MERLIN Properties SOCIMI, S.A. (BME:MRL)
Spain flag Spain · Delayed Price · Currency is EUR
11.52
-0.18 (-1.54%)
Nov 16, 2026, 5:35 PM CET
← View all transcripts

Earnings Call: Q3 2026

Nov 13, 2026

Summary

Nine-month FFO rose 11.9%, supporting higher 2026 guidance, while traditional assets sustained growth and 95% occupancy. Data center expansion and funding remain strong, but Spanish regulation could add substantial battery-storage costs.

Inés Arellano
Director, MERLIN Properties SOCIMI

Good afternoon, everyone. Welcome and thank you for joining MERLIN's nine months trading update. As usual during quarterly reports, Ismael Clemente, our CEO, will walk you through the main highlights of the period. This time he will thereafter address what are the most common questions that we get every day in a presentation that is already available on the website. For those who want to raise questions afterwards, please, as always, dial star five in your phone. If you haven't been able to register due to the Outlook issues, we've enabled in the link space for you to write those questions. With no further delay, I pass the floor to Ismael. Thank you very much.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Thank you, Inés. Welcome to MERLIN's third quarter 2026 results presentation. As you have probably seen in the publication yesterday late afternoon or this morning, despite the emotional rollercoaster during the quarter, the results of the company have remained steady and convincing. I think it's been a tremendous effort on all the team, part of which is gathered here today in this hall. The remarkable performance of the traditional asset classes that has been our hallmark during so many years recently has been sustained with 3.4% like-for-like growth in offices, shopping centers, and logistics, with a particularly stellar performance of shopping centers in the quarter. The very high occupancy overall of the company remains true at 95% and likely to be exceeded as of year-end, also because as of year-end, we will also report occupancy of our data center assets.

There's been a very significant income growth, which is normal. I mean, normal for those of you who have been dealing with us in past investors conference is simply that little by little, the cash flow from data centers start kicking in the P&L of the company. Of course, this is creating now a growth in the top line that hopefully in the future will result also in a significant growth at the bottom line. The FFO has increased also by 11.9%. Very importantly, that kills the dilution caused by the March capital increase within the same year, which I believe is a very interesting point of information.

If you look also at the comparison between nine months 2025 and nine months 2026, you will see that our margins are maintained overall or even slightly improved, which means basically that the company continues converting top line into intermediate and bottom lines with ease. In light of this FFO behavior, we are in a position to increase our expected full year 2026 guidance to EUR 0.58 versus EUR 0.55 before. I know many of you believe we will also exceed EUR 0.58. I don't fucking know if we will be able to exceed it or not. I really don't think it, but let's at least give you the most accurate guidance that we can give you at this point. Very importantly, the MEGA Plan continues rocking. Beyond the 160 MW IT-let, we are in advanced negotiation now for 206 MW, up 26 from last information disclosed to market.

That belongs mainly to a new client in Lisbon for a potential repowering of 20 MW, and two new clients in Madrid for a potential repowering of 6 MW that we are dealing with as we speak. Many of you will be surprised that the 180 MW big contract in Portugal is not yet signed. I have to say that it is a major contract, which total backlog will compare to total market cap of the company is going to be very significant. It has a number of sovereignty issues. It has a number of complications that we need to overcome, and particularly and more importantly, neither the client nor us are in a rush, because the first delivery of assets pertaining to that contract will only happen in December, at the end of December 2027, and the second batch in December 2028.

It will happen when it will happen, and if and when it happens, it will be the subject, given its sheer size, of a relevant fact notification to the Spanish stock market regulator. You will get notice of the signature of this contract through a stock market regulator communication. Regarding the non-core divestments, we are at 75.4 with 66.8 to be executed before year-end, of which the first 30 have already been executed. We are perfectly on track to be in the region of 140 as of year-end, which was our target for the year. We have already prepared part of the way of 2027 with approximately EUR 90 million, which are in the process of being signed during 2027. There have been no valuations in the quarter. The net tangible assets per share stands at EUR 16.13.

We will only proceed to a valuation of assets as of year-end. If you look at the different asset classes, in offices, the like-for-like has been 2.3%, with a re-leasing spread of 1%. In logistics, 1.6%, and a re-leasing spread of 4.3%. Shopping centers are 7.1%, with a re-leasing spread of 5.1%. Very importantly, as commented on some other occasions, now we are translating into customers a significant chunk of inflation on every year, and this is, of course, taking a dent, taking a toll in the reversionary potential. You cannot expect very brilliant re-lease spreads going forward because we are already making significant hikes in our rents every year through CPI compulsory indexation. Occupancy overall has gone up by 38 basis points to 95%.

Importantly, in offices, where we were a little bit shy at the beginning of the year because we knew we would lose in April Meta in Barcelona, and we didn't know whether we would be able to replace it through performance of the rest of the portfolio. We gave you an indication for the year of between 93 and 94. Mid-year, we said midpoint to high point, so 93.5 to 94. We are already at 93.7, and frankly speaking, I believe we are going to end up the year above 94.

Hopefully as close as possible to last year figure, which is, again, a remarkable effort by all the team and also a signal of the strength of the Spanish and Portuguese office market, which in cases like, for example, the Art's Building in Lisbon, we have been able to completely re-lease it after the exit of BNP Paribas, and we are also progressing in Barcelona, improving occupancy, and Madrid is now clearly outperforming Barcelona, both in terms of occupancy and a perspective of passing rents. That's basically it for, let's say, ordinary performance of the company. We have prepared a presentation that is very simple and very blunt. We have called it Addressing Market Concerns. As you know, the quarter has been absolutely varied. The month of July was easy-peasy. I mean, tailwinds everywhere and wine and roses.

All the sudden, in August, the market started fearing high interest rates, which, in my humble opinion, should be a small preoccupation in a company like this. I mean, if we were 50% LTV, I would be worried about higher interest rates. Being 25, I am not that worried about the effect of higher interest rates in the company, and we have prepared some numbers to show you in very big figures, more or less, what the picture of the company looks like as of today and what it will look like in 2030, so that at least from a rationality standpoint, we can allay your fears regarding the hike in interest rates, which, of course, will affect us not only by increasing significantly our debt service, but also by probably reducing a little bit our total value, at least particularly in traditional asset classes.

I believe we will be able to compensate that with the value creation that we are carrying out through our data center development program. So, in the presentation that you have in the web of the company, we are going to be splitting it. I will comment you on the famous or infamous royal decree to be issued by the Spanish government, and what is our contingency plan as a reaction to that legal initiative. Then my colleague Fran Rivas will comment on CoreWeave, which is another one of the topics that we have been suffering, that one particularly in the month of September. Then Javier Echave, we will take the opportunity to introduce Javier to all of you. He's just new to the company.

He has joined coming from Heathrow Airport in the U.K., and he's going to be our CFO, so he will introduce himself and talk to all of you about what can we expect from the hiking interest rates regarding, or at least in what pertains to MERLIN. On royal decree, I'm sure that you are more or less familiar with the contents of the royal decree. There was a first draft. The first draft was a disaster. Then there was a significant reaction by the sector with 663 allegations or challenges or legal challenges to the contents of the royal decree that were submitted to the government. Basically, they have paid no attention to any of them. I mean, they have passed all of them by the 'arc of the triumph'.

They have created a second draft, which is, curiously, in the friendly media, they say it is a significant softening of the first version. This is only because general citizens do not know the so-called SRAP mechanism that I can explain to you, which is a POS, and basically kills the possibility of any decent data center company accepting or adopting the measures of the second draft. In the second draft, the changes to the first one are basically they have increased the minimum size of the data centers that are affected by the royal decree from 1 MW to 10 MW. Well, what a great difference.

I mean, in reality, the only reason why they have increased from 1 MW to 10 MW is in order to safeguard from the effects of the royal decree all the mini data centers publicly promoted by the government and regional and autonomous communities. I mean, because those tend to be very small data centers between 2 MW and 4 MW, 6 MW, and they have excluded them from the application of the rules of the royal decree in order to self-protect themselves from a norm which is inherently abusive in its current drafting.

