Good morning, everyone. Welcome to Arasur. We are going to do now the main part of the presentation, which is the introduction to Phase III. The presentation is going to start with a discussion on how the AI workloads are evolving in the world and what is the relative situation in Spain. Then we will comment about what is the MERLIN's relative positioning and which are our main competitive advantages, particularly in design and construction. You will have the opportunity to meet our design and construction team. We are going to specifically focus on Phase III, what it's all about. No mystery, no wonder, it's simply making use of all of our existing power land.
I mean, it would be criminal not to do it because now that everyone is looking for power land and we have it, of course, the first thing we have to do is make full use of it. Fran is going to zoom in on the pipeline because at the end, you need to give continuity to the current effort we are making in data centers. He will distinguish the different types of pipeline and which are the definitions we apply internally to consider that something is long-term versus short-term pipeline, and you will see what is coming in the future in terms of total capacity. Okay. Data Centers Opportunity. The global demand for AI workloads is estimated to more than triple between 2025 and 2030.
These are McKinsey numbers. This creates an enormous opportunity for countries like Spain and for companies like us, which despite being relatively new to the business, we are native in AI data centers. The cost of compute will exceed $7 trillion, and you know that there is mounting criticism to the hyperscalers because of the sheer amount of CapEx that they are devoting to this business. I believe this will eventually end up in a certain bifurcation and an opportunity for people like us because we will become producers of facilities that will eventually be leased either by new clouds or in many cases, in many more cases than you think, hyperscalers. What happens in Spain? Spain is in the Flintstones age of data centers.
You can see that the current estimated capacity is little above 300 MW, of which almost half is hyperscale, but the vast majority continues to be colocation. I'm not sure that these 300 MW are in full operation as of today. It is very, very difficult to know exactly what is consuming electricity and what is just being finished. Because if you look at this building there, it looks finished, but it's being equipped. It will not enter into operation till at least next year. Okay? Regarding our competition, if we look forward to 2030 to those 1.8 GW of total capacity that should be ready by 2030, these are what we believe are the most credible competitors at the time.
Iron Mountain, which is doing a facility in San Fernando, Madrid, with two different phases, and they will be reaching close to 80 MW there. ACS, which through the joint venture with GIP, we expect that they are in a position to develop around 130 MW, including their facilities in Alcalá de Henares in Madrid and what they are developing in Zaragoza. Iberdrola? But at least the project they are doing in southern Madrid with Echelon, I believe this is going to be Echelon or Avalon, is going to be finalized. QTS in Calatorao in Zaragoza, which is backed by Blackstone.
We still do not know whether they are finally going to develop the project from A to Z, or they are simply going to develop the first phase of a project, create the expectation, and then flip it. You know, eventually, it's a credible project. AWS, which is by far the most advanced AI operator in Spain, but with a characteristic peculiar characteristic, which is that they are completely focused on self-building. They are self-building, and therefore, they are not our competitors. They are not competing to capture the clientele that we are trying to capture. However, it is true that ultimately they can be competitors of our clients. This is why we include them in this graphic.
Out of the EUR 1.8 billion by 2030, we are going to be in the region of EUR 630 million, okay? Reaching more than EUR 700 million in the year, in the following year. Why Iberia? Why are we so happy to be here and why have we developed this, you know, line of production focused on data centers in Iberia? Because as commented in the presentation that you saw before with Nemesio Fernández-Cuesta, this is a land with abundant renewable energy resources which, you know, first feed into your data centers and give you the possibility to open new projects when you locate your projects close to where the generation is made.
Second, they feed you also with a very interesting cost, competitive advantage embedded in the price that you can obtain from the generation operators. It is also a country with a relatively scarce population that is relatively abundant land. That's important because, you know, the reason why more data centers are not being developed in the Netherlands is not electricity. It's mainly land. I mean, it's a little bit like logistics. I mean, it's complicated to develop new land in countries which are relatively tiny. Same is happening with Japan.
The key characteristic which differentiates our market compared to many other, the only one that really resembles Iberia is the U.K., which is very significantly interlinked with the rest of the world, is the cable, submarine cable communications. I mean, in the Iberian Peninsula, you have, like, three main entrances of cables in Lisbon, here in the north, and then in Barcelona. You know, through the Iberian map, you will link those three entrances and funnel them into Europe, okay? It's a very adequate place to be located if you are particularly when things rotate from model training to inference. Because although today the latency with the U.S. is still here in the north of Spain, in the Basque Country, is around 60 ms.
The new cable that Google is laying through the Atlantic that will reach Lisbon, which is Nuvem, is said to be around 30 ms latency, which is more than enough for most inference charges, okay? It's very interesting. The limit, the physical limit is the speed of light, which in theory will put us around 10 ms from the U.S. Technically, the evolution of the cable business will be such that, you know, from that 30 ms, it will be always eating into, you know, the latency up to the moment in which they've reached the physical limit, which is speed of light.
