Hello, everyone. I am Andre Medina, IR director at Patria, and welcome to the sixth edition of our PAX Talks, today to discuss Patria's role in building the data center infrastructure in Latin America. This will be a panel discussion, so feel welcome to submit your questions. We will try to get through as many as we can. Before we start, I have to read the forward-looking statement. I would like to remind everyone that today's call may include forward-looking statements, which are uncertain, do not guarantee future performance, and undue reliance should not be placed on them. Patria assumes no obligation and does not intend to update any such forward-looking statements. Such statements are based on current management expectations and involves risks, including those discussed in the Risk Factors section of our latest Form 20-F annual report.
I also note that no statements on this call constitute an offer to sell or a solicitation of an offer to purchase an interest in any Patria fund. Okay, with that cleared, I am very happy to have with us Cristiano Souza and Rodrigo Abreu. Cristiano, who will lead the questions, is the Founder and CIO of Zeno Equity Partners, a London-based equity investor focused on concentrated long-term oriented global equity strategy. Prior to founding Zeno in 2023, Cristiano was a senior partner at Rio de Janeiro-based GNO Gestão de Recursos, where he spent 30 years. On the other side of the table, we have Rodrigo, who is Patria's operating partner for digital infrastructure sector.
An electrical engineer with a Stanford MBA, Rodrigo has held senior leadership positions in large technology companies in Brazil, LATAM, and the U.S. for over 30 years, including the CEO post at two of the largest public telecom operators in Brazil. He is an enthusiast of transforming and creating businesses, and is currently focused on developing Patria's thesis on large AI data centers in LATAM. As you can see, two heavyweight names for today's discussion. Without further ado, I would like to pass the word to Cristiano to kick off the panel. Cristiano, the floor is yours.
Thank you. Thank you, Andre, and thank you for the opportunity to have this discussion. Rodrigo, maybe we should start with a little bit of context. Can you give us an overview of Patria's experience and history in this space, in Latin America, and so where you guys are coming from in this space, and what is the current position? What exactly is the current structure that you guys have in place for the data centers?
Absolutely. Good morning, everybody. Thank you, Andre. Cristiano, a pleasure to be with you. As you mentioned, Cristiano, to give a little bit of context of what have we done in data centers and what are we doing today. As you know, Patria had a previous very successful experience in data centers, in the data center space. Curiously enough, it was launched right at the time I was at Patria as an advisor back in 2015, 2016. By that time, the Patria team identified that the cloud trends would really speed up in LATAM, and there were no dedicated cloud data centers in LATAM. The data centers were here for a while. They had been here for almost 15 years, but they were still growing in a very small scale.
On the other hand, this was happening everywhere else in the globe with the growth of cloud. Patria decided to invest in a greenfield thesis by that time, and we created ODATA. ODATA was created from scratch. It had a first project, which was a small data center outside São Paulo. We say small today, but by that time, it was a 3 MW data center, so not a good size data center for 2015. It was built in 2016. It was launched at the end of 2016, 2017, and then the company started. It was an interesting experience because the company started really as a greenfield with no previous experience in the sector, just the reputation of Patria, its contacts, obviously the management team that Patria was able to attract by that time.
The company ended up being very, very successful, which transformed into one of the largest data center platforms in Latin America. We admitted a partner along the way, so we had a partnership with CyrusOne in the middle of the way. In 2022, after a very successful run, ODATA had developed seven data centers throughout LATAM in Brazil, Mexico, Chile, and Colombia, had contracted significant capacity with hyperscalers, with off-takers. At the end of 2022, Patria successfully signed the sale of ODATA to a U.S.-based company called Aligned Data Centers. The story that I used to tell about this is interesting because at the end of 2022, when Patria sold the company or signed the selling of the company, the closing occurred in 2023, ChatGPT had not been publicly announced yet. ChatGPT was announced publicly a couple of weeks later. Literally a couple of weeks later.