The second, in case there was any doubt, they have created a new figure called strategic national interest that is basically intended to safeguard the Spanish EU Gigafactory candidacy from the application of the royal decree because they know for sure that it cannot comply with the royal decree, so they have exempted it from complying with the royal decree. They have left the compliance with the royal decree only for the private operators of the market, which are the bad guys. The rules on what is considered additional renewable source have been significantly amended to move wind sources from 18 months precedency to 36. So, you will consider now good for royal decree purposes any wind park which was built 36 months before the day in which your data center is energized. For solar, anytime is good.

They have gone back to zero in order to protect the photovoltaic sector, which is in the brink of insolvency, and they needed to protect it, and they have done it through this mention. The four main points about the new draft is that they keep the 80% self-consumption with hourly correlation, which is absurd and very difficult to comply with, but they have created a transitory period subject to becoming part of the SRAP system without any compensation whatsoever. What is SRAP? For those of you who are not familiar with the Spanish electric system, there are two mechanisms by which the national grid authority can disconnect an intensive consumer from the grid in case there are problems. The first one is called SRAD, is active response to demand system. The SRAD is a more or less civilized system which is intended to prevent blackouts.

When the national grid authority believes that there is going to be an overall excess in consumption versus generation at a certain point of time, with certain time of antecedents, they call you, or the technical room of national grid authority, they call the intensive consumers and they inform you that you have to disconnect your facility for a certain period at a certain point in time, which is normally 15 minutes later, half an hour later. That means two things. First, that you can move safely into emergency procedure, so you can start your generators and everything without any problems of heart attack. This is very important. But the second most important thing is that the line remains energized, so you remain connected to the main grid.

That means that if you experience problems, for example, igniting one of your gensets or whatever, you can always go back to grid and keep your facility working. This is particularly important, for example, for steel mills, because there are points in the day in which if they stop operations, the melted steel will basically solidify, and it creates a big problem then to melt it again. So, for that reason, in case generators do not work or whatever, you can go back to the normal system. So far this year, they have called, or they have made use of this emergency procedure eight times. So, the Spanish grid has been at risk eight times during the year, and in principle, we should have no problem becoming part of that system because happens very now and then. It doesn't happen that often. And second, it is not sudden.

I mean, there is a pre-notice with which you can start your generators and be on your own without problems. And it is also a remunerated system. So, if you need to disconnect, you are going to be remunerated by the equivalent cost of electricity that will be applicable at that point of the day, which is normally a peak hour. So, the compensations can be significant, and when you receive those compensations, you partially offset the cost of your fuel, which is not a very efficient and is a very expensive way of generating electricity. Okay? So, this is the way the SRAD works. The SRAP is completely different. The SRAP applies to electrical nodes and is randomly generated by an algorithm and without any pre-notice.

So basically, when the algorithm detects that there are frequency variations or tension holes in a certain electrical node, basically, they disconnect everyone around that node, okay, till the, let's say, normality in the grid is recovered. So, the number of occasions in which that may happen is very significantly above the ones of the SRAD because the SRAD is the general grid, the general consumption of Spain, whereas in the case of SRAP, it is calculated by nodes, it is predicated by nodes, and is subject to an IT program. Okay? And second, and more importantly, it is not remunerated. So, the immediate consequence of accepting that measure is that you will need to include that provision in all your existing lease contracts, informing your clients that at certain times during the year, at the will of the public authorities, they are going to go on generators.

The last line of defense of any given data center in the world, which is generators, become the main energy source for a certain period, and you need to tell that to the client. It is not technically a problem, but it is something that you need to tell your clients, and of course, it will be a good reason to either exit a certain contract or renegotiate it. It is important that people knows about the consequences of joining that SRAP crap. It is very important for people to know. You need to do it both in existing contracts and in prospective contracts. For example, in Portugal, we will need to tell the government and the client that we, in case Portugal adopts the same measure, we will need to disconnect. You can imagine the mess you would introduce in a certain contract.

Thanks God, the Portuguese government has confirmed that they are not going to follow the leadership of Spain in this aspect. In exchange for joining the SRAP, they give you a certain transitory period of two years in which you need only to cover 70% of P1 hours, then two other years in which you need to cover 70% P1 and 60% P2, and then from there onwards, 80%. Our decision has been to try and comply with 80% from scratch because the contrary, if we don't do that, guaranteeing to a client 99.9999% availability becomes complicated on a statistical basis. I mean, if you are running at times your data centers just on generators, guaranteeing that this is going to kill 99.9999 probability, it's very, very complicated.

Many people will say, "Why don't you use batteries?" Because batteries, the battery flow comes through the same line. If the node cuts the line, even if you have batteries in a different place and you are trying to suck from your batteries, unless you have batteries on site, but if you have batteries on site, you have many other problems regarding particularly fire. So, it's impossible to comply with the royal decree. On page five of our presentation, you will see that the royal decree has now gone through the non-binding opinions of both the antitrust authority and the Council of State. Both of them have raised very significant objections to the contents of the royal decree.

In the case of the Council of State, there was a significant it's been a soap opera because the first working draft came with 14 illegality flaws and the recommendation to withdraw the approval of the royal decree. But at the end, through political pressure, I mean, the chairmanship of the Council of State is basically stooge, and they have managed to reduce the number of essential observations, but still, there's very significant essential observations. One of them is that the royal decree cannot penalize anyone with 500 x the grid tolls or 100% or whatever. They cannot impose any penalties with an administrative measure, with an administrative order, like a royal decree. If you want to impose penalties in a state where there is rule of law, you need to resort to a proper law. You cannot do it through an administrative measure.

Anyway, I believe, as you can see on page six, that in all likelihood, the royal decree is going to be approved, because this has become a little bit of a hold my beer kind of thing for the government. It has been included in the electoral campaign. So, with a significant degree of probability, we believe that the royal decree is going to be approved on the Councils of Ministers of the November 17, November 20, or November 24 , or eventually approved between 29 November and December 23 in the transitory period before a new government is formed, whatever its color might be. Okay? In this case, you need the opinion of the State Attorney's Office, but I believe they will get it because the State Attorney's Office is not independent in Spain.

There is a remote case in which the royal decree maybe is not approved because there is certain pressure from the European Union on grounds that the TRIS mechanism has not been complied, and on grounds of paying attention to the illegality flaws raised particularly by the Council of State and also by the National Markets and Competition Commission. But we prefer to prepare for the worst. If that approval, as depicted on page seven, if that approval happens on the November 13, it did not happen. So, on the November 17, November 20, or November 24, the royal decree will be valid from approximately mid-December onwards. Okay? And we need to prepare for that scenario. There will be a new government. The Spanish Parliament will be constituted on December 23 2026, following the elections, and the new government could potentially be taking possession in early January 2027.

In which case, particularly if there is a change of wind in the politics, the decree will probably be repealed. Many of you will wonder why are we doing all that noise if the royal decree is going to be very, very short-lived. I believe this is simply the main motivation is to create legal confusion and dissuade data center investors from investing in the country on the grounds that even if the royal decree is repealed, they will be signaling kind of their legal stance for future governments that might be formed in the future. The damage, in a way, is already done in that respect. The center-right parties and some left-wing parties have signaled that they will repeal the royal decree in case they form a government. As commented, it should be relatively short-lived.

How do we, let's say, plan for the future in light of what we just commented? What we have done is we have created a response plan in order to tackle market uncertainty. I mean, at the end, our worst enemy is uncertainty. So, I prefer to cut my hand and give you a worst-case figure that you can input in your models rather than let you dream and be superstitious about what are the magical effect of this royal decree. What we have done is we have assumed that the royal decree is in place and remains in place for the foreseeable future, and we have quantified what is the CapEx that we need to add to our existing CapEx in order to comply with the royal decree. That means basically that we can only improve from there. Okay?

In case we can defeat the royal decree at the Supreme Court, in case we can have the European Union intervening and stopping it on the grounds of loss of market equivalence in the interior market of the European Union by these different criteria and different countries regarding the development of data centers, particularly now that the European Union is putting together a new directive called CADA, Cloud and Artificial Intelligence Development Act, that is basically intended to fast-track the development of data centers across the European Union, copying the norms of an autonomous community in Spain called Aragón, where there is a regulation called PIGA that fast-tracks the development of data centers from rural land to construction permit.