Moving now to the second part and second chapter of the presentation, this section aims to drive you through the main three pillars of MERLIN strategy when we started at the data center business. The first one is that we built on expertise, so we have an extensive know-how that you will be able to know today meeting our people, and it allows us to meet customer needs, but also to manage very well the supply chain and access to competitive and sustainable power. The second pillar is that we try always to lead from the front, we are trying to deploy our capacity over what we call the Iberian Digital Diagonal that we will go in detail in a minute, and also proven development capacity in all the different locations that we are choosing.
Finally, third pillar, we have the capacity to deploy capital, obtaining returns, as you have seen in Phase I, more consolidated, Phase II, ongoing, while maintaining a balance sheet strength, which is also a very key element for us. Entering more, in more detail on our extensive know-how, we have, like, four areas of excellence that we divided in four teams. These four areas are the following. Our energy team, it allow us to access to the grid to have a very renewable and efficiency in the way we access to that power, and we have a special presentation right after this one. Also be aware of all the regulatory framework that is affecting all of our access, which is very key in this industry.
The second area is our design and construction team. We have the luxury of having an in-house project management, in-house, but also the capacity we have of constructing based on modular projects. This makes that our cost control and our high density is very forefront in the sector. Our third area is the technology team. For that one, as you all know, we have a long-term industrial partnership with Edged, global data center technology specialist. With that partnership with Edged, we access to proprietary ultra-efficient cooling technology, very, very relevant, not only because we have a PUE of 1.15 on average, days like today, probably much lower, with no water consumption.
We have access to clients from the technological departments to understand what they are thinking that we'll need in a couple of years. Any time that we start a building or we start a development, we need to know what the market will be in two years so we can meet those expectations when we have product completed. Finally, our fourth area is the operations and maintenance team, a 24/7 operational management that continues our relationship with clients, but also with their maintenance, allow us to get, in our case, nine certifications that proves, you know, the excellence of this, of this area. In total, we have 301 dedicated people, and most of them, you will see them, today.
Of course, all of this wouldn't make sense if we are not able to meet customer needs, which is at the end, the goal of this project.
We are now excited to share how MERLIN is positioning itself at the forefront of the data center revolution by focusing on the infrastructure needs of our AI and hyperscaler clients. Let me start first with what the market is demanding in terms of capacity density. Today, we are witnessing a dramatic shift in capacity density requirements as the industry is evolving from digitalization to AI workloads. Traditionally, colocation required no more than 10 kW / rack. Today, AI inference and AI training requires up to 200 kW / rack. That's a 20-fold increase in a very short period of time. This density revolution requires ultra-efficiency cooling systems, both air-cooled for moderate to high density requirements, but also liquid cooling for very, very demanding AI training, and AI inference.
As in MERLIN, we expect that 70% of the demand will come from high-end AI load densities. We are already future-proofing our infrastructure to be ready to adapt to everything from hyperscaler and cloud demands of 20 kW/rack-70 kW/ rack all the way up to 200 kW /r ack for the more demanding AI workloads. Across the risk-return spectrum, we identify four different business models. First, at the lower end of the risk-return spectrum, there is powered land and shell, which is essentially leasing of land and buildings with power. At the opposite end of the spectrum, we have retail colocation with contracts at the rack level for a short period of time and a very large number of clients.
In MERLIN, where we find the sweet spot is at the hyperscaler and wholesale, colocation, business model. In this business model is what allow us to dedicate and to deploy, significant capital under, attractive, risk returns, okay? This business model, what we do is leasing whole entire facilities to the best-in-class AI and hyperscale clients on a long-term basis. We remain responsible for the mechanical and electrical systems, and the controlled operational exposure provide us with, increased returns. What sets MERLIN apart, it's this commercial platform and client relationships. We have built a dedicated in-house commercial team with international presence with, representatives in London and Amsterdam, but also across the pond. We are a qualified NVIDIA preferred partner, which allow us to participate in very early stages of our clients', planning processes.
At the hyperscaler front, we have successfully passed the standards of qualifications of three out of the four main hyperscalers. Our facilities are specifically designed for large-scale developments, making MERLIN a reliable partner for long-term strategic expansion projects. On the AI player side, we support ultra-high densities, up to 200 kW/ rack. We have very solid relationships with all tier one players, and we have proven our support to NVIDIA cloud partners. In summary, MERLIN is at the intersection of three key pillar factors from the AI and the hyperscaler perspective, which is we have proven expertise in high-density deployments, we have solid client relationships, and we have demonstrated capacity to deploy capital at scale.
