Switch Inc (SWCH) 2021 Q1 法說會逐字稿

Good day, and welcome to the Switch, Inc. First Quarter 2021 Earnings Conference Call. (Operator Instructions)

Please note that this event is being recorded. I would now like to turn the conference over to Matthew Heinz, Vice President of Investor Relations. Please go ahead.

Thank you, operator. Good afternoon, and welcome to Switch, Inc.'s First Quarter 2021 Earnings Conference Call. On the call today are Thomas Morton, Switch's President; and Gabe Nacht, Switch CFO. Today's call may include forward-looking statements, including references to expectations, projections or other characterizations of future events or market conditions. Actual results may differ materially from those expressed in our forward-looking statements, which are subject to certain risks, uncertainties and assumptions.

Our statements are made as of today, and we assume no obligation to update our disclosures. We describe some of these risks in our SEC filings, specifically our Form 10-K in the section entitled Risk Factors. In addition, today's call includes discussion of non-GAAP financial measures, which should not be considered in isolation from or as a substitute for financial information prepared in accordance with GAAP.

Please refer to today's press release and supplemental package for further information, including a reconciliation of non-GAAP measures. Our first quarter 2021 earnings press release has been furnished to the SEC as part of our Form 8-K and is available on our Investor Relations website at investors.switch.com. I will now turn the call over to Switch President, Thomas Morton.

Thank you, Matt, and good afternoon, everyone. Thank you for joining us today for our first quarter 2021 earnings call. Switch is pleased to report another strong quarter as we continue to execute on key business initiatives driven by strong customer demand. Our first quarter 2021 financial results reflect continued business momentum and improved operating efficiency across all of our PRIME campus locations. First quarter revenues were $130.9 million, increasing 5.5% year-over-year when adjusted for nonrecurring fiber transactions compared to the year ago quarter.

Adjusted EBITDA was $73.4 million, representing 19% year-over-year growth with an adjusted EBITDA margin of 56.1%. Excluding the benefit from $2.8 million of nonrecurring license fee income, the first quarter adjusted EBITDA margin was 54% representing 600 basis points of year-over-year margin expansion. We continue to see an elevated level of demand across our business and our sales team once again stepped up to deliver a strong bookings quarter, coming off of a record quarter 2020. In the first quarter, Switch signed $18 million of incremental recurring revenue and total contract value of $117 million.

This marks our third consecutive quarter with at least $18 million of incremental annualized revenue bookings and more than $100 million in total contract value. We believe this consistency in sales execution further highlights the value of the strategic hires that were made to bolster our sales force as these individuals have made a significant contribution to our organization since coming on board in 2019.

We are also pleased to report that Switch ended the first quarter with a record recurring revenue backlog of more than $57 million, eclipsing our previous record of $50 million set last quarter. The demand across all PRIMES came primarily from existing customers who continue to expand their footprint with Switch. As discussed last quarter, an elevated volume of multi megawatt transactions over the past several months has resulted in large space and power reservations that have limited the amount of contiguous and immediately sellable inventory. Given these near-term inventory constraints, we continue to be strategic about how we allocate capacity and approach our customer technology requirements.

We maintain a proactive dialogue with customers with reserved cabinet space and power to optimize the utilization of that infrastructure. One such instance in Q1 resulted in a 2-megawatt power upgrade by a customer in the Core Campus enabling us to monetize power without reducing physical space or cabinet inventory. As previously disclosed, Switch recently executed a definitive agreement to acquire data foundry for $420 million in an all-cash transaction, the first major acquisition in our company's history.

This multiphase strategic transaction will provide immediate capacity to sell in both the Austin and Houston markets, but more importantly, provides the foundation to launch our fifth PRIME campus. We believe this expansion of our footprint into the Texas market creates a long runway of growth and strong returns for our shareholders due to the immediate benefit of geographic diversification, the addition of new strategic customer relationships and the positive underlying dynamics of technology and enterprise migration to the Central Texas corridor.

I would now like to discuss some of Switch's notable first quarter activity and key metrics across the existing PRIME campus locations. In the first quarter, Switch added 24 new logos, accounting for approximately $2 million of annualized recurring revenue and $6 million in total contract value with their initial deployments in our ecosystem. We completed a 5-megawatt expansion with a Fortune 100 semiconductor company at The Citadel Campus, totaling $7 million in annualized revenue and $38 million in contract value.

In the Core Campus, we signed a 3-year renewal and expansion with a Fortune 500 biotech firm, totaling over $9 million in contract value. And also completed a 2-megawatt expansion with an existing hyperscale cloud customer, totaling $6 million of incremental contract value. In The Pyramid Campus, a leading regional bank customer executed a multiyear expansion worth over $4 million in incremental contract value. In February 2021, Switch and its joint venture partners closed on the sale of SUPERNAP Italia to IPI Partners, resulting in a $5.4 million gain in Q1.