They are kind of converting that into a European regulation, and they are trying to promote the development of data centers within, if possible, 12 months to obtain the permits and then civil construction and equipment on the private developer side. What we have done is, first, we have tweaked a little bit our MEGA Plan footprint. Why is that? Because we were having issues with the delivery of electricity in Arasur 0 4 and 05 in Bilbao, because the generation partner was in the middle of a legal rift with the Basque government, and that has resulted in losing their electricity licenses. We are working with them in finding an alternative location, which, by the way, will be much better in terms of self-consumption potential, and that would be equivalent in size.

The second possibility is to receive electricity from the general grid, and the Basque government will probably help us in doing so. That electricity will be subject, because it's future electricity, will be subject to the royal decree. In 150 MW, it will be very difficult to hedge it with PPAs in a, let's say, economically meaningful way. We have decided basically to move those two buildings to the freezer for the moment and have replaced part of that capacity with a repowering in the Portugal data campus, with a repowering in Madrid, and with the addition of a second development project we were working on in Aragon, nearby the first one, nearby Zaragoza-WIND, with a high self-consumption capacity for 50% of the power. The other 50% needs to be covered with normal PPA resources.

With all that, the total MEGA Plan moves from 724 MW to 704 MW. Basically, we make good much more than 90%, 95% of the MEGA Plan. We get out of this uncertainty surrounding the company because of the effects of the royal decree relatively unscathed, and now we will move into quantification of the additional CapEx we will need to incur in order to comply with the royal decree, which we have compared to the total program. Of course, if you consider project by project, it's a ramping. Because we are considering it against the whole program, it's more or less affordable, and we are a company. We are not a fund. We cannot simply walk away from an existing commitment that we have made in front of municipalities, regional authorities, et cetera.

We need to continue operating in this country in the foreseeable future, if at all possible. Of course, we need to comply with our word. This is why we have decided to do it that way. Moving to page 11, you will see that another collateral effect of tweaking a little bit phase III is that we advance the stabilization year from year 2032 to year 2031, and we keep reasonably similar the gross yield on cost. We have updated the underwriting rents by EUR 2 in phase II and by EUR 2 in phase III only. I mean, although we are clearly signing well above those levels. Okay?

With this, we get to a situation that you will see on page 12 in which we need to complement, particularly for hourly correlation, because in the case of Zaragoza-WIND and Zaragoza-MAGA , those two projects, if the correlation was yearly, we would exceed the required 80% by far. But if you need to do it hourly, you need to significantly invest, particularly invest in battery storage in order to comply with the royal decree. When we do the math with all the projects, we come to a total BESS CapEx of EUR 863 million, which represent 11.3% of our total investment. Okay? Importantly, we believe, and very prudently believe, that we are going to be able to recover approximately 1/3, 35% of that investment through BESS arbitrage and/or energy sales.

Since we believe we will recover that investment, the sunken cost, let's say the non-recoverable expenditure, is EUR 553 million, which is 7% of our total MEGA Plan. Again, measuring against the total MEGA Plan. If you go project by project, in some cases it's 20%, in some cases it's 30% cost increase. Okay? The total BESS investment would imply an additional cost of approximately EUR 1.2 million per MW, of which EUR 0.75, more or less, EUR 0.78 will be non-recoverable, and the rest will be recoverable through BESS arbitrage and/or energy sales. You can basically, if you want to run, I mean, we will convert these numbers in definitive if we need to continue working forever with this new royal decree. If the royal decree is repealed, you can forget all these calculations that we have made, okay?

This is simply a worst-case scenario so that we create a true order of magnitude so that the market stops guessing how much it's going to be, how much it's going. At the end, it is affecting us much more in the share price than the real effect it will have in our numbers. Okay? I leave you with Fran, who will comment on the CoreWeave issue that has also been killing us for a month now, about the financial health of CoreWeave.

Fran Rivas
Director, MERLIN Properties SOCIMI

Many thanks, Ismael. Another market concern, which has, as Ismael said, recently gained increased attention due to the growing weight of the data center business in our portfolio, is client concentration, and more specifically, the situation of our largest client so far, which is CoreWeave. Let me start by reminding everyone that this is neither the first nor the second time we have faced this challenge. at the time of our IPO in 2014, BBVA, through our bank branch portfolio, represented 78% of our total gross rental income. At that time, BBVA had a market capitalization of approximately EUR 48 billion, with EUR 21 billion of revenues, which compares to CoreWeave today at the recent lows of $32 billion market cap today, and with expected $13 billion of revenues for 2026.

Concerned about this level of concentration, the company managed over the years to reduce BBVA weight to final 15.9% by the time of the portfolio sale, following, of course, the acquisitions of several assets and portfolios, including Testa, Metrovacesa, and Saba Logísticos. We experienced exactly the same situation at the asset class level with logistics. We started by acquiring existing assets and, more importantly, developing new ones with a significant presence of, for example, a leading sector player such as XPO, which represented 46% of our total income in logistics in 2014. Today, this client represents only less than 9% of our total portfolio. Now, the name of the game is CoreWeave, which, as at the first half of 2026, represented 80.6% of our total gross rental income, although this share is expected to increase over the coming year.

Based on our past experience, there are two main ways to mitigate client concentration risk. First, by having a well-diversified group, and MERLIN has exactly that, combining offices, logistics, shopping centers, and data centers. Second, by managing concentration at platform and asset class level, limiting the exposure to individual clients within each portfolio. Of course, this is easier said than done, because diversification is a process that takes time. We will only achieve our target portfolio mix once the remaining capacity has been fully built, equipped, and leased. Until then, we need to remain disciplined and stick to our strategy rather than react to short-term market trends. Over the years, you may remember that we have received numerous suggestions to sell one business division and invest into another, and these recommendations tend to change with the market cycle.

Just one year ago, some were suggesting that we should sell all of our traditional assets and reinvest the proceeds in data centers. On the contrary, we took the decision to reduce our loan-to-value and not mix operational risks, which in case of data centers implies significant amounts, with financial risk. Today, when client concentration becomes a concern, we can clearly see the value of having a diversified portfolio. It provides resilience and gives us the flexibility to navigate market changes and lease our available capacity quickly. Moving to page 15, lease and credit protection, which is another topic frequently raised by the investment community, and this relates to the lease structures and credit protection, particularly in specific cases. As you can see on that page, we have summarized five key aspects that MERLIN takes into consideration when signing contracts with any data center client.

The first and most important one is our legal protection under the contracts. All of our contracts are governed by Spanish and Portuguese law, and in the event of insolvency, and as with our traditional asset classes, rental payments accruing after the commencement of such insolvency proceeding qualify as claims against the state under the applicable legal requirements. In addition, as with our other asset classes, all of our contracts include different types of guarantees, statutory rent deposits, parent companies or third-party guarantees, and additional cash deposits. These guarantees provide a company with a level of financial protection designed to cover the cost of recovering possession of the asset and securing a replacement tenant, helping to minimize the potential economic impact. The second point I would like to address is how easily MERLIN could find another client in data centers in the event of a vacancy.

Here, it's important to highlight that we are not building tailor-made assets for individual clients. In other words, our assets are not designed and constructed around the specific requirements of a single customer. We follow exactly the same philosophy as we have in offices, shopping centers, and logistics. Across all our four asset classes, our objective is to bring to the market top-quality buildings that are sufficiently modular and flexible to accommodate any major player in the sector. Our ambition is simple: any major client anywhere in the world should be able to lease capacity from us because we can provide the right infrastructure from a technical, sustainable, and service perspective. This explains why we accommodate different client profiles, namely cloud providers, telecom operators, or AI companies within the same building.

It also explains why, in projects such as Arasur Data Center, clients may initially consume all the contractual power in only two-thirds of our available rooms because of their typical higher densities. But we need to preserve flexibility because we cannot know whether future services will require lower power densities. There is another trend that has become quietly fashionable lately, driven by the immediate demand for computing capacity. Some clients are asking us to reduce the resilience levels of the facilities by removing redundant equipment and even backup power generation in order to bring forward the ready-for-service dates. At MERLIN, we have deliberately chosen not to play that game, because what may look like an opportunity to save time today could become a mere obstacle tomorrow when it comes the time to relet the asset to another client. Another recurring question concerns the client-specific fit-outs of a data center.