Having the luxury as landlords to have these prestigious tenants, of course, we need to make sure that we have an excellent management of the supply chain. The only way to achieve that is if we have an industrialized process, both on the development and in operation that my colleagues will describe in depth.
I'm going to talk about the first phase of the life of a data center, that is the design. The design involves from the very beginning of the process till the end when we end the construction, and we hand over the facility to operations. This means the first thing that we do is a fit of a potential plot that could be interesting for us. In that case, that is good for us. What we do is start with a schematic design development, construction documents, and also the design team is also involved in all the construction phase till we finish the facility, and we hand over to operation.
Another thing that we have been developing, apart from specific projects like these ones, is different typologies of buildings. What we are doing is just to be ready for the future, to be very quick in terms of the phases or the phase of the designs. These new typologies that we are doing are going to be ready for all the different hyperscales and for the AI players that we are going to have in the future. All our typologies are going to be ready for this next few that we are going to be.
Given the long track record of the company as a leading real estate company, we are familiar with construction. We feel very comfortable with construction. We know the risks, we know the players, and we know how the industry works.
In fact, we have proven the ability of the company to develop data centers. Phase I and Phase II are clear examples of this. The projects we will be presenting to you today represent a bigger job. In order to deliver as well as we did with Phase I and Phase II, the company has put in place two key steps. First one is an in-company project management team that will oversee all the processes, all the stages of the development from the land acquisition till the operation, from the design to the execution or to the construction. This is the first key step. The second one is working together with top-tier construction companies and top-tier engineering companies.
We have designed a fast-track process that will help us shorten time and de-risk all the execution. Later on, I will give you a couple of examples of how we put this in practice.
Time to talk about equipment. I will use three concepts to describe equipment. First one is proximity. We look at EU as our, you know, preferred space to look for equipment, which means they are close to the final site. We reduce the risk of logistics, and we minimize the time to deliver the equipment from factory to the site. Secondly, we make it repeatable, okay? We, when we say we standardize the design, we are also standardizing the specifications of those equipment. We don't have to rethink about, you know, specs of a given piece of equipment every time we need to buy a new one, so we can get that easily from the different manufacturers.
Finally, we make them always that is possible unique, and this is especially relevant when talking about our cooling systems. We have a proprietary design for the cooling system, which give us, you know, an advantage in the sector because no one else is using the same equipment we are using, and that is giving us some special features we are proud of, such as, you know, low PUE, no water usage or things like that.
Regarding operations, we have industrialized all the operations task, transforming from individual task into integrated industrial processes. How did we do? Building around two pillars. The first one is looking the excellence. We have created our own certification, which is, which designs the standards, define the rules, the processes, and everything across our data center. The second pillar is the training. We are looking for highly trained people. This is working with the universities. Even we have our own MERLIN Data Center University. Operational excellence cannot exist without highly trained people. Combining both pillars is the only way that we have to adapt and be fully aligned with the growing demands that we have in the portfolio.
Our objective is to provide a competitive and sustainable energy. We'll see more details later, but we have seen two different kinds of PPAs. The traditional PPA is when the generator is connected directly in the public grid, transport or distribution grid, and our data center are connected in another place, in this grid, in the transport or distribution grid. The generator delivers the energy to the public grid, and our data center consume this energy. For this reason, we use the grid. For this reason, we have to pay the grid tariffs. On the other side, we have the self-consumption with surplus PPA. In this case, our data center is connected in the generator grid. We don't use the public grid.
All the generation is consumed by our data center, and only the surplus energy is delivered to the public grid. In this case, we don't use the grid. For this reason, we don't have to pay the grid tariffs.
With all of these aforementioned capacities in mind, we do believe MERLIN is leading from the front. In fact, we're consolidating the Iberian Peninsula as a data center hub, mainly thanks to five key advantages. First and foremost, our location. The Iberian Peninsula is Europe's digital gateway. It's strategically positioned to connect with Africa and the Americas. In second place, workforce. As Enrique was mentioning, we have access to a highly qualified engineering talent. We come with top-tier universities and competitive labor cost bases. Additionally, we also come with a network connectivity that acts as a landing point for major subsea cable systems linking global traffic and acting as a key nexus for transcontinental data flow. Most importantly, power. We have a twofold renewable advantage. First, access to abundant renewable energy.
In fact, our renewable energy mix is about 61% versus our European peers, which come with a 45% renewable energy mix. Second, our power cost advantage, which is approximately 35% lower than our European peers. Finally, the space and development capacity. We've got room to scale.
