. So we are -- we select a deal very, very carefully. So we have more pipeline than before, but that let us can select -- use a very high standard to select a target acquisition deal. So I think of it -- but we are still in the early stage, to evaluate there's a lot of the projects right now. Dan, you want to add some more color?

[Daniel Newman, executive]

Well, just on the other 2 questions that Jon asked, the margin for the stabilized part of the portfolio -- the stabilized part of the area in service, it fluctuates, but in the third quarter it was over 55%. And about the projects -- if I understood correctly -- the projects which are under construction but not yet precommitted. So I think we provided some information about when those projects will come into service. So we just look at the 2 projects that will come into service in the first half of next year which are not yet precommitted. One of them is Beijing 4, for now or for this section, we positioned that for enterprise and financial institution customers who typically don't pre-lease. They -- we may begin to take commitments 3 months before it comes into service. I said for now, because there is some pressure from large-scale customer, so we may reposition, reallocate that capacity where -- we will make a decision in the near future. The other data center which is due to come into service in the first half '19, which is not yet precommitted, is Phase 3 of what we call our Shenzhen 5 data center. It's a very huge building relatively centrally located in Shenzhen. It's highly marketable. Right now, it's a toss-up as to whether it goes to an existing customer in that building or whether we are free to offer it to other customers. Either way, that's a very nice resource to have in our hands. Does that answer your question, Jon?

[William Huang, executive]

Jon, I want to add on a little on M&A deal. Number one, the current -- currently, what we see in the pipeline is much stronger than before, but it does not say we would take action immediately. This allowed us to have more opportunity to get a deal, but our strategy is still keep on -- or getting growth. That's our priority.

[Jonathan Atkin, analyst]

That does answer the questions. And then just lastly, given the strong demand and the desire to grow, including potentially in the new Tier 1 markets, how do you think about your various financing options to fund this expansion?

[Daniel Newman, executive]

Yes, Jon. There's debt, and there's equity. Right now, the positioning in terms of the banking markets in China, as I said in the prepared remarks, is very favorable to us. And partly, as a policy response to what is happening, the Chinese government has reduced reserve asset requirements, I think, 4 times for Chinese banks. And that's made it easier for us to obtain the project debt that we require. Not that we were having any problem, but I think it just gives us the ability to say that the leverage will be there with a very high degree of confidence. On the equity side -- fortunately, earlier this year, we did a couple of capital raisings that put us in a relatively strong position going into this current situation. And I outlined the amount of capacity that we could build just using existing capital. And you can assume that's either organic or it's a combination of organic and acquisition. But the total of what I said came to around 240,000 square meters. We -- up until now, we have always raised the equity capital we need at the holding company level, but we are always continuously approached by investors who are seeking to partner with us at the project level. So that's something we keep an open mind about. So I think it's really -- something really that I'd like you and the market to take away, that we're not beholden to the public equity market to raise the capital that we need. It's simply a question of looking at what our options are and what is the lowest cost of equity capital available to us at the time when we need it.

[Operator]

Our next question comes from the line of Robert Gutman from Guggenheim Partners.

[Robert Gutman, analyst]

Can you just talk about the trend in accelerated move-ins? It looks like part of the beat in the quarter was the Chengdu site that was delivered early. I believe that was 94% preleased. What's the outlook for the pace of -- the accelerated pace of move-in timing as we roll through the next few quarters?

[Daniel Newman, executive]

Rob, it's Dan here. It's a curious one, isn't it, because it happened at a time when all the indications from the tech supply chain was that the opposite should be happening, right? And it wasn't 1 customer in one place. If you really drill down into the details in the appendix to our earnings presentation, you'll see that there is, I think, 7 or 8 data centers which came into service at the beginning of this year, which were already somewhere between around 50% to 80% utilized, occupied, which is very rapid by anyone's standards. Why is this? So there's a variety of reasons. Because it's not one customer, and it's not one place. And some of it is related to new products. Some of it is related to some promotional activities. And further, a lot of it is just correlated with the growth of the cloud. Going forward -- maybe it's too early to say, we get delivery notices from our customers. They give us around 3-months-plus notice of their move-in intentions, so we can forecast around that. But beyond that time frame, we tend to forecast based on what it says in the customer contract. So there's a bottom line, a delivery schedule, we call it minimal contractual commitment in the customer contract, and that's -- that underwrites whatever guidance or forecasts that we make. For now, I won't change our usual practice. I will leave it to [ kind of an upside ] and a pleasant outcome if indeed the customers move in faster.