To answer this question, we must remember that MERLIN's real estate responsibility ends at the individual data hall level. We provide the building and its electrical and mechanical infrastructure, ensuring the required levels of security, temperature, humidity, power availability, and including the corresponding redundancies. This is no different in principle from the obligations we assume as landlords in a shopping center before our tenants, where we take care of the common areas and we provide the same required levels of security and temperature, et cetera. Inside the data hall, the space belongs to the client, and so does the fit-out. We may sometimes install certain elements ourselves to facilitate an early client entry, but it doesn't change the underlying ownership principles. Therefore, specific busbars, racks, and other equipment used by a client, such as CoreWeave, form part of the client infrastructure, not to the landlords.

Taking all of this into consideration, we are confident that our data centers are and will continue to be suitable for a wide range of clients in case that capacity will be available for resale. Finally, let me mention one additional point, and this is relatively. Data centers are really a new asset class for MERLIN, so we have not yet experienced a full reletting cycle or being able to report a meaningful re-leasing spread. However, if you look to our reported figures, you can already see that rental levels signed in phase I are being exceeded by those achieved in phase II and phase III. This is not explained only by the scarcity of available infrastructure, but also by the higher prices that GPU operators are charging their own customers. A third topic concerns how our clients are operating and behaving within our data centers.

Here, we have observed some very interesting trends. First, visually, all of our capacity delivered is being already utilized. Clients are entering their facilities very close to our ready-for-service dates, and followed by a ramping up at a speed we have not previously seen in the industry, have all the capacity at work. Why is this happening? Because computing capacity has become one of the most valuable assets in the AI economy, and as a result, a significant proportion of this capacity is already committed to end customers well before the infrastructure is being delivered. Second, we have not seen any material amount of contracted capacity remaining unused. In fact, the percentage of usage is pretty high, with only a minor exception that we have in a network client.

In fact, all of our clients have already increased their capacity initially contracted with us, and in some of the cases, they have even committed to additional capacity that we will only be delivered over the coming years. We also cover in this page 15 pre-lets and debtors' best interest. In the specific case of CoreWeave, the company has developed a very smart financing structure that we believe comprises all these elements we are showing here. Essentially, it brings together with a ring-fenced financing vehicle, which we graphically represented as a box, the three elements that any financier wants to see and pledge: the lease of the underlying real estate asset, the ownership of the computing equipment and GPUs, and the contract with the ultimate end customer. By ring-fencing these three elements, the revenue generated by the computing capacity and received from CoreWeave's customers flow first into this structure.

It is then allocated to cover power, rent, operating expenses, interest payments, and debt repayments. Only after these obligations have been met, thus the remaining cash flow returns to the parent company. If we put some numbers around this structure, CoreWeave reports the following approximately breakdown on a global basis. 16% corresponds to power and rent, and given the fact that power used to represent up to 60% of the rental cost, we can break this down in approximately 6% for power and 10% for rent. 9% refers to operating expenses, 8% to interest payments, 41% for debt repayment as an asset level, 10% for capital funded by the parent company, and finally, 16% as a residual cash flow retained by the parent company. There are several important considerations to be in mind here.

First, after five years, once all this debt is being repaid, the only remaining operating cost associated with the revenue that CoreWeave receives will be the power, the rent, and the operating expenses. The rest will effectively become additional cash flows available to the business. Second, let us compare this to the shopping center business. We currently report an historically low OCR of 10.7% for our shopping center tenants, and this represents, as you know, total occupancy cost as a percentage of tenant sales. However, remember that our shopping center tenants still have to cover all their operating expenses, and their occupancy cost is not the total cost base, but certainly not the profit margin, which is not the case in case of CoreWeave. Third, and this is particularly relevant, CoreWeave has reported that short-term contracts can be priced at twice the level of four to five-year contracts.

If we apply this pricing differential to our current condition, to the current economics that I mentioned before, the occupancy cost in our data centers could potentially decrease from the current 16% to approximately 7%, and being the rent only 4.3%. This is our real OCR in data centers, and this is easy to understand why interrupting rental payments is not a good business for them. Finally, on page 16, we show the evolution of CoreWeave's capacity with our portfolio. Once the full 704 MW committed by Ismael has been deployed, CoreWeave will represent approximately 19% of our total data center capacity. As I mentioned at the beginning of my presentation, this is precisely why we believe in diversification, both at the group level and within each individual asset class.

We have faced client concentration challenge before, we have successfully managed them, and we believe our current strategy puts us in a strong position to manage this one as well. Then I pass the word to Javier, that will cover the interest rates.

Javier Echave
CFO, MERLIN Properties SOCIMI

Thank you, Fran, and thank you, Ismael, for the very kind introduction. I will take the opportunity also to thank everyone around this table for their extremely warm welcome. It is only nine days, but I already feel part of the team, and I am super excited to be joining this team. I actually should say I am mega excited to be joining this team. I look forward to meeting investors and banking partners over the coming weeks and months, where hopefully we will be able to see my real face rather than the one that you can see in AI on this slide. You will be disappointed that actually I have put a couple of kilos from that picture.

Fran Rivas
Director, MERLIN Properties SOCIMI

You look more like Dario Amodei now. You look like the head of Anthropic.

Javier Echave
CFO, MERLIN Properties SOCIMI

Yeah, nine days and I'm already mutating. Ismael and Fran have walked you through the quality of our high-performing assets and shown how we can convert a growth pipeline into value for shareholders. Today, I want to add one more competitive advantage to that list, which is our balance sheet, which is now ready for growth. When investors look at a growth plan like ours, two questions usually come up on funding. The first question comes from our REIT structure. As you know, because we distribute most of our earnings, we cannot rely on cash flow of our existing assets. We need to pre-fund our growth. The second question is interest cover, whether our cash flow can keep pace with our high interest burden while new projects are still being delivered. Slides 18 and 19 hopefully address these two questions directly.

Let me start on slide 18 with where we stand. As of September 30, our balance sheet is very strong, evidenced by three metrics. First of all, cash and cash equivalents stand at EUR 1.4 billion. Second, leverage, loan-to-value is 24.9%. This is well below the industry average and 4 percentage points lower than at last year end. Third, interest cover, which is at 3.6 x against a covenant level of 2.5 x. All of this supports a strong investment-grade rating with a stable outlook. That was the picture as of September 30, and since then, we, this is the real we, actually, the team, have successfully refinanced our debt portfolio. On October 30, we announced a new successful sustainable financing platform that fully removes our refinancing risk for the next three years and gives us the financial muscle to deliver our pipeline growth.

The new platform has two components. First, two term loans totaling EUR 2.25 billion, both maturing on April 2032. The first tranche of EUR 725 million is already fully drawn, and the second tranche of EUR 1.525 billion will be drawn no later than the end of April 2027. Secondly, an expanded revolving credit facility, which has doubled its size to EUR 1.5 billion.

This new platform supports a healthy sources and uses plan that you can see on top of slide 18. On the sources side, we have the new EUR 2.25 billion of term loans that I have just referred to. Over the last months, we have also completed additional EUR 1.042 billion of issuances. You might remember the EUR 550 million bond issued August 2025 that will be due in 2033, and EUR 492 million of unsecured loans. That provides a total of EUR 3.3 billion of funding firepower.

On the user side, the first EUR 2.425 billion goes to meet debt maturities. We have already repaid the EUR 800 million bond due on November 2nd and the EUR 725 million term loan due on 2028, well ahead of maturity. Also, we are earmarking funds to repay the EUR 500 million bond due on July 2027 and the EUR 400 million for the bond due on September 2029. As you know well, we do debt portfolio funding, and therefore, we will be allocating funds as they come. That doesn't mean that we are specifically refinancing monies for specific maturities. Obviously, if you put that together, that leaves EUR 867 million available to pre-fund CapEx, which is enough to cover 47% of the debt component attributable to phase II.

On top of that, the expanded EUR 1.5 billion RCF, which is fully undrawn, would be enough to address the rest of phase II and around 25% of the debt associated to phase III, in the worst case of markets being unavailable to us. Our intention, as mentioned several times, is not to draw the facility without need. To complete the picture, it's worth noting that the remainder available cash and cash flows from traditional assets will fund our general corporate needs as well as ordinary dividend distributions. Obviously, liquidity is extremely important and in good shape, but the refinancing has also improved the quality of our debt portfolio. Average maturity has been extended to 5.7 years. Our interest rate swaps have been novated into the new structure, so close to 100% of our debt remains fixed. That was a long story that can be put simply.