As Ismael and Fran mentioned, MERLIN is tracing what we call the Iberian Digital Diagonal, connecting power, fiber, and submarine cables. In fact, we count with the data center hubs in Bilbao, Barcelona, and Lisbon, the three main landing subsea cables in the Iberian Peninsula. All of these are interconnected with our remaining data center hubs in Zaragoza, soon to be announced, in Madrid, Extremadura, and Bilbao. Additionally, MERLIN has proven a sound development capacity. We have shorter development timelines than our peers. We continue with our land acquisition and development. In fact, as you all know, we have more than 2.8 million sq m of a logistics footprint, and we're now focusing on developing our land acquisition program for our data center development program. This includes looking for land plots with access to power and capacity to scale.
We have a specific team for managing the electrical connection. We negotiate the energy contract with generator. We negotiate the energy contract with commercialization companies, and the rest of the energy, we buy the energy directly in the spot market.
About design, the typical time for develop a complete project is more or less 8-9 months. We are trying to squeeze, as I've told you before, with that typologies of building that we have developed, but we are trying to squeeze this 8-9 months to six months. We are going to be very quick, ready to start construction.
I will take you through the next three sections. As you know, a data center requires several permits: municipal permits, regional government permits, and also some energy bodies nationwide. Getting the permits at the right time is crucial. It's very, it's a key for the achievement of the delivery. Here, what we do is first we rely on the experience of our in-house management teams, and we apply, as I mentioned before, a fast-track process, meaning that in most cases, even before we have secured the land, the energy team is working with the design team and the planning permitting in order to bring forward all the permitting, all this process. Second, equipment. Equipment is also crucial. Here I'm talking about chillers, generators, and UPSs.
You will have the opportunity to see them, and as you can imagine, you cannot buy this at a general store. Here what we rely on our teams and taking into account that some of these equipment are fungible, meaning that we can put them in one data center or in another, and the great visibility we have over the future projects, we can bring forward all the acquisition processes, securing this way all the development and de-risking this process. Construction, as I mentioned before, we have an in-house management team that oversees all the process. This way, we can monitor every step, we can control the planning, we can control the cost of the whole project and de-risk all this process.
Everything you've seen so far in this program leads us into the possibility of bringing the equipment on site sooner than others are doing right now. By doing that, we have the chance to, you know, overlap in a longer program the startup process and commissioning of the critical equipment that will be at the end almost the last, you know, link in the chain, which is critical because you've taken all the contingencies all over the process, so you have to be, you know, focused on getting those equipment ready by the time we need them. A good example of bringing those equipment at an early stage is if you look through the window, you know, the building we are now, you know, in, under construction, you will see one red crane there.
Even though the building is still being built, you know, you see the shell is still ongoing, we are already sliding in equipment so that we can start with that startup that I just mentioned, you know, earlier than others will do normally.
Thanks to this strategy and the operational model, we are able to accelerate the onboarding process, ensuring the customer reduce their time to market. The people is ready, the processes are completely prepared, and the services and the training of the people are able to give the services to the customers.
With all of this in mind, you are seeing that the date of today is what when we will start with Phase III. In reality, we have 18-20 months of work behind us. Any other announcement of any other project, if they need to go through all the different processes that my colleagues have described, means that at the end, we have 20 months ahead of those players.
In terms of capital deployment, the reality is that phases one and two have been a great way to showcase the strong performance and execution capabilities of our team. Once again, it's helped us show the market that we're here to deliver on our promises. If you look at what we said in a Capital Markets Day back in 2022 when we announced our entry into data center industry and we unveiled Phase I, and then you fast-forward to today.
You compare the figures, three metrics. The dates, we said we were gonna stabilize in 2027. That's actually what's gonna happen. Full rents from Phase I will come in 2027. If you look at the CapEx, we said this would cost us EUR 9.6 million per megawatt. That is precisely where we're gonna land. In returns, we said we're generating EUR 75 million rents. The reality is that we're gonna bring in an additional EUR 22 million, so we're gonna land in EUR 97 million rents, which is way above and helps us to completely offset the loss of rents from the sale of the BBVA portfolio. We're landing at a 15.8% yield on cost, which hints to a potential revaluation of the investment 2x.
Just to give you, to wrap up and summarize the figures from phases I and II. Phase I, this is, you know, stabilizing next year, so works are largely completed. 64 MW fully LET. Three sites, Getafe, 20 MW. This, where we are right now, Arasur 3, 22 MW. In Barcelona, an additional 22 MW. EUR 600 million of investment, and, you know, the yield on cost which we just mentioned of 15.8%. If we move to Phase II, this is where the focus is right now from the development team. Stabilizing in 2030, almost 4 x as big as our Phase I, so 254 MW, across four sites.
Madrid Tres Cantos, Getafe 2, right in the back of Getafe 1, Arasur 2 and 1, so the extension plots you can see right behind us, and the opening, the launch of our Lisbon platform with 80 MW. This is now, even though it's meant to be, you know, stabilized in 2030, this is already pre-let at 20%. Total investment short of EUR 2.8 billion, and expected to generate returns of 14.4%, so also very highly compelling returns.