As part of the same transaction, Switch increased its equity interest in SUPERNAP Thailand, resulting in a 30% interest in the Thailand asset while also maintaining control over our international intellectual property license. Lastly, in late April, Switch issued its annual ESG report, which is prepared in accordance with the GRI, SASB and TCFD reporting frameworks. Some of the key highlights from the report include: continued environmental leadership with 100% renewable power and 0 scope 2 emissions across all data center operations since 2016, a 40% ethically diverse workforce, nearly 1:1 parity in our male-female employee pay ratio, a Board with 3 out of 8 female directors and Board committees comprised of 100% independent directors.

Now turning to our construction milestones and project pipeline. Following the receipt of all necessary construction permits in late Q4, we have continued to accelerate construction on several projects within our development pipeline. We currently expect to complete construction on 3 new data centers in the Core Campus, The Citadel Campus and The Keep Campus between Q2 of 2022 and Q2 of 2023.

At full build-out, these 3 facilities will total 1.3 million gross square feet and up to 160 megawatts. As these facilities are constructed, we are also completing site preparation work on 4 additional facilities, including 2 in the Core Campus, 1 each in the Citadel and Keep Campus locations. These assets have planned completion dates ranging from 2024 through 2026 and will comprise an additional 1.6 million square feet and up to 200 megawatts upon full build-out.

We have provided new disclosures on Page 7 of our investor presentation that details our pipeline of active and planned data center construction over the next 5 years. Additionally, realizing that in the current times, site visits are not as feasible, we have added a library of construction photographs to our investor website, showing the progression of our ongoing development efforts in the Core, Citadel and Keep PRIMEs. I will now turn the call over to Gabe to discuss our financial results. Gabe?

Thanks, Thomas. Today, I'm going to review our financial results for the first quarter of 2021 and discuss our outlook for the remainder of 2021. Switch reported first quarter 2021 revenue of $130.9 million, an increase of $2.8 million or 2.2% compared to the first quarter of 2020. Excluding the impact of nonrecurring fiber transactions, first quarter revenue growth was 5.5% year-over-year. Colocation revenue for the first quarter of 2021 was $107.3 million, up 6% compared to $101.2 million in the year ago quarter.

Connectivity revenue was $21.9 million, increasing 2% sequentially and 3% year-over-year when adjusted for nonrecurring fiber revenue. Other revenue, including professional services, accounted for $1.7 million in Q1 of 2021, essentially unchanged compared to the same quarter last year. Importantly, the customer revenue reductions initiated in Q4 were fully absorbed in the Q1 run rate, and we do not anticipate further material events of this type in 2021. As of March 31, 2021, Switch had approximately 17,200 billing cabinet equivalents, reflecting 600 net cabinet additions compared to the prior quarter.

Our average monthly recurring revenue per cabinet also increased on a sequential basis, exceeding 2,500 in Q1 of 2021. This metric can be influenced quarter-to-quarter based on the timing of large installations within the period. However, the underlying trend toward increased power density demanded by our customers continues to support strong unit economics in our business. We had more than 8,900 billing cross-connects as of March 31 and cross-connects accounted for 4.1% of total revenue in Q1 2021. This compares to 3.7% in the year ago period reflecting 12% growth in cross-connect revenue.

Now turning to bookings. During Q1, we executed 534 contracts comprising approximately 11 megawatts, representing total contract value of $117 million and annualized revenue of $37 million at full deployment, inclusive of both renewals and sales of incremental services. Excluding renewals, these signings represent $18 million of incremental annualized recurring revenue, including $16 million in incremental bookings from existing customers and approximately $2 million from new logos.

As of March 31, 2021, our recurring revenue backlog stood at just over $57 million, up from the prior record of $50 million set last quarter. We expect our backlog to contribute approximately $16 million of incremental revenue during 2021, with the remainder contributing in 2022 and beyond. Customer churn was 0.1% in Q1 2021 compared to 0.4% in the year ago quarter. As a reminder, we define churn as the reduction in recurring revenue attributable to customer terminations or nonrenewal of expired contracts, resulting in a full customer exit from the Switch platform, divided by the revenue at the beginning of the period.

First quarter cost of revenue increased by $4.7 million compared to the year ago quarter primarily due to an increase in depreciation. Excluding depreciation, amortization and equity-based compensation, our Q1 2021 adjusted cost of revenue decreased by 5%, driving a 190 basis point improvement in adjusted gross margin and a 5% growth in adjusted gross profit, which increased to $97.3 million.

First quarter SG&A expenses were $35 million, down from $40.1 million in the year ago quarter. This 13% year-over-year decrease in SG&A was primarily attributable to lower professional fees and general and administrative costs. I would note the year-over-year SG&A cost comparisons beginning in Q2 of 2021 will be less favorable as we lap the 1-year anniversary of COVID protocols. Q1 2021 income from operations increased 15% to $24.2 million compared to $21 million in Q1 2020.

Growth in operating income was attributable to the $5.1 million reduction in SG&A. Interest expense increased by $1.3 million to $8.8 million in Q1 2021, primarily driven by higher debt balances, partially offset by lower LIBOR rates compared to the same quarter last year.