We are talking about November 2022. By that time, ODATA had around 45 - 50 MW of operating capacity and had almost 140 MW of contracted capacity, contracts that were already in-house and were being developed. It was a very successful experience. What happened in 2023 after ChatGPT was launched was that the AI scenario really turned the data center capacity demand on its heads because things were already growing, but they started to grow faster enough, faster yet. During 2023, I started talking to the Patria team, and they invited me to come back to the firm to lead this new thesis of, hey, what can we do about the AI world and the large-scale data centers now? At the beginning of 2024, we started the development of our new thesis.
We worked a good part of a year in, I call stealth mode, not really saying what we were doing, but just gathering the thesis, talking to potential off-takers, potential energy partners, developing projects. By the beginning of 2025, we announced the creation of our new platform, which is OMNIA Data Centers. We launched OMNIA Data Centers by the beginning of 2025. I am the CEO of OMNIA. During 2025 or this period, we really successfully developed not only a good energy partnership but also attracted a global hyperscaler to Brazil. At the end of 2025, we announced Latin America's largest data center project, which is OMNIA's first project, which is a 200 MW data center for ByteDance in the northeast of Brazil. It is going to be the first large global data center base installed in Brazil. This is where we are right now.
We are in the middle of construction. We have a plan of expanding again to all of Latin America, looking at the usual suspects there, Mexico, Colombia, Chile, and obviously Brazil as the key focus for the company. We are replicating the story here from zero. We are, again, at the forefront of what is happening in terms of large-scale data center development in Latin America here.
Thank you. One thing that I am really curious about is when we look at data center deployment around the world, obviously it has been very centered around the U.S.
Yep.
Why is Latin America a good place to build a hyperscaler data center? What are the attributes that we have in Brazil or in other places that make it a place where, yes, of course, it makes sense to build a data center there?
Well, Cristiano, there are several factors that ended up developing over this time. You are right, the U.S. continues to be and will likely remain the key focus for data center investment. If you look at the overall global data center capacity today, hovering around 100 GW, give or take. The U.S. concentrates more than half of it, which is incredible. But in the end, several things started to happen, which actually benefits Latin America and call for the needs to decentralize a little bit global data center deployments. They were first, just the sheer magnitude of growth that is going to happen. We are talking about the most conservative projections, talk about the growing of the data center capacity in the next four or five years, multiplied by three, to get to close to 300 GW globally. It is virtually impossible to grow everything in the U.S. for several reasons.
The power required to do so, obviously, every time more is the key component. It always was the key component, but every time more is the key component of this equation because it needs to be installed at high voltage, high capacity developments where you at once deploy 300 MW, 500 MW, 1 GW project in terms of power requirements. This is not easy nor fast, especially in a country such as the U.S. where the grid is not integrated. So developing these large power projects in the U.S. is very time-consuming. They have been growing, they will continue to grow, but some projects can take five, seven years. Projects started to spread across the globe. There was a global base developed in Southeast Asia, in Malaysia. There were projects developed in Ireland. Obviously, the projects in Europe already existed. They continued to expand.
All of the hyperscalers realized that they need to diversify this geographic positioning of new capacity for data centers globally. Then Latin America entered into the map because primarily, the power requirements and power availability that we have, obviously, coupled with the other requirements, connectivity, fiber connectivity. We are very good connected to the world right now. Brazil is connected with 18 subsea cables to the rest of the world. Availability of land, geopolitical neutrality. But primarily because if you look at LATAM, we have over 200 GW of renewable energy potential development. Not only that, but at very competitive rates. If you look at Brazil today, the price per megawatt hovers around $50, give or take, all in.
If you compare this to prices in the U.S., we can talk about the cheapest prices being in the range of $70-$80, the most expensive ones in the range of $150-$180. Europe is north of $150. Southeast Asia has already gone over $120. Latin America remains a very reliable source of renewable power at very competitive prices. People started looking at LATAM as a potential base to expand. We successfully were able to convince one of the key hyperscalers, global hyperscalers that was looking to develop a global base to actually come to LATAM. We see this potential actually just starting. Our project is the first large-scale project, and we believe Latin America is going to receive much more projects like that.
Understood. The energy angle is clearly super important. A two-part question. Can you take us through what you guys think is the size of the opportunity?