We should also highlight MERLIN's balance sheet strength as the key driver of the future returns. MERLIN offers leverage in the lower range of the REIT universe, enabling us to deploy CapEx. We have solid credit metrics by both Standard & Poor's and Moody's, which render access to the debt capital markets. We have ample liquidity, providing funding flexibility. Operational excellence, compelling returns, and a disciplined approach are the keys that have allowed us to access capital markets since our IPO back in 2014. Since then, we have raised EUR 3.9 billion in equity. We have issued EUR 13.5 billion in debt, and we have established an EMTN funding program with EUR 7.5 billion in capacity. All this is what is gonna support the scalable long-term growth of the company to come.
Moving now into the definition of Phase III. As commented before, it is very simple. It drives straight from, you know, the experience achieved in Phase I and Phase II. It's a phase which is ready to go in the sense that we own the land, we have power secured, and our licensing in many cases has already started. We continue to follow the campus approach, so we are basically finishing the total construction capacity of two of our campuses, and we are opening a new one in Zaragoza. Our campuses continue to be fully interconnected through dark fiber, which allows us to obtain redundancies in case of need and operate remotely again, if needed.
The Phase III consists basically of two buildings in this campus in Bilbao, which are right across the rail tracks. You will see them from the rooftop when you visit the data center. Both buildings are going to host a capacity of up to 162 MW. In Lisbon, we are going to build three additional buildings, 3, 4, 5, which are just behind the admin building, and the first two, which are under construction as we speak, will be ready at the end of 2027. We will immediately or simultaneously start construction of 3, 4, and 5. Most importantly, if you look at Lisbon, in the top right of the picture, you will see like a little green land plot there.
This is the reserve of land for buildings number 6 and 7. We still have capacity for potential building 8, 9, and 10 in other parts of the industrial park. This is logistics turned into data centers, okay? Which is much more profitable for the company than operating plain vanilla logistics. In the case of Zaragoza, well, it looks like the Nevada desert, but it is only owing to the bad quality of the render. In Zaragoza, we are building two buildings with up to 150 MW capacity. You can see in the render that we have bought more land to be prepared for a potential extension in the future if and when we get extra power from our generation assets in the area, okay?
The three campuses are going to be high density, so they are already designed or designed from scratch to host AI workloads, okay? In reality, that means that the buildings tend to be a little smaller than other buildings we have done in the future, like this one, for example, which was prepared in case of need, was prepared for colo. You know, these things and the place in which they are located, they are mainly devoted to AI workloads. They will add capacity in the Iberian Digital Diagonal, in the one of the entry points in Lisbon and at the midpoint in Zaragoza, plus the entry point of the American cables in the north of Spain.
In total, we are going to develop 412 MW of additional capacity with a total investment in the region of EUR 4.5 billion at a cost ratio of EUR 10.9 million, which is very similar to the one we have achieved in Phase II. You have to take into account that in this phase we have bought land and power in many cases, so, you know, keeping the cost at bay is not an easy task. We are, you know, compensating the increase in cost of equipment and in some cases in construction through efficiencies we are gaining little by little with experience, mainly related to the supply chain. I mean, we are employing more and more local suppliers, and this is helping us to manage the cost.
When fully operational, this phase is going to give to the company in the region of EUR 650 million of additional revenue, and the gross yield on cost should amount to around 14.7%. Remarkable that we are not losing yield compared to Phase II. We are keeping the same range of yields, which is very, very important. It's very important because also this is predicated on average rents of EUR 1.33, okay? Which compare to the empirical rents we are getting in Phase II, which are 1.30 versus underwriting of 1.18, and the rents we are getting in Phase I, which are 1.26 versus an underwriting of 1.12, okay?
This is why in Phase I there is a big delta in profitability, but in Phase II we are closer, 1.18-1.30, and here we are projecting at 1.33, which is taking into account the two years lag compared to Phase II. I believe is a relatively modest assumption given the pace at which rents are moving in this market, and we are seeing that in every new negotiation we are undertaking as we speak. In the case of Bilbao, the total capacity will consist of the two new buildings that we just commented, 4 and 5 with 80 MW and 70 MW, and then we expect to get a repowering for building number 1, which is already being designed with a maximum capacity of 60 MW, okay.
That maximum capacity you will see is only sourced because we are waiting for final approval from the grid authority. We demanded 96 MW for that building. We were given originally only 70 MW, but we continue insisting on the other 26 MW. 26 MW will be good for 18. However, the physical constraints of the building, the design of the building will only allow us to make good 12 MW, okay? But it's, you know, it's better than a kick on the teeth because, you know, reinforcing capacity in existing buildings is always a very good business because you save a lot in terms of CapEx and equipment, okay? Regarding Lisbon, three buildings designed for 40 MW. The reason why you see 40 MW, 40 MW, 20 MW is simply because, you know, this is the maximum power we have.