As of March 31, 2021, we had approximately $1 billion in total debt outstanding at a weighted average interest rate of 4.1%, inclusive of interest rate swaps on our $400 million term loan balance. First quarter net income was $24.4 million compared to a net loss of $3.5 million in Q1 2020.

The growth in net income was primarily attributable to a $3.2 million gain on interest rate swaps in Q1 2021 compared to a $17.6 million loss on the swaps in the year ago quarter. Other significant items impacting net income include a $5.4 million gain from the sale of our interest in SUPERNAP Italia and license fee income of $2.8 million related to an agreement to use Switch's intellectual property for the construction of data centers in Saudi Arabia.

Adjusted EBITDA totaled $73.4 million for Q1 2021 compared to $61.5 million in Q1 2020, reflecting year-over-year growth of 19.4%. Our adjusted EBITDA margin for Q1 2021 was 56.1%, increasing 810 basis points from the year ago quarter. Excluding the $2.8 million in license fee income, adjusted EBITDA margin was 54%. Adjusted funds from operations, or AFFO, was $60.9 million in Q1 2021, a 26% increase compared to $48.5 million in the year ago quarter. AFFO per diluted share was $0.25 compared to $0.20 in Q1 2020. Please refer to the appendix section of our investor presentation for a reconciliation of net income to AFFO. Maintenance capital expenditures were $2.1 million for the first quarter of 2021 or 1.6% of revenue compared to $1.3 million and 1% of revenue in the same quarter last year.

Growth CapEx for data center construction and improvements was $98.3 million for the first quarter of 2021 compared to $78.9 million in the year ago quarter. Please refer to our press release and investor presentation for a detailed breakdown of capital expenditures by campus. As of March 31, 2021, the Switch PRIMES had capacity for 24,200 cabinet equivalents within our open sectors, of which 91% were committed under contracts compared to 88% in both Q4 and in the year ago quarter. Please refer to Page 7 of our investor presentation for utilization rates across each PRIME campus.

Looking now at the balance sheet. As of March 31, 2021, the company's total debt outstanding net of cash and cash equivalents was $1 billion resulting in a net debt to last quarter annualized adjusted EBITDA ratio of 3.4x, unchanged compared to the prior quarter. As of March 31, 2021, Switch had liquidity of $538.9 million, including cash and cash equivalents and borrowings available on our revolving line of credit.

As we mentioned in regards to the data foundry acquisition, we are currently evaluating financing options to fund the acquisition. We expect to finalize our issuance of new debt securities in the coming weeks prior to closing the acquisition. As of March 31, 2021, there were 241.5 million total shares outstanding, including 127.5 million Class A shares and 113.9 million Class B shares. As disclosed in recent 8-K filings during the first quarter of 2021, our members redeemed 7.7 million common units, resulting in the issuance of an equivalent number of Class A common shares. An additional 3.3 million Class B shares have been redeemed in the second quarter of 2021, bringing our Class A public float to 54% of total shares outstanding.

Now turning to guidance for 2021. We are increasing revenue and adjusted EBITDA guidance to reflect our stronger-than-expected first quarter results and continued expectation for a ramp in customer installations toward the second half of 2021. I would note that this guidance represents Switch stand-alone expectations and excludes any contribution from our announced acquisition of Data Foundry.

We intend to update guidance accordingly after closing the acquisition. Revenue in the range of $543 million to $555 million, reflecting 7.3% organic year-over-year growth at the midpoint. Adjusted EBITDA in the range of $281 million to $290 million, reflecting an increase of 6.4% compared to 2020 and an adjusted EBITDA margin of 52% at the midpoint. Lastly, our guidance range for capital expenditures, excluding land acquisitions, remains unchanged in the range of $330 million to $370 million.

We continue to have a significant amount of space and power reserved for large customer deployments in the second half of 2021, although installations have paced slightly ahead of their contractually committed time lines year-to-date. We remain optimistic the revenue growth will continue to accelerate throughout the year, returning to a more normalized level of growth as we exit 2021. And now I will turn it back to Thomas for some closing remarks.

In conclusion, we firmly believe that Switch is favorably positioned for the rapid digital transformation among enterprises as they continue their migration to hybrid multi-cloud architectures. We are working hard to accelerate delivery of additional data center capacity to meet the strong level of demand we are currently experiencing, and we are confident in our team's ability to execute. We would once again like to take this opportunity on behalf of our management team to thank our employees, customers, partners and our shareholders for their continued support of Switch.

We would now like to open the line for questions.

(Operator Instructions)

And our first question today will come from Sami Badri with Crédit Suisse.

Noticed there's been a lot more commentary or messaging around the international opportunity. And I kind of just wanted to revisit what the Switch strategy really is going to be, at least going forward. And I know you just made the Data Foundry intense acquisition. I just want to double check here. Are you guys going to potentially consider international targets? Or will the international opportunities look a little bit similar to what you guys are already just doing today?