When combining debt and equity already raised, cash flow retention, and expected disposals, we forecast to cover all CapEx needs of traditional asset classes and more than 1/3 of phase III investment with phase II, and 1/3 o f phase III investments without the need to access debt nor equity markets until 2028. That hopefully answers the first question. The bulk of our growth is already pre-funded, and financing has become a strong competitive advantage for MERLIN. We move to slide 19 to address the second question, which is interest cover. Obviously, it's natural to ask the question these days about how to prevent a potential liquidity trap caused by fast growth, or what happens if, at the same time, rates go up. On this slide, we have run an exercise to show that MERLIN is future-proofed against that.

It compares our rents and our interest burden today with what we would expect by 2030. To run this purely illustrative scenario, we have been deliberately conservative. As our current hedges and debt are replaced over time, we have assumed that our average cost of debt rises by 110 basis points to 3.8%. Combined with the debt that we take on the fund growth, that would result in a total gross debt of around EUR 8.4 billion.

Debt service would increase from EUR 134 million in 2025 to EUR 1.576 billion by 2030. Sorry, debt service would increase from EUR 134 million in 2025 to a potential EUR 317 million in 2030. I got overexcited about that. That's around 2.4 x today's level. Over the same period, our rents are expected to grow much faster, from EUR 541 million to EUR 1.6 billion by 2030, or around 3x . The picture is quite simple.

While debt service jogs, FFO runs. Despite this position of strength, none of this changes our discipline. We remain fully committed to a strong credit rating and a prudent approach protection of our balance sheet. That means keeping loan-to-value below 35% and maintaining a strong buffer over ICR covenant, which is, as you have just seen, expected to strengthen rather than weaken. In terms of future funding plans, we are fully funded for the foreseeable future, so we do not need to rush to the debt nor equity markets to fund our growth. That said, we remain fully open and proactive to opportunities when market conditions are attractive. I hope this gives you comfort that our MEGA Plan is an attractive strategic opportunity, not a financing challenge. We have done our homework.

Numbers show that our balance sheet is ready for growth and can deliver it while keeping our commitments to both creditors and shareholders. With that, I'll hand back to Ismael to wrap up and Q&A. Thank you.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Thank you, Javi. It's a pleasure to have you here, and many successes in the future, and I hope you can start meeting our investors around the world because I need to take a rest a little bit and work. Well, as commented at the beginning, quite strange quarter. While we see that the company is performing like a Swiss clock, we need to face every day new challenges by the market, new doubts, and new superstitions. But we will try and keep our heads. This quarter, I have remembered on many occasions Rudyard Kipling when he said, "If you can keep your head when all around are losing it and blaming it on you." It's a very, very inspiring poem. We will fight the royal decree. We have no chance to do another or to follow another option.

Particularly what they have done with the grandfathering in plain words is criminal, because if you look at our Arasur 01 example, for example, in the Basque Country, by not putting a grandfathering protection to projects like this. We are talking about a project with 48 MW IT, which is fed from a line which is 70 electric, in which we have maximized buildability. So now, we don't have a space whatsoever, buildability whatsoever, to put up a BESS in the premises. And the line is not capable of feeding up batteries in case we find a place to put them, because if the client is working with a load factor close to 100%, there is no freaking way we can charge the batteries.

So, by applying a new legislation to a project that was started one year and a half ago, and which is now with civil construction finished, and we are already receiving all the equipment, part of it is on premises, part of it is on sheds, but we are already receiving all the equipment. Basically, they are screwing up the profitability of an otherwise super profitable thing. Of course, we will claim that loss of profitability from the government in due time, that in the best interest of all of you, while we do that, we will also react by protecting your interest through the worst-case scenario that we just commented, in case we cannot obtain precautionary measures from the Supreme Court, and there is no change of wind in the next elections.

That theorem has also another corollary, which is that if the situation doesn't change, that means we will stop new developments in Spain and move production to Portugal. I mean, it's not a caprice. It's simply we are obliged to do so because there is a delta of profitability which is very significant in favor of Portugal, where the government keeps a certain common sense in the way they approach legislation. With all this information, without further delay, let's move into Q&A. I'm sure there will be hundreds of questions, and we will do our best to reply all of them in an orderly basis.

Inés Arellano
Director, MERLIN Properties SOCIMI

Thank you, Ismael . The first question comes from the line of Mariano Miguel Hidalgo from Alantra. Mariano, the line is yours.

Mariano Miguel Hidalgo
Analyst, Alantra

Thank you, Inés, and thanks for the color on all those aspects. Of course, thanks for taking my questions. I've got a few. A couple of one on Portugal. First, I'm sorry for asking about regulation, but there has been some news on potential data center regulation, so simply wanted to confirm with you that we shouldn't expect any impact on your Lisbon campus or any other following developments. Then, also on Portugal, we were wondering if you have any update on the power auction you participated a few months back.

REN has said this morning that it expects it for Q1 2027, but I don't know you have any additional color on it. Then a final one on numbers, in order to welcome Javier, a brief follow-up on the cost of debt exercise. Could you give us some color on the expected evolution of cost of debt going into 2027 and 2028 in order to make the calculation of the FFO a bit more precise? That's all on my side. Thank you very much.

Fran Rivas
Director, MERLIN Properties SOCIMI

Thank you. We'll take the ones related to data centers. Regarding Portuguese, there has been, as you perfectly said, on November 3, some speech from the Ministry of Energy declaring that they're working on a regulation on data centers. We have been meeting yesterday the Secretary of State of Energy to address, I would say, 100% of the questions you have raised now. That legislation, as you said, they will do in two ways. The first one will be a public consultation before year-end, most probably. Once they receive the feedback for the different sectors, actors, they will take measures and approve it in the parliament, and it will be effective. It will be a little bit different than here, where we wake up a little bit every day with new regulation, and then you need to make allegations on it.

It will be a little bit different. We'll be more in consultation with the different actors in the sector. Second point that I will highlight is that we ask whether the measures will be linked to what we are seeing here. Of course, they cannot disclose information yet, but things like hourly correlation and this type of SRAP that Ismael described before, of this type of actions are not in the table right now on the regulation. They have presented at the beginning of the year a program to accelerate data center implants. Of course, they need to regulate that, like everyone else, but they will be more linked to renewable sources to access that and to give a premium to that type of capacities, sustainable levels, but not with the very, very almost impossible requirements that the Spanish regulator has put in front of us.

Okay, so that is on legislation. Regarding REN, as you said, there has been call of interest at the beginning of this year. We wanted to get more power for the capacity we have in our data center in Portugal, where you may recall that we are doing right now five buildings of 180 MW with a repowering of additional 20 more, up to 200 MW, as we described before. We have more capacity within the campus to enlarge that computing capacity. That reason why, seeing that there has been a lot of power available in the substation nearby, and we have a massive installation from EDP surrounding us, we wanted to take advantage of that.

We participate in that call, and that call has selected the entities that can move to the next round, which will be an auction process that, as you perfectly said, will be expected to be assigned in Q1 2027. We will give you more visibility on the annual results presentation in February, because by then, probably, either the auction has been already cured or is about to cure, so we will have more visibility there. But yeah, we can confirm we are participating in that auction.

Javier Echave
CFO, MERLIN Properties SOCIMI

Okay, because of that, well, thank you, Mariano. Well, probably I should say first of all that although I bring the grand job title of CFO, the actual boss is Inés, as you know, and therefore I am going to abide to the disclosure principles that she has been handling.

Therefore, I am afraid I will not be able to disclose details of spread over time. But probably the only thing I can remind you is that 99.8% of our debt is fixed debt, and we have no plans to rush into debt markets in the following months. Therefore, I think at this stage, probably the best that you can do is assume that there is going to be no major changes in the short term.

Inés Arellano
Director, MERLIN Properties SOCIMI

Mariano, as we always do, FFO guidance is always provided in February for the year, and the only thing that we have done with data centers to guide you is to talk on a stabilized basis. But bear with us, we have never given guidance on an annual, biannual basis on FFO because this can be very binary, okay? So, sorry, but we are not going to provide guidance on yearly FFO, apart from the one that we have already done. So stay tuned for February to get exactly the figure that you need for your model. Of course, we can talk offline about how you are thinking about, as we always do with all of you guys.