You alluded to 200 GW in—
Potential of renewable energy developments in Brazil. What do you think, realistically, when we think about a 5-10 year time frame,
What's the potential in terms of gigawatts of installed capacity for data centers in Latin America, maybe more specifically in Brazil? The second part is, what would a typical project look like? Sort of time to develop, main components of the CapEx, how do you fund it? The commercial strategy, how do you think about levered and unlevered returns? If you could first talk about the size of the opportunity and then go into the details of what the economics look like, that will be super helpful.
Absolutely. You are right. First, the opportunity, you are correct. We have 200 GW of potential development, renewable developments here. It is obvious that we are not going to capture the majority of the share that is going to occur in data centers. This is most likely going to the U.S. What we used to discuss here with the sector and with the players is that if we just catch up to our proportional participation in the global economy in terms of moving up from roughly 1% of the data center capacity globally that LATAM has today to something around 5%, 6%, which should be where we could have the economic representation of the economy at large. We could certainly think about having, in LATAM, over the next three to four years, a multiplication of four to five of the current installed capacity.
This means bringing the 1 GW of data center capacity that LATAM has today and bringing it up to 5 GW, just as a bare minimum. This, by itself, in terms of infrastructure investment, represents an opportunity of roughly $40 billion-$50 billion in infrastructure alone. I am not talking about the hyperscaler investment which comes on top of that, which is significant, but just on the infrastructure alone, $40 billion-$50 billion in infrastructure opportunities. If we are successful and capture even a slightly unfair share to the best part of it in terms of growth, there is no reason why we could not think about LATAM having 10 GW in the next five years. There is no reason why not.
Obviously, we still have to do our homework to get to the fair share of getting to 4 GW or 5 GW, but that is the size of the opportunity here. In terms of a project, if you think about the typical project, there is a metric that hovers around $10 million/MW . So a typical project for AI starts in the range of 100-200 MW IT capacity, which means 1.3x-1.5x this in power capacity, through power gross capacity. We are talking about projects that normally start at the 200 MW mark, but they have a characteristic in common, which is all of the hyperscalers, when they set up a 100-200 MW facility, they want to see expansion capacity.
In our case, for instance, our first project is 200 MW, but ByteDance, when it announced the project late 2025, they said that they expect this project eventually to grow to 1 GW. This is the typical size of projects. How long does it take to develop those projects? This is a complex development. We should not underestimate the requirements because we are talking about developing the whole power licensing, which, depending on the country, can be a very lengthy process, taking nine months to a year of developing just the power licenses. We are talking about environmental permits, which are significant. We are talking about big enterprises here. Talking about, obviously, construction itself.
There are different methodologies that are evolving today in terms of construction, just a precast, a steel structure, and most recently, modular structures where you pre-integrate components in factory and then just install them in containers in a precast structure on site. We can talk about development times that range, for the full project, two to two and a half years. Construction alone, 12-18 months, 12-24 months, depending on the size of the project. The key components are virtually two. One is construction, just basic construction and just putting the buildings together, and this is traditional construction methods such as precast concrete or steel structures. It is a large construction. Just to give you an idea, the project we are building right now, we are talking about 150,000 sq m of just a construction roof capacity.
And then obviously you have the key components of technology in the CapEx, which make this not an ordinary traditional civil construction project. We are talking about diesel generators for backup, batteries, cooling systems, water systems, security, temperature, environmental control, access control, all of the very sophisticated infrastructure that goes on these projects. This roughly represents, I would say, 60% to 70% of the costs of the project. So the CapEx, the roughly $10 million/MW of CapEx go primarily to equipment and construction at the 30% to 40%, 60% to 70% proportion, give or take. How do we fund that? We are funding that based on the most traditional proportions that we see in large data center projects. This is starting to come to LATAM as well. We are being initially slightly more conservative, but those projects can actually go all the way up to 80% leverage.
