Please welcome Chief Executive Officer, Brookfield Infrastructure Partners, Sam Pollock.
Well, good afternoon, everyone. It is great to see so many familiar faces and good friends. As a son of a Sam, was I the only one feeling uncomfortable with Howard's story there? Bruce was looking over at me. Last year, we said that Brookfield Infrastructure was approaching an inflection point in its growth trajectory. Today, I want to pick up where we left off and show you why we believe that it is underway. To do that, I am going to briefly revisit the strategy that we have, walk through what we have accomplished this year, and then I will finish with what we think the opportunity ahead is, and why we think it is as strong as ever been.
To start off with, we have got a slide here that I think most of you have seen this before, and that is pretty intentional. Our strategy has not changed, and there are three parts to it. We maintain a strong financial position, deploy capital at attractive risk-adjusted returns, and we crystallize value through capital recycling. When we execute those three things well, the result should be 10% or better FFO growth per unit over time. The full-cycle approach has been the foundation of the business for almost two decades now, and it remains the framework for how we allocate capital today.
As I just mentioned a second ago, a year ago, we told you we believed the growth rate was at that inflection point, and we had conviction because several things were coming together at once. Our organic backlog had grown. Recent investments were beginning to contribute to our business. Capital recycling had scaled, and some of the headwinds that were affecting our per unit growth were starting to minimize. This year, rather than repeat that statement, I want to show you some of the evidence. I think this chart tells the story pretty clearly.
From 2023 through 2025, average annual FFO per unit growth was about 7%, which is below our long-term experience. This year, we expect that to be approximately 10%. The important point is the direction of travel. The base business is performing well, capital from our backlog is coming into earnings, and recent investments are contributing more meaningfully. Since inception, we have compounded FFO per unit at roughly 14%. We are not suggesting that one year will make a trend or get us right back up there, but we believe the business is now moving back toward the growth profile that the investors in this room have come to expect from us.
Let me now walk you through the three elements of our strategy and where we stand against each of them. The key point is that the improvement in growth is not coming at the expense of financial discipline. We are maintaining a strong balance sheet, investing at attractive returns, and continuing to monetize mature businesses. Let us start with our financial position. We have generated 10% FFO per unit growth year to date while maintaining approximately $2.6 billion of corporate liquidity, BBB+ investment-grade credit ratings, and a 65% payout ratio. Those numbers matter in combination.
We are growing while retaining financial flexibility. That means that we can be patient when markets are expensive and move quickly when volatility creates opportunity, and that's been a hallmark of our success in the past. The second part of our strategy is deployment. We've secured approximately $1.4 billion of growth investments and expect to deploy more than two. The deployment isn't dependent on one large acquisition. The deployment is diversified across the investments we made into our semiconductor facility, the growth backlog that we have within our other businesses, and new investments made through various industrial partnerships.
The mix is increasingly representative of our opportunity set. It involves more organic investment, more strategic partnerships, and more opportunities sourced through the Brookfield ecosystem rather than just competing in broad auctions. The more important point about this year's deployment is the return profile. We expect average returns of 15% or higher on the capital that we're currently deploying, which is above our historic target range of about 12%-15%. There are a couple of reasons for that.
First is the infrastructure super cycle is creating a very large opportunity set for us to invest in. Second, AI infrastructure is adding another capital-intensive growth avenue for us to pursue. Finally, our scale and competitive moat is helping us source opportunities where we can achieve better economics. What I'd emphasize is that we're not targeting higher returns by taking greater risk. We remain focused on high-quality counterparties, strong contractual protections, and disciplined underwriting. The other side of the equation, obviously, is capital recycling.
So far this year, we've secured approximately $1.4 billion of asset sale proceeds, and that excludes the approximately $1.2 billion of proceeds that we were able to obtain from the Csquare IPO, which was used to repay debt. We remain well on track to exceed our $2 billion target for capital recycling for the year. The returns shown here from the mid-teens to over 40% demonstrate the value that we've created across various businesses and geographies throughout the year. This is how the full cycle strategy funds itself.
