Ladies and gentlemen, thank you for standing by for GDS Holdings Limited Second Quarter 2019 Earnings Conference Call. At this time, all participants are in listen-only mode. After management's prepared remarks, there will be a question-and-answer session. 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.

Thank you, operator. Hello everyone and welcome to the 2Q19 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's 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 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 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. Please go ahead, William.

Hello everyone. This is William. Thank you for joining us on today's call. I'm pleased to report another good quarter with strong results across all aspects of our business. In the second quarter, we signed up customers for almost 21,000 square meters of net additional area committed, or around 52 MW of IT power, which should generate over USD 90 million of annual recurring revenue when fully delivered. Maintaining data center supply in tier one markets continues to be a critical success factor. We made significant progress in securing key resources, both organically and inorganically, enabling us to maintain our resource advantage. During this quarter, we initiated three new projects and today we are announcing another new acquisition. We continued to deliver operationally, resulting in over 50% service revenue growth and over 80% adjusted EBITDA growth year-on-year. Our utilization rate moved up to 70% which drove up our NOI margin to 53% and adjusted EBITDA margin to 43.5%, putting us well ahead of the guidance track. Last, but not least, we are particularly excited about our new strategic partnership with GIC, Singapore's sovereign wealth fund, to develop and operate built-to-suit data centers, initially focusing on a program for one of our top customers. This is a milestone achievement both from a business and from a financing perspective. During the second quarter, our new business is up with the run rate for the past six quarters. We did two big deals of over 20 MW, each with two existing hyperscale customers. We also did one deal over 5 MW from a brand-new large enterprise customer, a market leader in China providing smart phone solutions. We are in a period when there are all kinds of macro factors which may affect the market. We do not know how long it is going…

Thank you, William. Starting on slide 12, when we strip out the contribution from equipment sales and the effect of FX changes, we see even stronger growth and margin improvement than is apparent in the reported numbers. In 2Q 2019, our service revenue grew by 10.6%, underlying adjusted NOI grew by 14.1%, and underlying adjusted EBITDA grew by 15.2% in consecutive quarters. Our underlying adjusted NOI margin reached 53% and our underlying adjusted EBITDA margin hit 44.3%, which is 9 percentage points higher than a year ago. Turning to slide 13, with two quarters completed, our revenue is well up with our expectations, having reached 46.9% of the mid-point of our original guidance. Service revenue growth is driven mainly by customers moving into the space which they previously committed. Move-in during the first half totaled 18,781 square meters. We are expecting move-in during the second half to be slightly higher than in the first half, on top of which we will have additional revenue-generating capacity from Beijing 9 when the acquisition closes. Our MSR has been pretty much flat over the past few quarters. However, we are expecting a 1 or 2 percentage point drop over the next couple of quarters. The decline over the course of 2019 may be slightly less than we were expecting. Slide 14 shows the strong quarterly trend in margin improvement at the underlying adjusted NOI and EBITDA levels. Most of the recent margin improvement has been at the data center level. Last year, we scaled up our operations materially. Now we are seeing the operating leverage on the enlarged base. In the next couple of quarters, we're expecting the NOI margin to stay at around the current level because we have a lot of new capacity coming into service, which will lead to a…

Thank you, sir. [Operator Instructions]. We have the first question from the line of Jonathan Atkin from RBC. Please ask your question.

Thanks. So, I have two questions. One is kind of on the topic of M&A and I wondered Beijing 9 which you announced, what does the pipeline look like over the next several quarters for additional tuck-in acquisitions and would they be on roughly the sort of same scale of 8,000 square meters or would it be markedly larger going forward? And then, on the GIC arrangement, I wondered how soon you would anticipate commencing any additional projects under the JV in addition to what you've already agreed upon in Jiangsu province. Thanks.

Hi, Jon. It's Dan here. Firstly, on M&A, in the past, our business plan for this year was to do one to two M&A transactions and we've now announced two. By the way, Guangzhou 6 has not yet closed. Hopefully, it will close in the next few weeks. And Beijing 9 will close very late in the year. We have an M&A team. We identify a number of targets. We've done diligence on quite a few data centers. We're very selective and we're financially disciplined. We find relatively few that we want to move forward with. I think we can sustain one to two deals like the Beijing 9 deal, that order of magnitude per annum. However, I would highlight that, from time to time larger opportunities come along, Guangzhou 6 and Beijing 9 being acquired from the same seller, second-tier data center operator which had a portfolio with more than 10 data centers. We diligenced them all and we found two out of ten that we wanted to move forward with. Right now, there are one or two situations in the market for small portfolios of data centers or single data center campuses, but there's no certainty that whether we'll proceed with those. But just to give you some color in terms of the kind of opportunity which is out there. With regard to GIC, maybe didn't make ourselves quite clear. We have signed an MOU with one of our largest customers. And under that MOU, we are committed to develop seven data centers at three campuses. So, that means three locations. We mentioned one location in Jiangsu. The other two locations are in other parts of China. And those seven data centers instantly aggregate over 130 MW of IT power. So, that's the extent of the commitment under that MOU right now. But we expect the same customer to have substantially more requirement than that. And, over time, expect the scale of what we undertake through this partnership to increase. We're also in early stage, but in discussions with three or four other customers who have their own equivalent campuses in remote areas and who are all following the trend of seeking to outsource. We're not close to do anything, but the opportunity is there to expand this to other customers as well. Does that answer your question?