So 80% debt to 20% equity. There is a proportion that starts at 65%, goes all the way up to 80%. In aggressive cases, there are even projects that go all the way to 85% debt. But we are being on the more conservative side. So talking about roughly 70/30 here in terms of equity debt. Obviously the funding strategy comes from our own infrastructure funds. So Patria Infrastructure Fund V is the key investor controlling the project on our end. We just closed a significant project syndicate with the large global banks to fund the debt part. It is true project financing. That is one difference as well that is starting to happen. We are talking about true project financing, no recourse, parent company guarantees involved, all of that is expected for large true project financing projects.
And those projects, as you may imagine, they all are born with the off-taker already contracted, or there is no sense in developing as in the past, just developing, quote-unquote, "speculative building," and where you just sell a portion of the enterprise and then go after other customers, other tenants to make up for it. So in this case, when we are talking about an investment of $2 billion, it only starts construction when it is fully contracted. We are talking about long-term contracts, PPAs involved, normally USD denominated, very, very rare not to have USD denominated in this case. And with returns that are compatible with the risk that we are taking here, normally buy and hold returns, I would say in the yields in the low teens. It is going to depend on the complexity of the project and obviously the region.
But working with the exit multiples here, very conservative exit multiples, we can see that levered returns can go all the way up to north of 20%, because if you look at just the exit multiples that we are seeing, the average exit multiples for the sector globally actually are above 20. If you look at LATAM, it hovers around this as well, maybe slightly lower, but not very far from that. So the leverage returns can go all the way up to 20+, without a question here.
Very good. Obviously, we are talking about Brazil and Latin America, and as we say in Brazil is not for the weak, right? Let us talk a little bit about risks, execution risks, power availability, the regulatory environment. There is an ongoing discussion right now about import tariffs on equipment.
If you could talk a little bit about how you think about the risk matrix and—
What are the main ones that you are concerned with, that would be great.
Yeah, we could talk about, very, very quickly, four key components. You mentioned three of them. I would add a fourth one, which is just construction and availability of good quality workforce, GCs, and just the sheer construction process, which is not minor. But in all four, in terms of construction, just starting for the last one I mentioned, we have experience working with large GCs, and LATAM as a whole has been having more experience with GCs being specialized in data center construction. So there will always be, in particular, the schedule pressure on the project. So it is not a matter of, hey, we will be able to build this project or not. You always will be able to build it. The problem is, will we be able to build it exactly in the timeline and schedule that is required?
Because this is what actually can hurt significantly the returns of a project. The penalties in contracts with the large global hyperscalers are very strict, because obviously they have to follow a global schedule of online capacity. If you delay delivery of projects just by a month, a couple of months, you can hurt your returns already. Obviously, if you delay more, then it can be very, very dramatic. The key risk here is timing. But gladly, we have been able to work with competent, capable GCs. We have good experience as well managing those large complex projects, and this is a risk that we can mitigate. On the power side, our availability per se has not been such an issue because Brazil and LATAM as a whole, Brazil in particular, have excellent power capacity. We have surplus energy at this point.
As we have an integrated grid, this is another interesting thing. If you are connected to the grid, there is no large [AI] project which is not connected to the grid, it is very, very rare, globally I am talking about, to have a project that is not connected to the grid. Power will always be available. The key components here is, do you have the right PPA and do you have a trustable and with the financial capacity partner for the PPA? Because obviously the PPA is going to have to be settled by this partner if it is not able to generate the energy required for the project. But the grid will always be there, so power availability will always be there, and this is critical. On the regulatory front, yes, there are several licenses and permits that are required.
Probably across LATAM, this is a micro topic in terms of the scale, but it is very important because it impacts directly the schedules that we discussed. Having track of all of the licensing and permits that are required for such projects, it is key. Obviously, because of the very strict timelines and pressure for delivery that we face in those projects, there is normally some risk that you incur at the beginning, where you do not have eventually 100% of the licenses and already have the commitments. But we have been able to manage that to a good extent. It would be good to see some of those processes improving in some countries. We know that, for instance, Chile is revamping their whole permit regulation right now. It was too cumbersome, very complex to obtain licenses in Chile. Brazil has done some work.