Mariano Miguel Hidalgo
Analyst, Alantra

Okay. Thank you very much.

Inés Arellano
Director, MERLIN Properties SOCIMI

Thank you, Mariano . The next question comes from the line of Thomas Rothaeusler from Deutsche Bank. Thomas, the line is yours.

Thomas Rothaeusler
Analyst, Deutsche Bank

Hi. Couple of questions. The first one is on the royal decree scenario. To what extent do you expect to be able to compensate the higher CapEx by higher pricing, and what is the impact on your expected yield on cost, finally? Then on tenant diversification, assuming you would be successful in closing this large-scale data center lease you refer to, how would this dilute the core REIT exposure? Is it the 56% exposure you show in some of the slides? Then on pricing, you highlighted strong ongoing pricing power. Just wondering what is the magnitude here? Is it still the EUR 120 per kilowatt, or is this a totally different number in the meantime?

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Okay. Look, regarding the royal decree, what we have done is simply a worst-case scenario if we see that there is no change of wind on November 29. First, if the royal decree finally is issued by the government, they have three possibilities to do it on the November 17, November 20, and November 24. If there is no change of wind in the next elections, and if we cannot obtain precautionary measures from the Supreme Court and/or obtain protection from the European Union, then of course, the worst-case scenario that we have put in front of you will become the scenario, will become the base case scenario.

If that base case scenario finally crystallizes, in order to calculate the yield on cost, you have to multiply our current yield on cost by 0.93. So, basically to reflect a 7% lower profitability as a consequence of increased CapEx. Then regarding the core REIT exposure, if and when we announce Portugal, Fran will comment on it.

Fran Rivas
Director, MERLIN Properties SOCIMI

Yes, Thomas, you have this in page 16. What it says basically is when the capacity is entering. So if you do a pro forma, then would be a big decrease on because if you are adding to the current capacity, all of a sudden, the 180 MW that we have in Portugal, we have repowering another 20 more, all what Ismael described, or the 68 booked, et cetera, then of course you have a drop, significant drop to the 19% immediately from signing perspective. Of course, because the capacity is entering in different moments, so you have 20, in case of Lisbon, the capacity will enter in 2027, and then will enter in 2028. What we have booked that we comment here is capacity that is coming in 2028, in 2029.

All of that capacity, you will see some movement. This is what I was advancing before. It is not immediate drop on the perception of cash, but this is an immediate drop on the percentage of contract signed or let or pre-let. That is where you would see that diversification. Regarding rents, we are disclosing this in page 11 when we talk about the stabilized GRI. This is, of course, a mix of the levels. To your question of the 120, I think we have already abandoned the 120 already in phase I. We are reporting an average of 126 in phase I, and 132 in phase II, which we have pretty good visibility on it. Then phase III, 135.

You are seeing how rents are going up a little bit as a consequence of clients being in a position to pay more for that capacity, justified as well by the fact that end users are also paying more to our customers. That is basically the translation from AI cost of token into more modest or humble real estate increase. That is the reason why you are seeing this. You also remember from the Capital Markets Day when we announced phase III and we reported phase II, at the time of capital increase, we said basically that we would be around 121.5 for phase II. We are reporting 132 already. As is my comment before, we are already three years more in phase III from where we reported just six months ago. That is how rents are progressing. I think that 120 is already a little bit of the past.

Inés Arellano
Director, MERLIN Properties SOCIMI

Okay. If there are no more questions, the next question comes from the line of Florent Laroche-Joubert from ODDO BHF. Florent, the line is yours.

Florent Laroche-Joubert
Analyst, ODDO BHF

Yes. Good afternoon. Thank you to take my questions. Maybe I would have two questions. Maybe the first one on the valuation of the assets in case we are in the worst-case scenario. What are your discussions with the appraisers on this scenario? Do you think that they could factor or they will factor this scenario if applied at the end of the year? I understand that, for example, the yield on cost will be 7% lower. What can we expect in terms of value equation? Maybe my second question, this is about more globally the resilience. How do you see the resilience of the demand for data center for potential tenants here in Spain or Portugal due to all this political debate?

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Okay. Regarding the sentiment of tenants are completely unaffected by all this. In fact, what this is now creating is a little farther tension because one thing that I believe is emerging clearer and clearer in this market of data centers is that potential supply is not going to flow as easily as people thought. For a number of months or years, we have been surprised, having experience in other asset classes, but not in data centers, we were surprised that the clientele evolved from arguing about your requested compulsory periods to shorten them, that was year 2023, 2024, to more recently arguing about your compulsory periods to lengthen them. You would offer the client 10 years compulsory, and they would say, "No, I prefer 15." It is the first time in my life that I see that.

You say 15, and they say 20. It's curious. Why they were doing that? What were they seeing that we were not seeing? What they were seeing was all the stupidity in the Western world surrounding NIMBY and all that bullshit. They were coming from the future. They were coming from the U.S., they were coming from the Nordics, and they knew that sooner or later, China would end up fostering opposition to data centers everywhere in the world, except in China, where you cannot oppose anything. It's been curious that we have been signing leases beyond what we initially expected, and this is why.

Regarding the valuation, the appraisers, I believe that if you subtract a 7% profitability to a 14.5% yield, you will end up in a 13.5% yield. The appraisers will simply compress that to the prevailing yield they want to consider as a scenario, say 6%. You might think, "Okay, 6%, but not on CoreWeave." It is a real pity what is happening with CoreWeave, because we are receiving a lot of stupid calls of people telling us, "Why don't you cancel the CoreWeave contract in the Basque Country?" For God's sake. I mean, CoreWeave has more computing capacity than the next 10 neocloud companies. CoreWeave is the best neocloud operator out there. CoreWeave, it is a pity that they are put in doubt constantly in what, to me, is a textbook short-selling attack. I mean, where they have first done the operation, and second, they have created the narrative.

A lot of people is following the narrative as if they were wildebeest in the Kilimanjaro National Park, and they are falling up the cliff. It is a real pity. I mean, we are in contact with them on a daily basis, and we see their numbers, we see how they are progressing, we see the evolution of the underlying contract, we see the utilization rate of our data centers, we see the amount of equipment they are bringing in the Basque Country, where they are taking possession on December 20. They have already unloaded EUR 2 billion worth of equipment. It is a real pity to see that all the sudden, CoreWeave is like a big disaster, and unless you lease a data center to Microsoft or Google, even Oracle is now out of this category.

Unless you lease to Microsoft or Google, you cannot lease a data center, for God's sake. I mean, thank God the appraisers are a little bit out of all the artificial intelligence noise, et cetera, and what they see is a compulsory contract that can be executed according to Spanish or Portuguese law. Therefore, it is a bulletproof contract in their mind. Maybe not in the mind of the market, because as during the COVID, as in COVID, many of you thought that we will lose our rents, and we did not lose anything. I mean, because we are still in a rule of law country. I mean, and we have courts, and we charged all of our rents, and we collected all of our rents without a problem, flawlessly. With the data centers, it is going to be absolutely the same.

But this is probably better perceived by appraisers than it is perceived by young people in the market. It is a pity that we are under this kind of discussion permanently regarding CoreWeave, because that discussion is going to be replicated when we announce Portugal. When we announce Portugal, there will be doubts about whether the client in Portugal will be able to pay. As I told you, when we announce Zaragoza, unless it is Microsoft, you will always raise doubt on the quality of the tenant we put in Zaragoza. Then when we announce Extremadura, unless it is Google, you will raise doubt about the people we will put in Extremadura to pay rents to us. So, it is an endless exercise of doubts on everything. I believe this is a real pity.

We should not get dragged into this situation, because it is a textbook short-selling attack, in my opinion. The company is performing impeccably well. Yes, they are incurring a lot of CapEx as a consequence of their expansion. By the way, we have talked to them about, "Why do not you stop your expansion in order to tell the market that you are more profitable than the market thinks?" What they say is that, "I cannot stop my expansion because I need to be relevant if I want to face, in the same level, I want to face the hyperscalers in the future in terms of AI computing." Okay? Well, this is the situation, and I believe valuations, as of year-end, will not move one single comma as a consequence of the noise surrounding CoreWeave and the artificial intelligence.