Yes, it does. Thank you. And then, just kind of a commercial question. I think you entered this year with nearly 6% or 7% of your area committed coming up for renewal in calendar 2019. And can you maybe provide an update on what you've seen in terms of customer behavior as they review contracts, maybe they depart for various reasons, are you seeing any sort of customer churn beyond the traditional run rate that you have seen? And are the contract lengths that you're signing with enterprises and hyperscalers relatively the same as what we've seen or have there been any changes? Thank you.

Yeah. Jon, we rarely mentioned churn because, fortunately, for us, it's statistically insignificant. In the last quarter, it was 0.2%. occasionally, just every few quarters, there may be churn event, which results from some change in our customer's own organizational or operations. But there isn't sufficient churn to be able to characterize it in any way. With regard to contract length, most of the hyperscale deals now are in the 6 to 10 year band. Actually, most are in the 8 to 10 year band. I'll just take this opportunity to make a comment. When these contracts are signed, it's almost invariably pre-commitment. These days, quite often, precommitment one year before the data center comes into service. So, some of the contracts have no right of early termination, for the customer to terminate early at any time over the life of the contract except, of course, if there's a serious failure in performance. The other contracts where they do have an early termination right typically only kicks in after the end of the move-in period. So, that would be, say, two years after the beginning of the service delivery. And then, there is a very severe penalty. It's tens of percent of the total contract value. So, our backlog really is very solid term in terms of underpinning our growth.

Thanks. And maybe a question for William in terms of just demand trends you're seeing, whether it's gaming or AI, cloud, social networking or enterprise, but any particular industry verticals or types of companies where you're seeing demand trends notably different than three months ago?

Yes. So far, we didn't see any change, especially on our customer base. But what we can tell is AI is overwhelmingly deployed in the different verticals right now. So, I think that's why our customers, their demands for power capacity at each order, the size is much bigger than before. Yeah. So, I think the current three key drivers in our view are: power is still the number one driver. And the second is Internet company and enterprise, especially the financial institution still maintain very strong demand right now. And the demand profile in a single order is much bigger than before.

Great. Thank you very much.

[Operator Instructions]. We have the next question from the line of Yang Liu, Morgan Stanley. Please ask your question.

Thanks for the opportunity to ask questions. The first one, I think Dan just mentioned that the debt financing environment is quite favorable for GDS now. Could you please give us some numbers in term of the current debt interest rate or refinance interest rate compared with the previous term from the banks? The second question is, when GIC acquired the 90% stake of the build-to-suit data centers, what is the premium in term of the valuation compared with the cost of GDS? Thank you.

Hi, Yang. It's Dan again. First of all, the current cost of debt, we've done some refinancing of data centers recently with Chinese banks. And, in fact, new relationships. And we've also done a financial lease. The all-in costs came to less than 6%, which is three quarters, almost 1 percent point lower than it was a few quarters ago. But, I stress, that is for refinancing of the data centers where we would expect to get a slightly lower price. We also got longer tenants. We've got 8 and 10-year tenants, which is quite exceptional for project term loans with back-ended amortization. So, really, just about as good as it gets, I think. Yeah, you asked about the premium that we pay for acquisitions. In a way, Yang, you can figure it out, right, because it cost us about $5 to get $1 of EBITDA. When we do these acquisitions, it costs about $8 to get $1 of EBITDA. So, you see the premium is around 50%. But in some ways, it's kind of academic because we've done the acquisitions because there hasn't been an opportunity for us to do the project organically. Sorry, can you repeat the question? Someone was talking to you when you were asking the question?

Yang, can you repeat the second question? Hello.

[Operator Instructions]

Operator, can we just finish answering the question?

Sorry, everyone, excuse me. So, if I understand correctly, the question was, what is the premium when we sell equity interest in a project – build-to-suit project to GIC? Yeah. The premium is 8%. If we've incurred financing during the construction phase, that will enable us to recover our financing costs.

Mr. Yang, your line is open, sir.

I have finished my questions. Thank you.

[Operator Instructions] We have the next question from the line of Frank Louthan from Raymond James. Please ask your question.