In our case, we have worked with the state government in Ceará to really speed up, streamline many of the processes which have actually benefit next projects that will occur there. This is obviously something that we have to track. For hyperscalers, the regulatory requirements, I would say, that are relatively benign. LATAM and Brazil in particular, have been known for being kind of in the middle between the U.S., as a completely regulation friendly environment, and Europe, which is more strict in terms of regulation. We are normally in the middle of the way. But we are very balanced in our application of regulation. We heard of no real issues stemming out of regulation in our case. Finally, tax and tariffs is obviously, this is the key critical point in particular for Brazil.
One of the key deterrents for large projects to come to Brazil in the past was exactly the high import tariffs, and high tax chain as a whole, not just the import tariffs, which could add a 50%- 60% surplus on top of the CapEx investments. Brazil has been addressing that in several ways. One of them is through specific regulation for data centers, tax regulation for data centers. It started two years ago. It has taken a lot longer than it could, but it is being resumed as we speak. This week we had news about it. The government expects to pass this regulation of having a special regime for data center investments called Redata by the end of the year. This is going to help tremendously to speed up the number of projects that have been kind of waiting for this regulation.
In our case, we also resorted to something that already exists in many countries and in Brazil as well, but we were the first ones to benefit from it, which are the special economic zones. In our case, we are using a free trade zone called the Special Export Processing Zone, where, as our project is targeted to a global base, not to a local base, we have a full tax exemption there because it is an export operation. This does not depend on specific regulations. It is already a law that has been deployed in Brazil several years ago and is an existing special economic zone. It is not a new special economic zone. This already exists as well in Colombia, for instance. Colombia has several free trade zones. Mexico obviously has the proximity to the U.S. and does not have this issue.
Chile has lower taxes in multiple cases to actually attract investment as well. We believe this is a risk, but some of the measures that the region is taking are providing a long-term stability. The ZPE, in our case, the free trade zone that we are using, has a 20-year timeframe for the stability of the benefits and the exemptions that are issued. Same thing in Argentina. We know that Argentina has been receiving more attention recently. They instituted RIGI and Super RIGI as special incentive programs. Colombia, same thing. There are several things going on that we believe address this issue of taxes and tariffs, and the key thing here is long-term stability, long-term visibility.
In terms of incentivizing long-term investment.
Yep.
Good to know that Redata is back on track.
Changing a little bit the scope. Obviously, everybody is aware of the deal that NVIDIA recently announced, the new funding structure, along with some of Patria's peers in the U.S.
This comes in the context of an ongoing debate about circular financing and returns. Do you have any thoughts on this and whether this funding mechanism is something that could eventually play a part in how you guys are developing the infrastructure in Brazil?
Mm-hmm. Well, this question is interesting because it prompts to another question that I believe everybody has, but nobody really likes to touch upon it, which is, are we living in a bubble, and is there a possibility of this bubble bursting? I responded to this question many times, and I like to say that, in a sense, yes, because obviously if you see the projections that go far beyond the next three, four years, they continue to be really incredible, outlandish sometimes. But what I like to say is that differently from many cases in the past, even if this bubble bursts, what's underneath it's already really, really, really big and much bigger than what we have today because capacity is in short supply.
It's different than the case of previous infrastructure investment cycles where people say, "Build it and it will come, demand will come." In our case, demand is already here and is not satisfied. You mentioned NVIDIA. NVIDIA just published the results, as you know, and the key deterrent for NVIDIA's revenue growth was supply. It was not demand. If they had produced 50% more than what they sold in the last quarter, they would have sold everything. This is because capacity is in short supply today. If we had globally another 10 GW, 20 GW, 30 GW of capacity, it would already be consumed. The problem is there is no capacity. Capacity is being built as we speak.
If you look at just the fundamentals of how capacity is going to grow, how capacity requirements demand is going to grow, it's obvious to see that there's no way other than up. Eventually, it may taper down and grow slightly slower than what we're seeing today, yes. But remember that probably today we're at the peak of growth. Last year, just the five U.S. hyperscalers alone invested $250 billion, which was an all-time record and a very healthy investment number for the global infrastructure economy. This year, they're expected to invest $750 billion. It's three times as much as last year. Do we need years of $750 billion every year to sustain this growth? No, we don't.