Sami, you're right, we just did the Data Foundry acquisition, which means that we've expanded from 1 to now 5 campuses since we went public. So that's a pretty aggressive rate of expansion. We did rightsize the international attributes that we had. And we have no problem with looking at other international opportunities should the need arise, but the -- or the opportunity arises. But currently, we are focused on making our fifth PRIME a success and also being successful in developing the 1.3 million square feet that we have under development at 3 of the 4 of the other PRIMES.

Got it. And then the other question I have is regarding connectivity. And specifically, the part that is maybe under more pricing pressure, the part that probably more indirectly competes against other IP transit providers or connectivity providers. Can you just give us an idea on just levels of pricing or pricing pressure that Switch may be seeing in the market?

Yes. The -- well, telecom pricing is always a decreasing pricing model, right? Every time when a customer renews, they renew at a lower price. I mean, I can remember when the gigabits were almost $100 per gigabit and now we're buying them for less than $1. So it has obviously been a decreasing. But what happens is people deploy new technologies and those technologies such as WAN, result in larger margins, and they increase the amount of bandwidth they need and the increase in the amount of bandwidth have outpaced the decreases in prices.

Also, we continue to increase the number of people participating in the telecom cooperative. So as we bring in new revenue and as current customers expand the revenue, this is a growing revenue item for Switch. And overall, we've been able to maintain a fairly consistent profit margin off of those customers.

Sami, this is Gabe. As Thomas mentioned, telecom is a declining price product. It always has been. So when you look at our telecom growth over time, it has remained 17% to 18% of our revenue for several years, which means it's keeping pace with our co-location growth despite the fact that unit economics decline on -- in the telecom space. So customers are making up for that unit economic decline by expanding their networks and buying more capacity and buying more circuits. So the fact that it remains 17%, 18% of our overall revenue is fantastic given the dynamics of the telecom industry.

And our next question will come from Erik Rasmussen with Stifel.

Maybe just as it relates to your guidance and specifically EBITDA, you raised the midpoint by about $1.5 million, but given the meaningful outperformance this quarter, the run rate would have probably suggested a higher number. Is there anything to read into this? Or what we should expect for the remainder of the year? That might be keeping a lid on your expectations there?

Sure. Yes, 2 things.

Go ahead.

Go ahead, Gabe. No. This is one of the things about being remote. So 2 things, and then I'll let -- I'll turn to Gabe to supplement. The first is that as COVID restrictions continue to ease, there is the potential for our team to do more travel. And if that happens, then there will be an uptick slightly in expenses regarding G&E. The second thing is that in Q3, we tend to see an increase in power costs, and we always want to know that, that's out there and be cautious about increasing too much with that coming out. As you've seen last year, we had the winter storms in Texas. The year before that, we had the Cal ISO issues with the pipelines and the fires in California. So in Q3, Q4, there is some uncertainty with respect to power costs. So we want to just be cautious about how much we raise in the front end of the year, knowing that there is an opportunity for power cost changes or dynamics in the back end of the year. Gabe?

You hit them both, Thomas. We're good.

Thank you, sir.

Great. And then in your prepared remarks, you talked about supply constraints and limitations due to lack of having that contiguous space. I guess that's suggestive of there's a lot of larger deals coming through the pipeline. But what can you do -- can the team do in the near term to sort of ease some of that? And obviously, your Data Foundry acquisition could help alleviate some of that, but also the robust development pipeline, it seems like there's a lot that will be coming online towards the latter part of the year.

That's right. We're doing everything we can to accelerate development as much as we can. But everybody has to remember at the end of the day, this is bricks, mortar, concrete and steel. It does take some time to develop. So we are moving as quickly as we can on all 3 fronts. We have also purchased a fifth PRIME, which will have the ability to bring some more opportunity online as we close the acquisition to begin to build out the remaining space in Data Foundry.

So we are working to expand our footprint as fast as we possibly can, and we feel the need to bring more space, power and cooling online as fast as humanly possible.

And Erik, just to clarify that the next large building will be coming online at the beginning of 2022. So it's slated for Q2 of 2022. And with another 2 large buildings coming online at the beginning of 2023. So we have about 1.2 million square feet coming online between '22 and '23. But we're building as fast as we can. Data Foundry, we'll add some capacity but the large expansion will hit in 2022.

We've also -- I think we mentioned -- I mentioned in my comments that we have added progress slides onto our IR website so that investors can see what's going on in the 3 campuses that we are currently deploying on and see the expansion as that begins to evolve.

And our next question will come from Aryeh Klein with BMO Capital Markets.

Maybe just following up on that. Is there a little bit of a risk of maybe an air pocket where you don't have enough available capacity at leasing flows given that a lot of the capacity isn't coming on till next year and the following year?

Yes, we believe we haven't -- well, we've given our growth rate for this year given the available capacity. We have capacity coming on next year. We've also bought Data Foundry, which will provide some incremental capacity for us to grow into. So we believe that we'll be able to continue our growth trajectory through this year and into next year. And then as additional buildings come online, we will begin to enhance that growth because we'll have the opportunity with additional inventory.