[Operator]

Our next question comes from the line of Gokul Hariharan from JPMorgan.

[Gokul Hariharan, analyst]

First of all, Dan, could you elaborate a little bit more on your comment about the operating leverage that could start to kick in going into the next year or so? I think you started talking about SG&A coming down to about 10%, 11% level and potentially could be even lower. Second, could you talk a little bit also about what are your rough CapEx expectations, given the pipeline of demand that you're seeing as well as the entry into potentially 1 more -- new Tier 1 markets? If we leave the M&A aspect aside, which I think you're waiting on timing, et cetera, the right timing, from an organic basis, could you talk a little bit what could be the kind of CapEx that we should be thinking about? Is the number that you are spending this year roughly the kind of ballpark that we should be thinking? And one last question, a quick one. Could you talk a little bit about the Hebei project economics now that at least one of the data centers is pretty much stabilized. Are NOI margins in Hebei quite high compared to the normal data center operation given the ASP and the utility cost pass-through structures there?

[Daniel Newman, executive]

My pleasure, I could have guessed you were going to ask those questions. We talk about the operating leverage at 2 levels, at the data center level and at the corporate overhead level. At the data center level, it does tend to fluctuate. So we brought a lot of capacity into service late last year and early this year, and then that had a result of actually reducing or lowering our NOI margin. And as we've gone through the year, the utilization rate has increased by, I think, 3 or 4 percentage points, and the proportion of our area in service which has stabilized has also increased by 3 percentage points. So that's what's raised the adjusted NOI. But I have to say when I look over the next 3 or 4 quarters, it's not going to go up in a straight line. However, leverage on the corporate overheads, I think that can be a very significant feature over the course of next year. We talked about this before. We have a lot of corporate overhead, which is related to growth, in terms of sales and marketing, project sourcing, design, procurement, construction management and of course funding as well as some pre-operating costs, which are included in G&A. But those costs don't really need to scale. We're still operating in the 5 Tier 1 cities. It may increase by 1 or 2, but we're not going all over the map. So I do think we expect to achieve quite significant operating leverage on corporate overheads over the next year. In terms of capital expenditures, if you calculate from -- we've been disclosing costs to date and cost to complete for our in-service and under-construction capacity. If you look at the under-construction capacity and you calculate the total costs, you'll see it comes to less than RMB 60,000, which, at current exchange rates, is under USD 9,000. That's clearly lower than what we've talked about in the last couple of years, and it's an indication of the initiatives that we've taken around supply chain management and scale economics coming through. When we look at the CapEx for next year, although it's too early for me to give guidance, I think I can go on record and say that it's not going to be materially higher than this year's number. It will be higher, but not materially higher. As regards Hebei, in total, it's 3 data centers which represent less than 10% of our total area committed. So I don't want to call out too much about those 3 data centers. Otherwise, it's going to overcomplicate. But you're right. The revenue per square meter is lower than our average. The data center level margin or NOI margin is higher, and the project return is at the low end of our range. But because of the build-to-suit customer contract, we were able to finance these projects with a lower overall cost of capital, so the spreads over our cost of capital was actually quite attractive. I've disclosed just for the third quarter and the second quarter what the revenue per square meter -- MSR per square meter would have been without the Jiangbei -- without the Hebei projects. If I look at the NOI margin or EBITDA margin, the impact of those projects was not very significant in those 2 quarters.

[Operator]

Our next question comes from Frank Louthan from Raymond James.

[Frank Louthan, analyst]

Great. Can you talk to us a little bit about the EBITDA margins? It's been a nice uptick. Where do you think those can ultimately go to? And remind us kind of what the margins are on the stabilized properties? And just as a follow-up, and I apologize if you've mentioned this earlier, but have you -- what is the current status of any deals that you sourced from the CyrusOne relationship? And when do you expect to see some projects from that?

[Daniel Newman, executive]

Okay, Frank. It's Dan here. I will go first on the EBITDA margins. I know what you and Gokul [ obviously ] are getting at. You want to know how quickly our EBITDA margins are going to go up. The first target we'd like to -- or threshold that we'd like to cross is 40%. Incidentally, we IPO-ed just 2 years ago, just over 2 years ago. We just had a 2-year anniversary. And at that time, we targeted to reach 40% EBITDA margin at the end of 2018. And I think you can see that we are on track to achieve that despite the fact that we are growing -- our growth is 2x to 3x what we expected at the time of the IPO, which also implies a very significant growth drag that we had not factored in. So I think it's a good achievement for us to have got here. I mean, if you look at the stabilized part of the portfolio, I mentioned third quarter at the NOI level, it was over 55%. But when you add in the part that is not stabilized, which is ramping up, it brings the total down to 48.5%. And then you -- that won't go up very quickly, that was Gokul's earlier question. It will go up and down. It will trend up, but it won't go up very quickly. But then you have to deduct from that the SG&A, which is currently over 10%, but which will come down, I think, to the lowest levels you've seen from any U.S. data center service provider and maybe less. So I think maybe that gives you some indication of where the EBITDA margin will go to over the next 1 to 2 years.