If we go back to just the $200 billion-$300 billion a year, which was the steady state that probably 2024, 2025 showed, the infrastructure sector would have a significant opportunity still for the next five, 10 years, and we can see this happening. We see this as something that's normally associated with the demands of the IT investments as well, not only the infrastructure investments. As a race here, there's also this component. We know that eventually this will stabilize. We have winners, we'll have different models. But the thing is that the applications are so new that honestly, I don't believe, nobody, and I mean nobody, has yet a clear view of what can be done with the technology we're developing. I believe that we haven't seen even the start of it.
If you think about, for instance, agents, that we're talking about personal agents, a single agent can occupy the training capacity required for an entire model. We're talking about multiplying the capacity that we're using to train models today by millions. Millions. Demand will be there. Obviously, it has to make sense economically. We've seen the growth of revenues, for instance, for Anthropic in all of the last 24 months, and it's nothing short of incredible. We're talking about adding $10 billion-$15 billion of revenue a month. A month. The economic, the fundamentals will be there. Eventually, they will not grow to the sky. There's a saying that say that trees don't grow to the sky, right? But they continue growing, and so there's still a cycle that is not at its end yet.
We still see a cycle that will continue for some time to come. But obviously we should not expect that this is going to grow forever or grow in the proportions that we have been seeing. Eventually it's going to start to slow down and start into a more regular growth trajectory. That's what I expect.
That makes a lot of sense. I have one final question. Obviously, as everything else, this is going to be a competitive race to—
Mm-hmm.
Get the gigawatts and to commercialize them. Why do you guys think that Patria has a right to win in this market? And who else you're competing with that you see in Latin America that is trying to do what you guys are doing?
Sure. No, that's a great question, and one thing we have to have in mind is that when we look at the type of investments that we're seeing and needing in infrastructure, digital infrastructure projects today, this has been changing. Digital infrastructure used to be a sector where there were all sorts of projects, small projects, very small projects, and local players. But when we're talking about 200 MW+ data centers, we're talking about serious commitments of equity and debt. Naturally, the number of players is already reduced because you have to have the financial backing, the financial capacity, the musculature to actually take care of that. Not only that, but you have to have the reputation, the means of actually fulfilling very complex construction contracts and development projects. Naturally, the number of players gets reduced.
Globally, what we're seeing is that the role of infrastructure funds and large private equity funds have is virtually 100%. There's almost not a single project where you don't have infrastructure development funds involved. In our case, we're obviously the largest development fund in LATAM. We obviously have a good experience in developing greenfield investments. We have the financial backing, the financial expertise, and the track record in this segment. In addition to that, I will mention that Patria has, I believe, some of the core competencies to develop those kinds of projects, which start in the financial capabilities, obviously, but then it quickly moves towards the engineering space where we do have a hands-on approach. Patria has its own team of engineers, of technical experts, and that's what we used, for instance, to start up OMNIA.
We brought the Patria engineering team to help, and to work alongside us in making sure we were developing the right projects with the GCs and with the suppliers, and having a very hands-on technical approach. We also have a power expertise, which is really required for those large projects. We have the capability of developing power partnerships. We do have our own power operations, we do have power trading, so we understand power and have the power skills inside, internally. Finally, Patria also has a very developed real estate capability across LATAM. If you think about data centers, large-scale data centers, we're talking about a combination of power, real estate, construction and project development, and financial capabilities. This is exactly the sweet spot of where Patria has in terms of supporting capabilities, founding capabilities here, and we're using that to the full extent.
Obviously, we're competing with global funds. Normally, the large players in the space, and we're talking about a handful of players here, we're not talking about 20 players. We're talking about five, six different players, large-scale players. They all have financial backing from global funds, and we believe that Patria is now a global fund, and that we have a right to play.
Got it. Yeah. It's so interesting that we finally found a way to export power in Latin America.
Correct.
And in Brazil. We've always had this—
Correct.