Okay. And then it's been, I guess, a week since you announced the Data Foundry acquisition. Can you just talk to the customer response since then? What have you heard from them? And any early leads that you can talk to?

Yes. We have had some customer inbound calls asking us about our availability in Texas and if there's space that we might be able to sell. Those come from existing customers and a couple of new customers. So that is encouraging. We have had, as part of the diligence process, calls with a number of Data Foundry's largest customers. Those calls all were received very proactively, very well. And so we do not expect there to be a significant amount of client flux in the course of this transition.

Okay. And just a quick clarification. Were the nonrecurring licensees previously included in the EBITDA outlook?

They were not.

And they are now. The guidance?

Correct. Yes, they are.

And our next question will come from Richard Choe with JPMorgan.

In the disclosures for the long-term plan development. I guess, to some of the earlier questions, maybe the development or strength you're seeing near term is a little stronger than you expected and you need to build a little bit faster, not for just '22 but also '23 and beyond that, which is kind of far away. What kind of -- is this a change in the way you're addressing the business or -- and CapEx should run relatively high for the next few years? Or is this always kind of planned?

Yes, I'll let Gabe address the CapEx. But what we had is, in the end of 2020, actually the last 2 quarters in 2020. We had some very large sales. Those sales committed space while the revenue rolls in this year as we chew through that backlog. That space is committed and therefore, not available to be sold. That reduced the amount of available inventory that we have to sell in 2021. There was also, as we've spoken about a slowdown in the issuance of permits, which has been relieved, which delayed new buildings coming online slightly.

We have been working day and night as fast as we can to bring more product online. As fast as we can so that we can serve the market and the customers that we are being asked to serve. And we've also expanded to a fifth PRIME, and we'll begin construction there to expand those properties as well. So it's -- in a lot of respects, it's a good issue to have because we have a temporary bottleneck due to robust customer demand and the fact that people would really like to be located inside of our data centers and receive our services. As to the CapEx trending. Gabe, do you want to comment on that?

Yes. And Richard, when you say -- when you ask, is it a change, it's not a change certainly since our last quarterly call when we talked about the need to build as fast as we could because our space was being committed based on the bookings that we were signing.

So if you look at the last 3 bookings, we had $18 million in incremental bookings in Q3 of 2020, and $36 million which was a huge quarter for us in Q4 and another $18 million this quarter. So we've had a lot of large contracts that have ramps. So as Thomas alluded to, that space is gone. It's committed. So revenue will ramp in over the back half of this year and into 2022 and in some cases beyond, but the space is committed.

And so we have to build as quickly as we can. So it's not a change from the last call when we talked about the fact that we were building as fast as we could in Las Vegas, in Reno and in Atlanta. Atlanta is 70% committed from a power perspective already, which means, again, we have to build. So we're going as fast as we can. CapEx, we've given guidance for this year. We haven't given any guidance for next year, but we'll continue to build as quickly as we can until those Reno and Atlanta facilities come online. And then, of course, based on customer demand, we'll look at the longer-range plan. The things that you're seeing on our -- Slide 7 of our investor deck under the planned development for 2024 and beyond, we'll scale those based on customer demand.

Got it. And a quick follow-up with the campus that are highly or almost fully committed. I guess, are you approaching those with just leaving that space that whatever is available for existing customers? Or how do you balance that with new customers at this point?

Well, more than 70% of our incremental demand each year comes from existing customers expanding. So they assume that they will take a portion of the space that we have. But we added 24 new logos this quarter, and we are always looking for new logos to continue to enhance and increase the robustness of the ecosystem. So we're not reserving any cabinets for anybody in particular. It's just the first person to place the service order will receive those cabinets.

That's correct. We have some customers. Once they sign an order, if it includes a ramp, and that space is committed and locked. But unless there's a service order for specific cabinets, it's sellable, and we will sell it to both existing customers as they expand and new customers as they move in. And we do have customers that have rights of first refusal. And as space is filling up, we're asking people to either make a decision on that or let that space go.

And our next question will come from Colby Synesael with Cowen.

Great. I guess, first off, AFFO is up, I think you said 26% year-over-year. I was wondering if you could talk about what's driving the magnitude of growth that we're seeing and any color on what that might look like for 2021. And I guess as part of that, these ramp-ins that we're talking about, what does a typical ramp-in look like? Is there such thing as a typical ramp-in? And of the space that's utilized, including just commitment to the space but not necessarily installed. I mean, how much do you typically give? Is there a rule of thumb or we could say, for every deal, there's an extra 10% of incremental inventory that you're holding on behalf of your customers? Just trying to get a sense of how we should start to think about that.

Yes, Colby, I'll answer on the incremental cabinet space. As Gabe mentioned, we have customers that have reserved cabinets that are committed because they have signed ramps to load into that, and we've built that into our backlog. Unless they have signed and committed to take the space, we do not hold space for people. There are people that have brokers. But what we do is we can give them notice that somebody else would like their space and they can either take it or release it back into inventory. So -- but there's a like...