We take capital out of mature investments where we've executed the value creation plan and redeploy it into new opportunities at attractive returns. That spread is an important source of long-term per unit growth. I want to take a second to discuss the simplification that we announced in July. You may recall that we listed BIP in 2008 as a partnership and created BIPC in 2020. We are now proposing to combine these two into a single publicly traded corporation called BIP Inc. Nothing changes about the underlying business or the investment strategy.
What changes is just the wrapper. We expect one corporate security to be easier to own, it will broaden the potential investor base to those who currently can own a partnership, and it'll materially improve consolidated trade and liquidity. The security holder vote is scheduled for October 14th and subject to approvals, we expect to close the transaction in the fourth quarter. The result should be a roughly $30 billion company and more than twice the trade and liquidity. This brings us back to the inflection. Our confidence doesn't come from any single transaction or results from in a quarter.
It comes from a combination of having a strong base business, a larger organic backlog, and more capital going to work at attractive returns. I am going to finish up on this slide because it demonstrates or at least shows why we are optimistic on the deployment front, and it sets out the focus for the remainder of the presentation. We have three substantial avenues for capital deployment that we are going to talk about: emanating corporate partnerships, AI infrastructure, and organic growth within our businesses.
The point we hope you take away at the end of the presentation is that our strategic advantage is the optionality that we have in the business to allocate capital to whatever opportunities offers the best risk-adjusted returns at that time. With that, I am going to turn it over to Scott, and he is going to come up and talk about emanating corporate partnerships.
Please welcome President, Brookfield Infrastructure Partners, Scott Peak.
Thank you. I do not know if anyone else was keeping track of the number of times scale was mentioned today. Every session has referenced it, either directly or indirectly. In fact, Bruce's start at the beginning when he was asked what gets him most excited about coming to work, he referred to the assets, the relationships, the dollars, the people that we have here, which is very different from where he started his career. That is all scale. With that is BIP's primary competitive advantage, and it is something that has been built patiently and deliberately business by business over the decades.
I will start with three main messages. First is that scale is no longer a nice to have. It is not even simply just important. It is a prerequisite for market relevance, for accessing the best opportunities and the best returns. Second, we have differentiated access to the world's largest companies who originate most of the coveted and high-quality market transactions we pursue. Third, the infrastructure landscape is vast, and it continues to expand, so we are able to remain highly selective in where we deploy our capital.
I would estimate about a third to a half of the infrastructure we invest in today was nascent or did not even exist as an investable asset class a decade ago. With that context, it may make more sense that the current forecast for global infrastructure requirements has doubled over the last 10 years to more than $100 trillion today. While traditional infrastructure needs have grown, there have been three developments over this period that I would like to mention. The first is that there has been advancements in technology. You heard the OpenAI panel.
That is not going to be a surprise to you. Think fiberization, think automation, think AI factories. These weren't prevalent 10 years ago, but each is highly relevant today and will be in the future. The second is previously less understood subsectors are maturing with large capital needs. Think residential infrastructure. Last, asset classes previously outside the perimeter of infrastructure, ones that didn't meet the rigorous definitions that we hold them to, have been adapting their revenue models, have been adapting their contracts to fit within the perimeter of infrastructure.
For example, semiconductor foundries, like what we did with Intel. All this means we have a much broader opportunity set to choose from. However, to capitalize on these opportunities, a few interconnected elements are required. Together, these create a very strong moat for BIP's long-term deployment visibility, and that's very valuable and difficult to replicate. The first is the most quantifiable. You need a big business, big dollars, lots of people and assets, and a long track record. The second element is more subjective.
You need operating expertise. This can be demonstrated through key performance indicators, a large and experienced team, and results over a long measurement period. Third, which can only be earned if you have the first two right, are trusted corporate relationships with the leading market participants. Let's remember, the leading corporations originate the majority of the large and attractive infrastructure opportunities in the market. Origination from government is infrequent.