Blessing of having more power than what we consume, but we could never do anything with it. Now, we're exporting it through bits. It's super interesting. But thank you for
That's absolutely right. Just as a complement to your point, initially everybody thought that would happen with green hydrogen.
Yeah.
But green hydrogen, it's taken a long time, so everything migrates to data centers. Yes, we're exporting power. You're absolutely correct.
Yeah. Great. Look, thank you so much for taking the time. That is all the questions I have. Andre, I will pass it back to you. Rodrigo, thank you so much.
No, thank you, Cristiano. A pleasure discussing this with you. It is a fascinating topic.
It is fascinating.
You are going to hear a lot about it.
Yeah.
Thank you guys very much. Very valuable discussion. We do have some questions here from the audience, Rodrigo. The first one is a little bit more technical. The question is how to handle energy intermittency given the need to use renewable sources when setting up a data center of this scale.
Yeah, absolutely. That's a very good question. We have to remember something that I mentioned during the talk, which is the data center is connected to the grid. It's not connected directly to the renewable source. It's not a behind-the-meter project where you just develop a generation project and connect it to the data center. The data center is connected to the grid, and the generation is also connected to the grid. So for the purpose of supplying energy to the project, we don't have intermittency because we're connected to the grid, so the energy's there, and as we like to joke, the electrons don't have color. They come from anywhere. They come from the wind source, they come from the hydro source, they come from the solar source, and if the grid has the capacity to supply it, we're going to receive this with no intermittency.
As far as the generator and the PPAs associated with our project, there is one thing that, yes, the generators are doing it. In our case, the first project is developed with a partner called Casa dos Ventos. What they did is to supply an average power of around 300 MW, which is the PPA size of our contract. They have developed in our building close to 700 MW of wind plants. As you have a lot larger capacity on the wind plants, and many of them are in different places, they compensate and are able, even with some intermittency, to generate the required average loads for the power, for the PPAs. In our case, we don't depend on that. Even if they didn't produce a single watt, we would be connected to the grid, and we will receive the power with no intermittency.
Excellent. Very clear. The other question is, we've seen around the world different governments partner with managers to build and scale data centers in their domestic AI development platforms. Are you seeing governments in the region start to take these steps, and how is Patria positioned to benefit from it?
We have seen that. It is just starting, Andre. We are not there yet. We know that there are some sovereign AI discussions happening pretty much everywhere. Some governments took the step of eventually creating their own model developments and their own data center capabilities with partners, but it is just starting. It is not that common yet. In the case of Brazil, for instance, Redata brought some considerations of having to have a capacity dedicated to local developments and to sovereign AI, but it was as part of a larger incentive. They want us to have part of this capacity, quote-unquote, "reserved for local developments." But it is starting. It is not there yet.
It is going to be a discussion, remembering that, in the case of these partnerships, obviously it helps to have a local team on the ground with local experience, with local presence, with local participation, with local institutional relations, and understanding their requirements to do things such as this, which require, for instance, public bids, participation in public bids. Patria has a longstanding history and track record of participating in public bids. That is what we do for most of our concessions, for instance. When it comes, if it comes, we will have a right to play because of that.
Excellent. The other question here, I will read it. I think it is read hyperscale, and it is just saying, I have seen reports that Amazon, Google, and Microsoft are looking to develop in data centers and related infrastructure in the region. Is Patria in discussions with these other larger global hyperscalers, and how is Patria positioned to be their partner of choice?
Mm-hmm. We are discussing with all hyperscalers because, in our sector, if you think about the large AI data centers, even though the number is slightly larger than this because we can include some other companies, which we call neoclouds, which are new players that demand a high capacity as well. If you think about the traditional big tech hyperscalers, we are talking about eight players globally, and that everybody know their names, and we are talking about five U.S. players, three Chinese players. A platform that focuses on AI data center and large-scale data center development has to talk with all of them. So we are talking to all of them.