Is the ramp like -- but do you get paid while you're waiting for the ramp? Do they actually have to install before you get your first dollar for that space? Is that ramp typically 6 months? Or is it typically a year? Just any other color like that would be helpful.

No, that's great. There's 2 ways that people can do it. Some people pay -- it's built into the price that they ultimately pay for the space. The way that, that is currently architected, either it's in the full price per KW, and it's amortized over the term. We figure we want to have a target EBITDA margin on that deal.

And based on the pricing and the ramp and the way we're holding space, we will know what sort of charge we want to make to that customer. The other way that people can do it is they can pay a reservation fee, while they decide whether or not they're going to ramp into that space.

And they will pay that reservation fee for a designated period of time. And at the end of that period, they've either exercised it or choose to exercise it at the end or it's released back into inventory. Those are the 2 structures that most people have with respect to reserved space. Gabe, do you have any comments on the AFFO question? Okay.

Yes. But first, let me chime in on the ramp. Colby, I think as we talked about in last quarter's call, historically, customers would ramp in. If it was a large deal, they would start installing in, say, month 3 and they'd ramp in over a 12- or 18-month period based on their need. But what we saw with COVID is that customers were a bit reluctant to sign a firm commitment for a ramp that was that fast. So we're seeing some deals come in with, say, a ramp in 6 -- was starting in 6 months. And instead of 18 months of ramp, they would ask for 36 months of ramp. Now historically, we've seen customers go faster than their minimum ramp. Because remember, when they're signing an agreement, that's a firm commitment.

And so we would see customers go faster. But obviously, we don't -- we base our bookings on signed commitments. We did see the customers win a bit faster in this first quarter, which is why we're adjusting guidance upward, that in conjunction with our bookings. But we have seen customers ask for a longer ramp period, whether they will exceed that is yet to be determined. But you asked what is typical. There's no such thing as an exact typical ramp. But historically, we've seen shorter ramp periods than what we're seeing today because of the COVID transaction.

And what's driving the COVID effect? And what's driving AFFO, frankly, it's the EBITDA increase. EBITDA went up 19%. So that's clearly what's driving the change in the AFFO as well.

Okay. And any color on AFFO for the full year?

Yes. We don't guide AFFO. And at this point, we're not intending to. So we do guide to EBITDA, and I think you can give -- get a pretty good indication based on our EBITDA guidance as to where we think the AFFO will go.

And our next question will come from Nate Crossett with Berenberg.

Question on churn first. It was good to see it come back down again in the quarter. I was wondering if you could give us an update on the space that went vacant last quarter. Has there been any update there? Has any of that been backfilled? And then how is, I guess, churn tracking so far in 2Q?

It's a great question on the churn. It is good to see it go back down. For those of you that need a refresher, we had 2 customers that reduced their service levels with us. So that they could take advantage of some cloud offerings with respect to cold storage, and that was about 2 megawatts worth of reduction in Q4. I'm pleased to report that we replaced those 2 megawatts that went to the cloud with a cloud customer that moved into our facility. So it's an ecosystem. It's kind of ironic that they left for the cloud and replace them with cloud.

But that has happened, and it happened in the exact following quarter. So pleased by that. We have no indication that there are any major churn events that are upcoming this year. Customers appear stable and growing with us. And you can see this quarter, the vast majority of our growth came from existing customers actually expanding, which is great to see, and we don't expect any to be exiting that we know of for the course of this year.

Okay. That's helpful. I guess my question and follow-up on that would be, what's the delta between the released rate versus the churn rate, if you're willing to give that?

Well, yes. I won't speak to the specific rates on specific space, but you can see in our disclosures that our average revenue per cabinet is now over $2,500. So it continues to increase. So we're seeing additional yield from the cabinets that we're selling. So we're not seeing significant rate reduction, and we're seeing customers continue to dense up their cabinets, which gives us more revenue.

Okay. That's helpful. And if I could just ask one about kind of the funding needs for the business longer term for you guys. Obviously, leverage is going to go up with the Foundry deal. I guess how should we think about the puts and takes longer term going forward? What's the company's tolerance for leverage? Historically, you've been at the low end of the space. I'm just -- how should we be kind of thinking about that?

Sure. Nate, historically, we have been at the low end. We've been in the 3x range for years because of the way we deploy capital. It's very modular. So we'll deploy just enough power and cooling as we open up a sector for those customers that have moved in. And as more move in, we deploy more power and more cooling, and that EBITDA follows. So it keeps our leverage profile low. But we've also said we're -- we've never needed to be levered, but we're certainly not afraid of it, particularly when you look at 97% of our revenue is recurring and is very, very stable.

And so this was an opportunity to use our balance sheet. We often get asked, why aren't we using our balance sheet more strategically to grow where we see an opportunity. Well, we found one. Historically, we haven't been able to find an opportunity to fit within our profile in terms of the quality of assets, the location of the facilities. And this really fit the bill. It's good quality assets. It's located in Texas, where we really wanted to be given our strategic relationships that we have with many of the customers that are particularly moving into Austin, which is a fantastic market, we think, in the future, so this fit the bill.