[Frank Louthan, analyst]

Okay. Great. And on any CyrusOne [ deals ] which you've made. Maybe on [ some of these. ]

[William Huang, executive]

Yes, I think just so far, we still cook some deal right now. But what we see is that currently, the China government opened the door and let a lot of the financial institution and other, let's say, multinational coming -- multinational companies coming. So what we see are we are working very close with CyrusOne to try to catch up with this -- select at set -- a couple of this segment and have a vertical to which the company has the highest intention to come to China right now. So I think that, given the time, we will have the result. I'm so -- I'm very confident for that.

[Operator]

Our next question comes from the line of Colby Synesael from Cowen and Co.

[Colby Synesael, analyst]

Just a few. First off, just in light of where the stock's gone and your continuing needs to finance. I was trying to get a sense of what comfort level you have on leverage and where that could potentially go over the near to medium term. And then secondly, I think when we look at the macro, there is obviously a lag in terms of data centers, in terms of how people are purchasing those types of services from companies like yours. And I think the concern is that, as we start to see potentially a slowdown in '18 as we go into 2019, we're not going to see that show up in your leasing results, perhaps for another year or so, because a lot of the projects, to your point, that you're signing now and likely will sign for the next few quarters are from deals that you've been working on for perhaps a year or 2 at this point. What comfort level do you have that those trends, as we go into 2019, can be sustained based on what your own observations are of the underlying businesses of your customers? And then just my last one. You mentioned in your prepared remarks you're looking at potential edge expansions in some of the current Tier 1 markets that you're already in, looking to potentially industrialize how you go about building data centers. I was wondering if you can just give us a little bit more color on what that actually means and what the potential financial impact to your business could be perhaps over the next year or 2.

[Daniel Newman, executive]

I think William will go first on the...

[William Huang, executive]

Let me explain a little bit about the next year's outlook. I think a lot of people say, oh maybe a little bit that -- a server shipment that goes down and -- this does not affect -- directly affect us, what I just explained because of a couple of reasons. Number one, we have the right customer and our customer what -- the business which we host actually is very early stage in China. They all grow in a very significant way. And based on my latest-stage conversation with all the business unit heads from our key customers, they all talk to me. And nothing changed in next year. So this is number one, this is a very bottom-up conversation in the last 4 weeks, this is number one. Number two, I would say, based on our sales pipeline, I can tell, to maintain this year's levels sales booking in next year, I'm fully confident because we already had 2.5x of the pipeline coverage of our target. So we are just finished our sales plan in next year. So I'm very confident to maintain this year's level, new booking in next year. And on the other hand, I think, as we -- as I just talked in my presentation, I say we keep add on new large-scale customer in our customer list. This has quite diversified our customer base. And all the unique Chinese unicorn we still have a long [indiscernible] which hasn't been our customer yet so we are confident to adding more customers like the JD, NetEase and Kingsoft in the next few quarter. Okay. Let me talk about a little bit the campus-type development. Number one, I think the current environment, it's a lot of the local governments, now they changed their attitude. A lot of the edge town of the Tier 1 city, they keep pursue us, tried to ask -- invited GDS to buy some land or lock up some resource and then make some future investment in their area. This led us -- and they offer more better deal than before. And they are eager to get GDS in then -- in their new investment -- investment list. So now that GDS has the position to have -- a stronger position to get a better deal, faster deal in the future. But we are still evaluating right now. This is number one. That means, in the future, we try to build our land bank in the future. But it's early -- it's a little bit early to talk about, to inject the capital right now, but we try to maximum squeeze -- the landlord and the municipal government -- local government, and we try to gather the maximum condition for us. So I think it's a good time to get into this project right now, but -- so we are -- we will have a bunch of way to get into better deal. And currently, it's not directly affecting our equity right now. But on the other hand, we are pretty -- we learned a lot from CyrusOne, which they have capability to -- they have the track record to build hyperscale campus-type of data centers. And we have a lot of discussion with them and with their technical team. A lot of the new technology like VR structure, a lot of the prefabrication type of data center skill, we learned a lot from them. This is what we -- the -- is the meaning, when say -- we talk -- we just say talked about, industrialize the future data center to fulfill their hyperscale customer and standardize our design.