I believe at the end of the year, there should be a significant valuation pickup in Arasur 02, because it will be full and it will reflect the value of the contract, although it is already known by the appraisers. They know we already have signed, and they have examined the copy of that contract, so they know the contract is already in place. But I guess there should be some value pickup, and then it will depend on having new licenses. Particularly, the bookings will play a role, because Tres Cantos empty is not the same as Tres Cantos booked, and the repowering of Portugal non-existing is not the same as the repowering of Portugal existing. Okay? So, I believe there should be some valuation pickups as of end of the year.

Florent Laroche-Joubert
Analyst, ODDO BHF

Okay. Thank you. That is all we have today. Thank you.

Inés Arellano
Director, MERLIN Properties SOCIMI

The next question comes from the line of Aaron from Citi. Aaron, the line is yours.

Aaron Guy
Analyst, Citi

Yeah. Thanks very much for the call. Could you just provide a little bit more color around the sort of demand dynamics? Last time we spoke, there was a queue of tenants, sort of essentially queuing up for each of the data centers that you were producing. Can you talk about the potential impact that you are seeing in the NIMBYism potentially from Have you seen an increase in demand of people saying, "Well, I cannot get the capacity in the U.S., I am going to come to you guys now"? I know you have talked about the kilowatt pricing increasing so far, but could that start to accelerate because of this?

And then, on the other side of the coin, could you just talk a little bit about the dynamics of the procurement market that you are seeing in terms of the electrical, mechanical infrastructure? Are there any delays or anything that we could see on that side?

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Okay. Do you want to do it, Fran?

Fran Rivas
Director, MERLIN Properties SOCIMI

Yes. Starting from the procurement market, this is not new to us, and I think we shared in several other calls that, little by little, there is more capacity to be built, and therefore, there will be more scarcity of equipment in the market, something that we have seen in the U.S. It will come at some point in time more heavily into Europe. That is the reason why we are refinancing in advance, taking lines in advance, and trying to anticipate all the CapEx or all the commitments that we report normally on a quarterly basis of the CapEx or the equipment that we are buying upfront. We are seeing in some of the components that there are long leads. So far, they are not exceeding the length of the construction timelines that we are managing.

If you are building on a 24 - 30 months type of development, so far there is no equipment that exceeds that length. But in some cases, mainly on transformers and other type of generators, et cetera, you are getting there. In the U.S., it is a little bit different. It is another type as well of equipment, and they are already experiencing those challenges. This is something that when we are taking, we anticipate some financing from the equity or from the debt perspective, is in a way to mitigate this type of consequences. In terms of the demand that you were saying and the pricing, as I said before, this is a little bit of consequence of what the GPU operators and end users are reporting to the markets, not in sizes, but also in the cost per token and in the cost of that capacity.

As I was commenting before, going to the call with point of view, they are getting roughly like $20 million-$25 million per megawatt when they are selling long-term leases. That amount has increased to $40 million-$45 million per megawatt on short leases. The impact on the growth of that capacity is making that they can afford to pay higher rents. To put it in context, our capacity, for example, in Lisbon, the 180 MW, from what we have heard for several clients, to give you an idea, there is no capacity of these deadlines of 2027 and 2028 deliveries available in the world.

Right now, clients more and more are putting pressure on us to contract capacity that in some of the cases, seeing, of course, the track record that we are having in delivering on time, trying to contract capacity that we haven't even started. For us, this is a good thing and a bad thing at the same time. Good thing because we can, of course, anticipate that the risk projects were in advance, but that's risk as well because we need to make sure we deliver on time. What we are doing in the same way that valuations only enter into production when we have construction started, we are trying to do the same with the capacity we have. That's the reason why we talk a little bit very confident on what we are seeing.

There is a lot around of demand out there, but the problem, as Ismael Clemente was pointing out in the question before, one of the question marks is whether that CapEx can be deployed in the timeline because there is no data centers. If you go to, for example, the Form S-1 of Anthropic that they submitted to the regulator in the U.S., the common risk that appears more in all the document is the scarcity of data centers. When they are challenged of what are the things you can find in the market that can prevent you to get to the expansion you are foreseeing, the number one is scarcity of data centers. Then it comes regulation, it comes then sustainability, but of course, computing is not in or is at the very end of the list of computing demands.

Data centers is the first one, and this is what we are seeing. As I said, apart from the tone in the market, I think we are in a very good position to meet our targets.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Regarding NIMBYism, Aaron, NIMBYism is very, very important at a regulation level and during the planning phases. When you are planning and you start publishing your plans, is when you normally experience opposition and NIMBYism. However, if you look at our phase I, II, and III deployment, it is all under construction, basically, with the exception of two projects in Zaragoza, which are still pending to obtain the licensing. Once you are under construction, there is no NIMBYism. I mean, you have a construction plot and you are basically finishing your civil construction and then your equipment, and you are relatively out of all that noise. But it clearly will affect in the future.

I mean, the world is becoming complicated, and if you put together the effects of, for example, regulations like the Spanish royal decree together with the NIMBYism, it is easy to see that in the future you are going to be a super high market share operator in a tiny country. Frankly speaking, I do not like that. I would prefer to be a 10% market share operator in a big data center market in Spain, Portugal, rather than a 40%-45% market share operator in a tiny Spanish market that has been constrained by the royal decree and by the general public stupidity about data centers. It is amazing the public reaction.

We have the luxury of having existing facilities in operation, and we are taking now the decision to move people, fly people, or move people in buses to places where we already have a data center so that they can check by themselves the heat island effect. We go with a thermometer to the freaking roof of the data center, and we measure with the thermometer the temperature right next to the fans, and then we go down to the entrance door of the data center, and we measure the temperature, and they see that there is no freaking variation. But it is the only way to convince people that all the information they are receiving, they are constantly bombarded in the media, in the press, is completely fake. Likewise, noise.

We move people to data centers with a sonometer, and we measure the noise, 65 decibels, and then they see the noise screens. When it abandons the roof, there is already 50 decibels, and 50 decibels basically is inaudible at 100 meters. Only by seeing that, people is starting to understand that they are constantly misled by the information published by a number of media and by a number of activist groups. We have been confronted on many occasions with the famous NASA reports. NASA, at the end, we investigated what this meant, and it was a reference to a study conducted by an assistant teacher in Cambridge called Andrea Marinoni that had used NASA photographs. The study itself had nothing to do with NASA, but the activists say publicly that NASA has said that data centers increase the world temperature.

The study, by the way, was about the thermal increase when you make waterproof part of what previously was rural land. When you convert rural land into a logistic shed, a hospital, a road, or a data center, the temperature of the land increases by about 2 degrees. This is owing to the tarmac, not to the data center itself. It is really, really painful to be tackling these type of issues every day. Of course, we do it in a very direct way. I am going myself to explain the data center projects to the villagers in every village we start a project. I go there, and I take my time to receive as many questions, talk to them openly, because we have nothing to hide.

Data center is quite an innocuous kind of thing from a sustainability standpoint, but we continue being bombarded by all that bullshit.

Aaron Guy
Analyst, Citi

Yeah. Understood. Very clear. Just a couple of quick technical follow-ups. On the property valuation, after practical completion, given that there is not really any transactions in the market, can you just remind us how the depreciation of the data centers will happen, over what time horizon, and how it will be phased? In the worst-case scenario of a tenant going bust, can you just remind us what rights are in your contract of seizing tenants' equipment, the computer equipment inside, and paying default or late payment fees? Thanks.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Okay. Well, regarding the valuations and the depreciation, again, this is a situation where there is a big, big, big mistake in the market between accounting depreciation, which equals, in reality, tax depreciation, which is what leads all this thematic, compared to real useful life of equipment. We have conducted a thorough analysis, piece of equipment by piece of equipment. Out of the total of a data center, you have 30% is civil works. The obsolescence of civil works is pretty much meaningless. Then you have 70% which people calls equipment. In reality, equipment, physical equipment, is around 45%. The rest is cabling and a lot of industrial installations, which are also more or less eternal. Then on the other 45%, there are things like gensets, for example, that we depreciate linearly over 20 years.