If we get an extra little juice, we will go to 40 MW, 40 MW, 40 MW, okay? For the moment it's 40 MW, 40 MW, 20 MW and, you know, pending the extension that Fran will comment when discussing pipeline, okay? In this case, very importantly, the lines are being brought. I mean, the trenches are being dug by us. We are bringing the line. In the case of Arasur is part of the Solaria One deal, the one we signed in November. It's subject, of course, to a certain counterparty interaction. I mean, they will need to get their permits, et cetera, to bring the line to us, okay? You have to bear that in mind. This is development, okay? It's not Hollywood.
I mean, you need to understand that there could be delays, there could be complications, but we are working or doing our best to avoid those complications, particularly by helping our suppliers to, you know, comply with the dates we have agreed with them. In the case of Zaragoza, 80 MW and 70 MW, we have secured 227 MW of peak power, which are good for 150 IT. In this case, again, we have a counterparty, which is a generation specialist in the Aragón area called Forestalia, and they are bringing their lines to us. What is interesting about this park is that it is the first experiment that we are making towards one day becoming off-grid.
We are achieving a load factor through combination of wind, photovoltaic and batteries of 86%. We are this close to becoming one day off-grid, which is a real possibility in Spain, particularly in the areas where you can combine wind turbines with batteries, with photovoltaic. Okay. It's very interesting because then if you add a little bit of gas generation, for example, which is super expensive, but it will only represent a little part of your total generation, eventually you can become really off-grid, leaving aside, of course, the diesel generation capacity, which is intrinsic to any data center. Okay. Again, I mean, this, for us, is probably the most advanced technical project that we have been developing for many, many months now.
It's really interesting, and I believe it will be the first of many. How are we going to finance this? Pardon me for being a little vague, but it's going to be 50% debt, 52% equity, as commented in many other phases in the past. Why is that?
Because, as you know, we are obtaining revaluations in our projects in the region of 2x when we inaugurate and, you know, developing with a cost of 50/50 means normally at when it's finished, you get under 30% in terms of loan to value, which is important for us because we have a commitment with rating agencies, and for us, the cost of debt is an important raw material for the success of the company, and we need to keep our rating at BBB+ and, you know, try to hone it in the future if we can. Okay. That's important. ABB continues to be our preferred option with a soft preemption to make it as close as possible to a real rights issue. I mean, in the last one we did, we identified 92% of our shareholders.
We contacted 85, and we closed finally the ABB with 84% internal capital versus 16% external. Now, this time round, given the amount of demand we have received and the reverse inquiries, including the identification of a number of hedge funds who, you know that we are going to give prevalence only to people who is in the book, you know, makes us believe that we could be in the 100% range in terms of, you know, tapping internal capital rather than resorting to external capital in the market. For the debt side, we are thinking about mainly plain vanilla green bonds, but we might also make use of other instruments like convertible.
Convertible is an interesting alternative, but we are of the opinion we prefer to pay a higher coupon in and reduce a little bit the term of the convertible because we are very good at determining what our NAV is going to look like three years from now, but not so good at determining what our NAV is going to be seven years from now. Which means basically that in the convertible, the trick is normally the price of the option. I mean, you have the plain vanilla bond part of the convertible, then you have the call option. You know, it's very difficult to price that call option, and the only way to price it a little bit more efficiently is by shortening the time of exercise of that option.
You know, it's an alternative that we are checking, but, you know, it's not necessarily the route that we will take. Putting all three phases together. You know, they will add up to 730 MW total capacity. This is why we are calling this From Mega to Giga, because, you know, when now Fran discusses with you the pipeline, you will understand that this Phase III could be a little bit flexible. There could be exclusions, there could be inclusions in this Phase III, and as a consequence, you know, this is why we think that eventually it's going to be more like 1 GW than 730 MW .
For the moment, prudence is warranted, and we prefer to define the phase just with 412 MW . Similar returns to everything we have done in the past. Please note that in Phase I, there was a lot of area, okay. There was a big jump in rents, and all the land was ours. We didn't have to pay for land in Phase I, nor we had to pay for electricity. In Phase II and three, in many cases, we have paid either for land, for electricity, or for both. So that has a reflection in the yield on cost in the region of 100+ basis points, you know, erosion, which, you know, I believe is going to be a little bit the norm for the future.