Great, thank you. The current guidance that you've given out, does that include anything for the build-to-suit projects with GIC, either on the revenue or the CapEx side? And then, secondly, you generally get a rolling look at your customers' business over a several year period, updated throughout the year. Any change in how they're looking at their business over the next few years going forward based on the current US trade situation? Thank you.

Yeah. Hi, Frank. The guidance does include the CapEx for the assets, the build-to-suit assets that we are incubating, if you like, during the construction period, but it's only a few hundred million. In terms of income statement, once we transfer the 90% equity interest and then the data centers come into service, during the first year, the customer is moving in and has some flexibility about how fast they move in. So, we wouldn't be recognizing any significant service fees probably until 9 or 12 months after the data center comes into service. So, there will always be that time lag from when we complete the project. Just going back to my previous answer about the CapEx, of course, when we transfer the 90% equity interest to GIC, the CapEx that we've incurred will then be reversed or 90% of it will be reversed.

Frank, the second question is customer's rolling demand, right? So, I think, typically, the big customer will assure their three years' demand to us, especially with the larger cloud and Internet company. So, we are pretty focused on keep talking to them regularly.

Okay, thank you very much.

[Operator Instructions]. We have the next question from the line of Robert Gutman from Guggenheim Securities. Please ask your question.

Thanks for taking the question. So, just curious, on the MSR, which is looking better than you had originally anticipated, I think guidance for the year was a decline of 5% year-over-year. What's underlying the fact that it's coming in a little better?

Rob, when we talked about 5%, I was talking about 4Q 2018 to 4Q 2019. I talked about 5%, hopefully, it was going to be a little bit less. And I'm hopeful it will be a little bit less. When we look at the average MSR for the whole of 2019 compared with the average MSR for the whole of 2018, we're still looking at something close to about a 5% year-on-year decline if you calculate it that way.

Okay. Thanks. And then, just in terms of the strong results in the quarter in terms of revenue, would you say it was more from sort of just faster move-ins and sort of a pull forward or was it greater-than-expected sales in the quarter with immediate commencements?

Actually, Rob, the revenue was pretty much what we were expecting internally. It's tracking the top half of guidance at least. And that's what we forecast. What is surprising was – which even surprised us was the profit, the EBITDA or the NOI growth, the amount of operating leverage we've been able to realize, that did exceed. That has exceeded our own expectations. Last year, we, I think, increased the headcount by 20% to 25% because our business has gone from 40,000 square meter net add business to an 80,000 square meter net add business on an annual basis. So, we had to scale up to take account of that. And then, since early this year, we've barely increased our operating cost base, and that's why it's come through in very sharp margin improvement.

Great. Thank you very much.

[Operator Instructions]. We have the next question from the line of Gokul Hariharan from J.P. Morgan. Please ask your question.

Yeah, hi. Thanks for taking my question. My first question, Dan, could you talk a little bit about how much further operating leverage there is likely to happen over the next year or so, given you've seen a pretty strong operating leverage improvement over the last four to five quarters? Second question I have is maybe for Dan and William. When you talk about 80,000 square meter pipeline capability in terms of every year potential new pipeline coming in and the ability to prospect that and build it out, now with this GIC partnership also, would that number start to go up and be largely dedicated to tier 1 and satellite sites or would some of the purpose-built site be also included in this 80,000 square meter ability to furnish each year?

Yeah. Hi, Gokul. We took a lot of operating leverage. We always look at it at two levels. Firstly, the data center level where we look at the margin for the data centers has stabilized and activity, I'd say, it's 55%. Of course, it's a little bit higher than that. And then, we have the dilution effect or dampening effect from data centers which are ramping up. And, over time, as the balance has shifted to greater proportion being stabilized, that's been raising the margin. And I think, you said over the next few quarters because I think in the next couple of quarters, I expect the NOI margin to stay around the current level. And going into next year, we could be looking at least another 1 to 2 percentage point improvement. I think if we took our operating leverage at the SG&A level, my ambition is to get SG&A down to 5% of revenue because that's lower than any data center operator has ever disclosed, right? So, that would indicate that we've got 3% to 4% still to go. But that will take some time, but I think we continue to make steady progress in that direction. The second question about the 80,000 square meters, let me make sure I understood correctly and don't give the wrong answer again. The 80,000 square meters refers to what we do in tier one markets. We're not including the first project or the future projects that we undertake through this joint venture in that 80,000 square meter number. If we did, I would be adding another 7,000 square meters to our area committed because that's what we have in project number one in Jiangsu. So, that's not been in any numbers that we've talked about before. It's entirely additive.

Understood. So, just wanted to ask, is there any discussion or any interest in some kind of partnership, like, for some of your satellite to tier one kind of city projects as well or is it something that, on an economic basis, GDS feels that it's better served to actually fulfill them on its own?