The good thing that we believe in our case happened is that as we successfully were able to implement a very large first scale project, which, upon launch, will be the largest data center project in Latin America ever. This obviously qualifies us to have live discussions and very engaging discussions with pretty much all hyperscalers looking towards developments in the northeast, where we are locating our first campus, but also talking about other countries as well, where we have presence as Patria in terms of real estate and energy. Yes, we are talking to them. It is a competitive field. Obviously, we do have our expertise, and we do have our competitive advantages, in particular by having developed this first very large project in a zone which nobody had explored before. We are ahead with our knowledge of the special economic zones.
But as the number of projects is so large, eventually, we believe that there is space for, as I mentioned again, 4, 5, 6 different players actually sharing this growth that is going to come. Let us remember what we talked at the beginning of the conversation. We see the possibility of LATAM receiving another 4- 5 GW in the next three years. It is impossible for a single player to address all 4- 5 GW. We are going to have to have a number of players working here to deploy this type of capacity.
Great. I think the last question here, and you pretty much touched base on it already, but with regard to the energy sources, standing up renewable, very strongly in the region, how much that is playing as a competitive advantage on getting us with talks with hyperscalers, which would not consider the region or which would be more interest in the region because of the use of renewable energy?
Yeah. No, you are right. Energy is the key differential. I would say that there are two points to the energy. One is the competitive pricing, and we mentioned that, we have very competitive pricing in Brazil and competitive pricing in general in Latin America. That by itself should be one of the deciding factors in bringing projects here. Because if you include energy plus the infrastructure leasing as the key components of operating costs for a hyperscaler in a data center, we are talking about energy being responsible for between 30% and 40% of expenses. It is big. It is a large cost. But on the other side, there are the commitments to carbon neutrality, and this is something that all of the hyperscalers have done.
It's facing challenges because as capacity is growing so much, and in particular in the U.S., it's hard to keep track of carbon neutrality when you're deploying gas power plants or even coal power plants. That's why when they have the capability of developing projects with renewable energy, with no need for carbon compensation, this is attractive. This is really attractive even in the economic sense, because here when you deploy a project with full wind power or full hydropower, you don't need to buy carbon credits. This ends up getting factored into the equations as well. We do see that this is happening every time more. We deal with some hyperscalers that when they look towards projects here, they say, "I will only deploy a project in Latin America if it's 100% renewable." That's the key motivation as well.
It's pricing, but also the commitments to carbon neutrality, which can be helped by those kinds of projects here.
Perfect. We have one last question here, Rodrigo, and I think you already discussed it, but maybe we can double-click on it. It's with regards to the additional opportunities that this will open for Patria. It says that many firms have developed multiple products to service the data center demand, from development to compute financing to power financing to buying and investing in stabilized assets. What is Patria's strategy here, and what's the current plan strategies do you have to take advantage of this?
Mm-hmm. Well, we're at the very beginning of our journey. As we mentioned, Patria launched OMNIA back in 2025, so we're developing our first project. We already are looking towards developing other projects in Brazil and in the other countries. But what we see-- First, we see the possibility of expansion, pure and simple, and we're talking about other avenues that include the connectivity, that include ancillary services that normally will come in those projects. As I mentioned, Patria has expertise and presence in the power sector and the real estate sector, and those are two adjacent sectors that without a question will be significantly impacted by data centers. That's what we're trying to do as well. We have internal discussions with the real estate team, with the power teams to see what can we develop together, and it will always be an arm's length discussion.
We're going to do whatever is good for the project itself and for the funds. We open up opportunities for power, we open up opportunities for real estate, and in some cases, we open up opportunities for credit and for core funds. Remember that we're starting to work with core funds as well, where we pretty much buy stabilized, more mature assets for a different profile of risk return. Eventually we're going to have those kinds of assets as well in the data center space. Normally, an asset from construction to stabilization takes around 2.5 years to be developed. After that, there's multiple examples across the globe of projects from stabilized assets moving to continuation funds or moving to core funds, and it shouldn't be different here in the long term as well.
Further questions from the audience. Cristiano, Rodrigo, thank you very much. Very interesting discussion. Thank you for your time.
No, thank you, Andre. Thank you, Cristiano, here for your questions. Very to the point.
Thank you very much, everybody.
All right. If you have any additional questions, just send us through my email, and we'll try to answer you. Thank you.