[Daniel Newman, executive]

Okay. On Colby's first question about how we think about leverage. The main focus is actually on the leverage at the project level. We inject equity into individual data center project companies and then leverage it with project debt. And typically, it's 60% debt, 40% equity. And then once the data center is stabilized, that translates into around 3x debt to EBITDA. And if you look at our -- the disclosure -- if you look at the data centers which are in service, you'll see many of them have lease rates or commitment rates of 95% to 100%. In fact, I think, 17 have commitment rates that are between 99% and 100%. The customers, the counterparties behind these commitments are almost entirely investment-grade, very well-known companies. And they are entering into 6- to 10-, I would say mostly 8- to 10-year contracts with us. A significant portion of those contracts don't have a right of early termination. They all have very severe penalties. And I think only once in our history has a customer terminated. So if you look at the cash flow that can come from those contracts over multiple years relative to the amount of cash required to service the project debt, that is covered multiple times. So we feel that the way our projects are financed is actually very conservative. However, when you add it all up, we have stabilized data centers, we have ramping-up data centers, we have data centers under construction. The debt drawn down for data centers that have not yet reached optimal level of EBITDA. There is debt drawn down for data centers which are preoperational and have no EBITDA. And there's the aggregate of all that, which you see reflected in our consolidated net debt-to-EBITDA multiple. I think interestingly, as of now, we don't have a single financing facility with a consolidated net debt-to-EBITDA multiple covenant in it. Nonetheless, appreciate that equity investors tend to focus on that metric even though it may not be all that meaningful. And we would like over time for that metric to come down to the kind of levels you see in the U.S. However, for now, with the very significant growth opportunity which we have in front of us, I expect that, that multiple will stay at the kind of levels it's been now and in the last few quarters, which is to say 7x to 8x, maybe closer to 8x.

[Operator]

Our last question comes from the line of Colin McCallum from Crédit Suisse.

[Colin McCallum, analyst]

I'll just keep it short. There's 2 quick ones for me. The first one is just -- I guess, for William. It's regarding the larger customers have finally this year obviously come under a little bit of share price pressure partly because of their margin. They've been missing and their costs a bit higher than people expected. Have they been turning around to you guys and putting further pressure on your lease rates and what they're paying you? And now I realize that, as you mentioned, you're in long-term contracts with them, so obviously, we wouldn't expect something to change in near-term numbers. But for forward contracts, 2, 3, 5 years out, are you seeing any or hearing any pressure from your customer -- from your large customers on your lease rates? That's the first question. Second question, just a quick housekeeping question one for Dan. I saw that the net finance costs increased quite sharply quarter-on-quarter. Were there any one-off costs in third quarter? Or is it just related to the increase in net debt that you alluded to in the prepared remarks? Because I see that the interest rates barely changed. Those are my 2 questions.

[William Huang, executive]

Okay. Colin, thank you for asking question. I mean, for the -- I think that GDS' current strategy is still stay in the Tier 1 market. But in general, I still will say the -- let's say, in the Tier 1 market demand and supply still have that huge gap right now. So we are -- we try to stay in the current price level for -- in short term. For long-term point of view, I think the -- now our customer more smart, very smart. They tried to deploy the difference -- different application in a different type of the product of data center. So different product of the data center -- actually, the price is different. The cost is different. So it's early to talk about the price difference. So our target -- I will repeat that, our target is to maintain our IRR. Because in the last couple -- last 12 or 18 months, we do -- we made a lot of effort to optimize our design, and we also drive down our development costs. So this is let GDS have the ability when one day the price -- we get a pressure from the price -- get a pressure from our customer, we still can maintain our IRR.

[Daniel Newman, executive]

Yes, Colin. On your question about the effective interest rate, there's a simple reason, which is the convertible bonds which we issued, which have a 2% coupon. If you excluded that and we just look at the effective interest rate for the rest of our debt, basically project term loans and working capital facilities, the effective interest rate would have been 6.7%, which is actually almost exactly the same as what it's been over the first 9 months of this year.

[Operator]

Thank you. I'll now turn the call to the company for closing remarks.

[Laura Chen, executive]

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 all.

[Operator]

Thank you. This concludes today's conference call. You may now disconnect your lines. Thank you.