And at closing, we expect leverage to be about 4.7x EBITDA, and we'll continue to fund our CapEx growth as well. And we don't see the need to lever beyond a position where we're comfortable. We're certainly comfortable being in the 4s, in the upper 4s, and even going above 5 temporarily should we need to. But we're comfortable with our leverage profile. We're still going to be at the low end of the peer group even after this acquisition.

Our next question will come from Eric Luebchow with Wells Fargo.

I was wondering if on your development pipeline, do you have the ability? Or have you historically seen much activity in terms of preselling or almost like pre-leasing like some of your peers do. Just wondering if there's any opportunity, for instance, at like LAS VEGAS 15 to sell some of that capacity in advance of opening the facility or typically, would you -- you don't see as much of that given you have more of an enterprise like customer base.

And then the other question I have was on the MRC per cabinet equivalent increase to over $2,500 maybe if you could talk a little bit just how that's trended over time? And is that -- the growth in that metric primarily come from adding more power density? Or have you also seen growth in whether it's contract ramps or connectivity services like cross-connects?

All right. Thank you, Eric. And so a couple of things. One is usually, it's the wholesaler model where you see people doing pre-leasing or preselling. And in our portion of the market, we don't see as much of that just historically. We have people taking larger deals, but we generally, people like to tour the facility, see the facilities and know that the facilities are coming online so that they can plan their infrastructure that's going to -- go into that facility.

So that's the quick answer to that one. On MRC per cabinet, I'll let Gabe talk to that. But quick notes, we do have increases in MRC per cabinet that come from density. We've seen a pervasive increase in the amount of density that customers are using. And then we also see an increase in connectivity, which includes an increase in cross-connects. Cross-connects continue to increases a portion of the connectivity revenue and a portion of our revenue overall. And of course, with Data Foundry, we have picked up another 2,000 plus cross-connects as part of that acquisition. Gabe?

Yes. And as far as the trend on revenue per cabinet, over the last several years, we've gone from about $2,300 per cabinet to $2,400 and now just over $2,500. So the trend is increasing. Despite what you see in the industry with repricing risk, we're seeing additional yield from our cabinets. And it really does come primarily from density increases and additional utilization. But the telecom network is also expanding and that for certain customers, it expands more rapidly than others, but we see increases from both density and connectivity. But density is driving the majority of it.

Our next question will come from Tim Long with Barclays.

Two questions, if I could. One, could you talk a little bit about kind of developments with Switch EDGE and maybe give us an update on the FedEx deal and how that's going? And then second, can you talk a little bit about -- I think you mentioned some interest from customers for the Texas for the new PRIME. Could you just talk a little bit more broadly about how you're seeing kind of cross-selling lately across the other 4 PRIMEs as well?

Perfect. So as to EDGE, we continue to work towards the development of EDGE facilities. Remember, EDGE is something that is strategic for us but not a huge contributor to revenue. We don't have any revenue in the 2021 forecast for EDGE. So it's not going to be a financial contributor in this fiscal year.

As to Texas, yes, we have seen customers express interest in the megawatts that are available in Texas, and we continue to see people wanting to cross-sell. And the fact that it is existing customers of Switch proper that are interested in Texas shows us the power of that cross-selling.

Remember that in Atlanta, about 50% of the customers in Atlanta are new logos and the other 50% are existing customers that have expanded into that location. So cross-selling is something that we see as a very valuable attribute of the PRIME ecosystem. Gabe?

Yes. And additionally, in this last quarter, some of the contracts that we signed were actually renewals with incremental increases that involve multi campus expansions, moving from Las Vegas to Las Vegas and Reno, for example. So we continue to see the benefit of having multiple locations and are very happy with the way the customer base is expanding.

Our next question will come from Brett Feldman with Goldman Sachs.

And maybe just bringing it back to sort of the M&A strategy. In the past, I think a lot of investors would not have anticipated that you would be pursuing deals mainly because you have it in the past. But also the standard of the facility that you build a proprietary facility is really unlike anything else there in the market. And so obviously just can't buy assets that look like the assets you currently own.

So I guess the question is, what do you have to get comfortable with when you're looking at someone else's portfolio to be comfortable that you can fold it into your portfolio and deliver an experience to the customers that aligns with the brand identity you've built? Because you obviously have some incredibly unique elements to your brand identity such as having had 100% uptime. So that's the first question. Just because it would seem like you might have a much more expansive opportunity over time in the M&A field if we can get a better handle for what you're willing to do and interested in doing it.

And then the second one is a different topic. Just from a supply chain standpoint, a lot of angst around stress on global supply chains. Has that showed up in your business at all in terms of any of the infrastructure assets you need to acquire to build out your campuses? Or have you heard anything from your customers that maybe need some flexibility to deal with their own supply chain issues?