Adding up all the capacity, what is notable here in this slide is that while in Phase II, we took four years to deploy the 254 MW, or will take four years to deploy the 254 MW. In Phase III, we are trying to shorten that. We are trying to squeeze that in only three years, okay? You might have it might ring you a bell in some of the calls that we have had in recent times that we have been talking about things that we are doing already in preparation of Phase III. For example, in Lisbon, we are already in the licensing process. In here in Bilbao, we are already in the licensing process. We are doing land preparation works in anticipation of construction.
We have been preparing a little bit, doing some preparatory works, to try to squeeze the time to market to three years because of course it is important, not for the yield on cost, but yes, if you measure in terms of IRR. Regarding CapEx management. Well, CapEx commitments up to now have amounted to around EUR 1.8 billion. 2026 is going to be another busy year with close to EUR 1 billion CapEx commitment. 2027, 2028 and 2029 will range between EUR 1.3 billion and EUR 1.5 billion of CapEx commitment each with, you know, a little tail end, EUR 850 million in year 2030. We have industrialized our construction and development approach.
We have built now a very significant team, and we are confident that we are going to be perfectly capable of you know sustaining this pace of activity. We are hardworking people, and at the end that is important. This is what really matters. I mean, when Fran says that we have been able to shorten most of the processes, the only reason why we shorten processes is because people who works 20 hours normally is 2 x as productive as people who works 10. Okay? You know, we try to have small and relatively lean teams, but pretty hardworking people. Okay? What is the effect in the company? Well, the effect in the company is you can almost calculate it back of the envelope.
I mean, it's the full execution of Phase III is going to basically triple the top line of the company. You know, similarly, it will have a similar effect in cash flow. In terms of stabilized rents, data centers will move from 6 to 65, so it will be tenfold, the weight or relative weight of data centers in the top line of the company. If you, instead of 400 MW, you move into the 600 MW, then that 65 move to 75 to 78.
This is the importance that data centers are having in the transformation of this company, which I believe is a very interesting example of self-transformation of a company from the inner part of the company, led by the teams that you are meeting here today.
In addition, Phase III described by Ismael, we want to have the opportunity to share with all of you as well the different pipeline that the company has been adding over the last years, and that, as you will see, comprises 5.1 GW of capacity. Of course, the 730 MW are included there, out of which 2.5 GW, we have relatively high visibility. If we need to label this additional pipeline, we have arrived at three categories. The first one is Phase IV, Navalmoral. It's a project that Clara will describe to you in a second. Comprises two plots of land, pretty large. First one available for 200 MW IT, and the second one of a maximum of 1.2 GW.
On that one, we are requesting construction permits, and the only caveat is that we need to get the authorization, final authorization from the grid authority. We are right now in a contest because there are several people trying to get some power on this node, and as soon as this is resolved, then we will move into construction. The rest of it is what we call short-term pipeline and long-term pipeline. As you will see in the short-term pipeline, there are certain elements that are missing there, whether this is grid authority permission, whether this is construction permits or to advance on the design. If you allow me, I will go one by one, so you see basically the status of each of the projects.
The first one is what we call Bilbao ARA 6. As you will see as well, when you have entry into this campus, there are more plots available in this area that we own, and we have requested through the distribution company and also directly to the grid authority for more power here. This is of course subject to the new planning of the grid authority for the next five years. Second one, Barcelona, building 2, 3, and 4. Those locations are assets that we control, that we own. What we are pending is as well to get more capacity in that substation where we are connected to. The third one, Madrid Tres Cantos, well-known because we have that included the first building in Phase II.
That first building comes through access to the power through distribution, through Iberdrola. The excess of capacity and excess availability we have there could move us to an additional 322 MW. All of these three, the particularity of these three projects is that Bilbao ARA Sur, the substation is very nearby. Probably when we enter through with the bus you have seen a lot of cabling arriving there. It's in that substation. Same in Barcelona. We have the substation like 50 m away of all these four assets. Same in Tres Cantos with the particularity even in Tres Cantos that the substation is within our plot of land. The expansion and the connection is pretty quick or should be pretty quick. The other three projects, what we call Madrid South.
This is the second transaction that we have closed very recently, and we are in the process of designing it and asking for the construction permit. Once it is obtained, that project will move into Phase III, as Ismael was mentioning before. We are a little bit flexible there. Lisbon 6, 7, and 8, as described before, those are two plots of land, one completely vacant right now, which is hosting the building 6 and 7. We are now as well moving forward through the construction permit, and this is able for 120 MW, 60 MW each. The second expansion is delivering us another 80 MW in the case that at the end of the leases we have with our logistics clients there, then we decide to expand that capacity.
Finally, Castilla y León is a project that we will make public probably within the year. This is another location that hasn't been disclosed now, and we are, let's say, putting all together as well to move forward. In the case of the long-term pipeline of 1.9 GW, this comprises two types of projects. One are 920 MW in Portugal that we have applied for different projects, but together giving us that capacity and the one of Valdecaballeros that you are aware where we control the land, but we are still waiting for, you know, heavy infrastructure to be completed.