Well, our business plan and our capital raising is based on what we see in tier one markets. I think we're well capitalized for the opportunity in tier one markets. But as I commented during the presentation, a lot of work went into developing the structure for this partnership with GIC and large customers. William said it's really a three-way partnership. The customer had to accommodate as well. And having done it now, but adapted specifically for these build-to-suit projects in remote areas, yeah, it's certainly something that is in the back of our minds that we could deploy a similar structure, potentially with different ownership level and we could be – deploy it with us having the majority and consolidating. Or we could do it with us having a larger stake. But, conceptually, it's the same. The differential return is a result of having an equity investment, plus a management and operating fee. I think, in this industry, given that a very large part of the demand is coming from aggressively shortlist 15, 20 very large customers, it's important to be able to access the lowest cost equity capital. And that's not always the public equity market. So, having established a channel – we're, of course, establishing a partnership with a brilliant partner in GIC – I think that positions us very well to be able to look at our requirement and see whether it's best to use our own equity, which effectively comes from the public equity markets, or whether it's best to use the equity of a partner like GIC. So, I don't rule that out at all in future, but it's not something that we're specifically contemplating in terms of any actual situation right now.

Okay. Just if I may ask a little bit more on the demand situation, I think previous caller alluded to your view on the demand. A lot of your customers on the hyperscale side have had a tough situation in their current businesses. The future businesses are still growing. So, has any of that really played into any of the capacity planning discussion, pipeline planning discussions that you had with them? And also, from GDS own perspective, how do you handicap any of those kind of [indiscernible] into future planning?

Yeah. Let me just summarize for William. So, Gokul was asking, because of the, supposedly – Gokul thinks that the hyperscale customers in China are maybe having some challenges in their own cloud businesses. That may or may not be correct. Does that come through in terms of what we see with resource capacity planning? Any changes in capacity planning and how do we adapt to that?

I think, currently, what we commit to the markets, it's 80,000 square meter, right? We can repeat that. That's what we are having seen, right? I think this will not change. And because we have very strong customer base and our customer base come from the different verticals. And even in the cloud side, we have all kind of the cloud in China. So, I think our customer base, whatever from the vertical point of view or industrial point of view, we are quite diversified. That's how we managed our demand and certainty. So, I have to say we will not change our CapEx plan. In other words, we are comfortable to deliver another 80,000 every year.

We have the next question from the line of Colin McCallum from Credit Suisse. Please ask your question.

Thanks for the opportunity. Hopefully, you guys can hear me okay. Actually, I have a difficult question for you. It's kind of fundamental question. I'm just wondering, on this GIC transaction, why would you choose to do it this way and only be taking a 10% stake, plus service fee on the GDS side? Is it – you alluded a couple of times to remove areas. Is it that you view these areas or the risk attached to these areas or this customer in particular being way above what you think your public shareholders have kind of signed up for? Or is it an issue just with finance raising? Or return that you would expect from these data centers? Because you kind of suggested that the returns is okay. And anyway, the customer is reliable. Therefore, the risk wouldn't be so bad. Then, you've said earlier on that the financing side is very favorable at the moment. So, just intrigued why you've decided to do this, particularly with really such a small equity stake for the current shareholders of the business.

Yeah. Hi, Colin. It's a good question. First of all, let's be clear, the kind of project we're talking about is totally different from our mainstream business. These are build-to-suit projects on sites where, in this case, typically, the customer actually owns the real state, owns the power infrastructure and is outsourcing the design, the construction, fitting out and the long-term operation of the data center. So, in that respect, it's a different product entirely. It's a build-to-suit data center at a customer's site. Secondly, we look at the quantum or the volume, I've always said that our value is in fulfilling customers' requirement for somewhere to locate their latency sensitive data and applications. Customers have a huge requirement, which is not latency sensitive as well. So, the volume of what gets put into remote locations is very substantial. And our customers are asking us to follow them there. And that's not part of our business plan. It's not part of what I said earlier. It's not part of what we've planned for in terms of our capital raising. It requires a lot of additional capital. And for relationship reasons, we want to be able to do this and nobody else can do it. If we can do this, as well as what we're doing in tier one markets, that gives us an even more edge and – to use a term – an even greater moat. The third point is about – yeah, undeniably, it's about the financial returns. We started off doing three projects for our customer in Hebei. Of course, in this kind of situation, the customer wants to outsource, but the pricing is low and the returns are lower. In the case of Hebei, we made it work from a financial perspective, by doing it entirely…

Understood. No, that makes a lot of sense. Thank you.

Ladies and gentlemen, as there are no further questions, I'd like to now turn the call back to the company for closing remarks.

Thank you once again everyone for joining us today. If you have further questions, please feel free to contact GDS Investor Relations through the contact information on website or the Piacente Group Investor Relations. Thank you.

This concludes this conference call. You may now disconnect your lines. Thank you.