But the average useful life of an MTU 85 L 20V Series 4000 generator can be easily 50 years or 100,000 hours, and we are doing 20 hours, 27 hours per year. So the generator will normally reach year 20 with maybe 500 hours, compared to useful life of 100,000. So, if there is anything that needs to be replaced, probably it is going to be the electronic management system. Other than that, sometimes the alternator. You can replace the alternator in year 30 if you want to extend the useful life. But this is one of the costlier pieces of equipment in a data center, and there is a big discrepancy between real useful life and depreciation, accounting depreciation, which equals tax depreciation.

If you are a self-builder, if you are AWS, if you are Microsoft, of course you want to shorten the depreciation of your equipment, because by shortening the depreciation of your equipment, you basically reduce your taxable income. So, that has misled the market to believe that there is a lot of obsolescence that even our clients exaggerate the obsolescence of chips. They go to authorities and they say, "Well, in four years, the chips are going to be burnt." So they depreciate over four years, 25% each year, and we can tell you that on year three, they are still running at 93% of capacity. So, there is no way they are going to be at zero in year four, okay? But everyone is playing with their own interest, and they are misleading the market, and the market is understanding that tax depreciation is the same as useful life.

Same with transformers. Transformers are, I mean, particularly dry transformers, like the ones we use, which are not subject to problems with oil, et cetera, they are almost eternal. So, you can use a transformer for many, many years. The only thing which is relatively short-lived is batteries. Batteries, normally, we underwrite for 10 years usage, and we know that as from year seven, normally, the performance starts dropping a little bit. So, this is why we normally buy batteries with 12 minutes capacity, because we know that on year seven, they are going to go to nine minutes capacity. Still, more than enough to compensate for the 90 seconds you need to start up a generator. So, this is the only thing in which we normally, the depreciation we use is relatively similar to the useful life.

But still, there is a lot of residual value, because that battery with 70% useful life is still highly valuable for many countries in the world. I mean, you discard it because you need it with 100%, but in many countries of Africa, they will kill to obtain that battery, and you will see that your residual value is very far from zero. It can be easily 20%, 30%, 40% of what you expected. So, there is a lot of noise regarding that. In our internal calculations, the loss of profitability due to depreciation can oscillate between 1.2% and 1.5%. That is really if you want to replace everything every year. If you were to endow a sinking fund, this is what you should be putting in the sinking fund. But you have been also as a real estate analyst for life.

So, imagine what would be the sinking fund endowment for a hotel, for a shopping center, if we were to endow sinking funds in every asset class in which we are operating. Of course, in logistic sheds or in offices is different, but in assets which are highly intensive in people's traffic, the sinking fund will be very significant, even higher in our internal calculations than the one we would use for a data center.

Fran Rivas
Director, MERLIN Properties SOCIMI

And in terms of the default situation, as I was explaining before, we have different guarantees. It's very similar, like an office building, like a shopping center, or a logistic. So you have what the law mandates, which is what we call statutory rent deposits, and this is by law, so you have this in the contracts. Then you have normally requests to the different clients, whether parent or third-party guarantees or additional cash deposits. Of course, all these guarantees, which are cash or access to cash, you cannot have them for the 100% of the length of the contract, okay? So this is basically covering one year, two years, like in every single asset class. So you cannot think that any client will give you a guarantee for 100% of their backlog first day. Exactly the same as we have with BBVA at the time, okay?

Second is that depending on the contract, and you can ask, sometimes it's capped, sometimes not, you can ask for all the amount of money of the lease in case of a default situation. As we were explaining before, the equipment in this type of assets is a significant amount. So, as we were describing before, 4x our total investment, and out of our investment, we get, as you have seen, like 15% yield on cost. So, the amount of lease that they need to pay to us with regards to the total amount of equipment they're supposed to, and that we are securing in our buildings, that we are maintaining temperature and humidity and air conditions through SLAs. Of course, all of this is subject, which is on the maintenance of the machines, this is subject to us receiving rents.

That's the reason why, as we were saying before, this is a bad business for them to stop paying, and that's what we believe. In all the structures we have seen as well from the financing perspective, as I was describing before, power and rent come the first, because it's more important to maintain the GPUs alive and in good maintenance conditions in a well-secure area than even think of continuing operating. That's the reason why default risk in these cases, we believe, is well covered.

Aaron Guy
Analyst, Citi

Okay. Thank you. Have a good weekend.

Inés Arellano
Director, MERLIN Properties SOCIMI

Thank you.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Thank you.

Inés Arellano
Director, MERLIN Properties SOCIMI

We have received several questions through the platform from people that could not register to do them through the phone. I will try to summarize them because some of you have actually raised different questions at different times. I am sorry if I miss any of them. I will try to put them all together. The first questions, they come from Jose Ocina from CaixaBank . They are all about data centers, review of phase III. He is asking if we are taking into account the government planning for increasing the total capacity of the transmission network in Spain and the 9.3 GW allocated for data center. This has been the news in Expansión today. He is raising the question as well, I think Ismael has already answered this one about the contingency plan.

Is it going to be shelved in due time, or is it the final firm plan to be undertaken by the coming years? I mean, what we've presented is our base case.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

We'll comment.

Inés Arellano
Director, MERLIN Properties SOCIMI

The contingency. There's a question as well about Bilbao-Arasur 04 and 05, that as you know, we've taken aside from phase III. The question is if Solaria is the energy provider. There's another question on the royal decree. Are we going to be challenging the royal decree at the National Court on the Contencioso-Administrativo? What is going to be our strategy? If we're going to be doing it ourselves or teaming up with some other affected parties. Hopefully, Jose, I've covered all your questions here, and then let us address them.

Ismael Clemente
CEO, MERLIN Properties SOCIMI

Okay. Well, first, regarding the increase of available power to 9.3 GW for data centers, that increase is subject to royal decree. If you make use of that power, or if you request that power and obtain it from the national grid authority, you will obtain that power subject to complying with the royal decree. That means with the reduced profitability of the royal decree. You will need to be very, very selective in choosing only to obtain power in those locations where you can carry out a very significant self-consumption strategy, particularly if you can do it behind the meter. Because if you need to get the PPAs through the distribution or the transport network, you are going to be subject to the tolls and taxes, and therefore, the price of energy will be prohibitive. Our calculations are made with the existing power we already have.

We are not taking into account the 9.3 GW that will be put or that will be made available by the government for data centers in the future. If and when that power is put in the market, we will review on a case-by-case the different locations and scenarios. In principle, looking at our initial numbers, it will be much better to go and look for opportunities elsewhere in Europe than to develop here. We will, of course, be open to review different opportunities that will be put in the market. Regarding whether the plan will be shelved, what we will be shelving is the BESS increased CapEx part of our program.

The definition of phase III would remain equal, because in the case of Arasur 04 and 05, until we cannot get 100% certainty that we are going to get the electricity, we will put it in the freezer, because we cannot announce to the market a project in which the electricity is still in doubt. Remember what happened recently to Australian operator NEXTDC, and we cannot do that. We only consider for deployment the projects in which we have 100% certainty that we are going to get the electricity, not only at the sourcing and granting level, but also at the infrastructure works level, which sometimes are done by ourselves, sometimes are done by a third party.

As commented with you, particularly when that third party is a renewable generator, we have always to be a little prudent, because those counterparties, in many cases, are not institutional quality. We have to be careful with that. Regarding who was the counterparty in the Basque Country, yes, it was Solaria, but thank God, we had an alternative plan. We are now analyzing with them the possibility of an alternative plan by moving the consumption to another area of Spain with a higher degree of self-consumption capacity, because it seems clear that they will not be in a position to supply us with the committed electricity in the Basque Country. Regarding legal strategy, look, I hope you will understand that we don't reveal our legal strategy.

Of course, we will fight, not by caprice, not by ideology, simply because we have been significantly affected by the royal decree, economically speaking. We have situations like Arasur 01, or we have situations like Tres Cantos, where at the time we designed the data centers, or even Zaragoza, despite the high consumption level, at the time we designed the data centers, which is normally like two years ago, we didn't take into account the problematic of the royal decree. We didn't take into account the need to reserve buildability for a BESS. We didn't take into account that we needed to lower the IT capacity in order to leave enough spare room in the capacity of the line to charge the batteries.

Now, the new royal decree basically makes those projects or renders those projects, I wouldn't say impossible, but at least very expensive to carry out, because you need to start from a blank sheet of paper, imagining solutions for what was not designed for that kind of regulation.