If as Ismael was saying, if we need to establish the criteria of the different phases, Phase III of course matches with all the different milestones that we require to move and to make it public and to start development. Of course, the land is owned. The power is not only sourced but granted, and we are not in need of heavy infra works in order to access and to get the licensing for starting construction. This is where we are. In the future, pipeline Phase IV, pipeline short-term and long-term, as you are seeing here, there are several milestones that are still needed to be completed. As soon as we achieve those milestones, then that capacity will move into Phase III.
Now I leave Clara to go a little bit more in detail on Phase IV Navalmoral de la Mata.
Giving a bit more of color on Navalmoral de la Mata, it is our landmark, a gigawatt scale campus, ultra-sustainable, that will focus on generative AI and advanced computing. It will have the buildability to host up to 1.4 GW of power, as we will mention shortly. Out of its key attractives, we highlight the abundant, low-cost green energy of Extremadura. In fact, out of the 15 regions that form the Spanish electricity grid, seven of them are in a surplus. This means that they produce more than they consume. However, Extremadura ranks first with a 40% more surplus than the next in line. This means that Extremadura produces 6.5 x more energy than it consumes. In second place is strategic fiber connectivity.
It is located midway between Lisbon and Madrid and connected with our data center campuses in Madrid and Lisbon. Finally, but not least, its cutting-edge sustainability design. It will come with a zero water consumption design and carbon neutral. As I was mentioning, this data center campus is halfway between Madrid and Lisbon in the A5 highway. It comes with a ready-to-build urbanized land plot, and it's less than 20 km away from the Almaraz nuclear power plant. This means it will come with up to 2 GW of backup power. Additionally, it is surrounded by more than 1 GW of solar photovoltaic production. This means that we will come with cost-attractive energy costs, and we'll be able to sign PPAs for our tenants.
Giving a bit more detail of what the short-term pipeline of Navalmoral is, we have requested the construction license for the first two buildings, which will come with 200 MW IT, 100 MW per building. We requested the license in September of this year and expect to be given the license by mid of this year. As Fran was mentioning, in February 2024, we requested up to 2 GW of power of the Arañuelo substation, which is currently under an auction. We are ranked first in the power contest and expect to receive this soon. Once we start construction, we expect to have these two first buildings ready for service by the end of 2028. This means that we will invest a total amount of nearly EUR 2 billion.
Putting everything together and showing their relative location within the Iberian Digital Diagonal, you will see that, you know, our current expectation of capacity of 730 MW, it's through the campus concept, is set to increase significantly. As commented, out of the full potential of 5.1 GW, we have a pretty interesting level of visibility on the first 2.5 GW because they consist mainly in, you know, making bigger part of our existing facilities in different places. We are insisting just on the areas in which we have already built. We know the ground composition. We know. We have a construction company which is on site that can continue doing another building and another building.
Now, this is easier, much easier than moving heavy machinery from one plot to another. It's also very interesting from a project management standpoint because we can, you know, make the most of the people we send to the different places because they are deployed to those locations for a long period. They get very familiar with the situation there and, you know, they are much more efficient in the use of their time. Little by little, of course, that allows us to have a very realistic pipeline. I know you are accustomed to people discussing pipelines a little bit out of the blue. EUR 4 billion, EUR 3 billion, 30 GW, 15 GW.
This is a zooming exercise on the pipeline so that you understand that what we are trying to put together is a manufacturing facility of power land and then data centers, okay? Which one day can be commercialized, and we can obtain cash flow long-term from those. That's it. Thank you for your attention. As very brief closing remarks, outstanding growth opportunity, okay? You are helping to build what is already the Iberian leader in AI infrastructure. We will simply try to increase that leadership in the coming years because our competition for the moment is not very active. We are trying to, you know, make more room to breathe. We operate in Iberia. We are trying to build financial muscle here rather than trying to open 20 construction sites across Europe. Why are we doing this?
Because we believe this is the most interesting entry point in Europe for the development of AI data centers. We have a proven track record that has been established during Phase I and Phase II and the beginning of Phase III, which, you know, we are discussing today. We are doing this, or we are launching Phase III, very importantly, only as a response to demand. I mean, the reason why we launch Phase III is because we have clients that tell us, "I need more power in your site." Okay? So it's, it is, we wouldn't do that, and in fact, one day you will see us stopping like we did in logistics. At some point, we will stop development. We are developing more and more because we see that there is room for maneuver in the market.
You have the pioneering effect. We have been the first movers, and we need to continue, you know, making wider the gap with our competition. We do all that with very compelling financial returns, which is what matters at the end for all of you, shareholders. Okay. You know, thank you for being here today.