All right. So a lot in there, Brett, thank you. First of all, in M&A, you're absolutely right. We have until now not found an asset that we were interested in acquiring. The Data Foundry asset was the best in the region, and it's a good asset. It's a very -- it's not a Switch level asset, but it's a very solid, well-run data center. And we were also very pleasantly pleased when we had the ice storms in Texas and Data Foundry managed to stay up, uninterrupted in both Austin and in Houston. So that was a wonderful unplanned test that they passed with flying colors.

They also have a stellar reputation for service. But as large as the Data Foundry transaction was at $420 million, it is the catalyst for our fifth PRIME. Our PRIMEs tend to be at least 1.5 million square feet. The idea is to use the Data Foundry asset as a way to catalyst for the growth and rapid expansion of that PRIME. As good as the facilities are, what we were also as equally or more interested in, in that transaction was the contracts, they have a very robust amount of very high-quality customers and also their team.

Their team has been with them for as long as 27 years. It is a long enduring, stable team that has been very successful in that region. So if the team, the asset and the contracts together with the location, made a very compelling model for us to consider. And so we took that on. And that was the first time that we've done one. There's a possibility of us doing more in the future, but this was a unique asset that ticked a lot of different boxes for us.

With respect to your question regarding supply chains, we have not seen any slowdown from our customers in terms of requesting any sort of deferrals on their deployment as a result of supply chain management. All of our customers have lived up to their take-or-pay expectations. Nobody has requested any reprieve from that. As to our own construction, we have very strong relationships with our suppliers. And steel is little difficult to obtain, but we have been able to get ourselves to the front of the line and obtain this deal and concrete that we need to continue our construction unabated.

And our next question will come from Frank Louthan with Raymond James.

This is Rob on for Frank. So you mentioned during your prepared remarks that the G&A cost savings are going to comp less favorably for the remainder of the year, which makes sense. Can you give us a ballpark percentage of some of the COVID-related cost savings that you estimate are here to stay more permanently? And then secondly, are the COVID-related permitting delays you've referred to in your conversations with us in the last couple of calls. Are they now mostly behind you guys? Or should we expect that those could still be a factor going forward?

Yes, I'll jump in on the savings. First, as we talked about last quarter, our margins in 2020 increased from about just under 50% in 2019 to 52.5% in 2020. And if you look at our guidance, we're guiding to 52% EBITDA margins for 2021. So that means that the 250 basis point improvement, we expect to make about 200 of those permanent. We do expect that there will be some additional SG&A cost in the back half of the year.

And of course, in Q1 of 2020, we were still operating pre-COVID. So the SG&A comparison is quite favorable in Q1. So the comparisons will get a bit less favorable as we move throughout the year because we shut down in Q2 of 2020. And we're not traveling. We weren't doing charitable contributions because those events shut down. So there were a number of things that just stopped.

And we'll see those comparisons play out throughout the year. But we do expect to make the majority of the savings that we experienced last year, permanent, and that's why we set the guidance that we did making about 200 of the 250 basis point pickup permanent.

Great. And then on the COVID-related permitting delays?

So we have -- we've gone through the COVID-relating permitting delays. We have been issued all the permits that we need to continue construction. And again, that's part of the reason that we put up the pictures of the current development that's underway in 3 out of 4 of the PRIMES, so that the investors can see what we've been building and track that progress from a visual point of view.

And our next question is a from Colby Synesael with Cowen.

Great. You mentioned selling SUPERNAP Italia. I think you said you recognized a $5 million gain. I'm curious how much you actually ended up selling it for. I assume that's different than the $5 million gain. And then is it correct to understand that, I guess, you took that capital and rolled that into boosting your stake in Thailand. And then I guess as it relates to Thailand, how are you going to be recognizing that, is that going to end up getting 30% of total revenue? Is it going to show up in your revenue, 30% of EBITDA in your EBITDA? Are you going to actually include that in your AFFO? Just trying to get a sense what's actually happening there.

Colby, we did with Italy, we had a stake in SUPERNAP International, which was our international joint venture. And SUPERNAP International held a stake in SUPERNAP Italia. So as SUPERNAP Italia got sold, there was a gain that basically flowed upstream to SUPERNAP International. And then SUPERNAP International also held an interest in Thailand. So we essentially exchanged some of that interest in SUPERNAP Italia for additional interest in SUPERNAP Thailand and recognized a gain. But we don't consolidate any of their financials on our books. We use the equity method accounting.

We'll continue to use equity method accounting for SUPERNAP Thailand as it exists. But one of the other important aspects to that transaction was that our -- when we first set up SUPERNAP International, this was years ago before I joined the company -- right before I joined the company, so it must have been 6, 7 years ago, we gave an exclusive international license to our partner overseas. As part of the sale, we took back our international license for all of our IP, all of our technology, and that's what allowed us to then license that IP in Saudi Arabia.

And recognize that $2.8 million license fee. So we have a bit more flexibility as to what we're going to do internationally, now that we have our license back, and we will have an increased stake in Thailand, which will continue to use the equity method accounting for that stake.

And this will conclude our question-and-answer session. Also concluding today's call. We'd like to thank you for attending today's presentation. And at this time, you may disconnect your lines, and have a great day.

Thank you all.

Thank you.