Because infrastructure assets are critical by definition, the same corporates rely on them both before and after a transaction. That is why, as we move to the middle column on this slide, corporate JVs, corporate partnerships, corporate carve-outs, they're all the most common. These corporate sellers require a counterparty that can operate the businesses critical to them safely and reliably. While price is important to them, it's rarely their primary consideration. The more complex a transaction is, an area where you've seen us excel, the greater the need for every prerequisite to be satisfied, and BIP ticks all these boxes.
Consequently, a selection from our partnership resume includes some of the leading companies around the world. The structures were varied, carve-outs, sale leasebacks, JVs, strategic frameworks, but each required a partner they knew well that brought a big business to bear with relevant operating expertise. Most of these were bilateral, beginning years earlier with a relationship. This snapshot captures the breadth of BIP's infrastructure franchise today, specifically the resources and capabilities that we bring.
$200 billion of assets under management, 50 portfolio companies in 15 countries, a track record of over 100 closed transactions representing over $ 75 billion of equity deployed, with an incremental $ 20 billion of equity projects currently underway. These credentials more than satisfy the requests of corporate boards, of regulators, and other key stakeholders. It's important to flag that the corporate leaders of these companies are the ones who ultimately drive counterparty selection, and what they really care about lies beneath the surface of this slide.
Corporate leaders regularly engage with our 50 portfolio companies across the globe. They've seen firsthand that BIP is an experienced operator and deeply understands their business. In each sector where we operate, our industry scale rivals the largest strategics. We're not a financial buyer standing outside these industries. We're one of the larger operators inside of them. We operate critical and high-profile data centers, rail lines, and logistics assets. This translates to deep credibility with corporate leadership.
Let's connect what we've talked about. Let's connect our scale, our operating expertise, and our relationships, and connect how that ties to what Sam talked about in terms of our FFO per unit growth plan. It starts with our ability to access and selectively pursue larger acquisitions and do more with these businesses once we own them. As the size of our acquisitions has increased over time, our average returns on invested capital have too. We're deploying more capital while generating more FFO per unit of capital.
Two reasons why this is occurring. The first is that larger businesses often have fewer credible buyers, which can translate to a better entry value point for us. They're typically also accompanied by more levers and more avenues for growth for us to create value very quickly during our ownership. Put simply, we're able to buy better, and we're able to create more value with what we buy. Another perspective to share is that bigger deals are not simply a larger version of a small deal. We're not buying a single big thing at the end of the day.
These are dynamic businesses with different segments in different regions and different assets included at different stages of maturity. Ultimately, a large business is typically a collection of smaller businesses that have been assembled and accumulated over time, and their integration over this time period is rarely perfect. Accordingly, these big businesses are often rife with duplicate systems burdened by legacy cost structures and constrained by bureaucracy, all serving as ideal opportunities for us to quickly add value.
Our toolkit's working. We start by identifying and supporting the right leadership team. Then we improve margins and contracting. We impose capital allocation discipline. We optimize the capital structure. We support the business through incremental CapEx at accretive returns, and we ultimately ensure long-term optionality for our exit of the business. We're seeing our toolkit translate to enhanced returns. We don't simply acquire businesses, we actively manage them. Triton, Hotwire, Colonial, all very different, show the playbook in action.
During the first 12- 24 months we've now owned them, we've completed bolt-on acquisitions, materially improved margins, monetized stabilized assets. None of this we paid for at acquisition. We underwrite these businesses to a mid-teen return, but through active management, we're already adding roughly 200- 400 basis points, lifting returns into the high teens. As larger transactions become more common for us, this same value creation playbook can be repeated with greater frequency and greater materiality.
The leading businesses within the infrastructure market have grown, so it's natural that the enterprise value of the transactions BIP has done over the years has grown as well. From 2015- 2020, many of our market-leading transactions had enterprise values in the low single-digit billions. Since 2021, the size has shifted upward, with our focus on larger businesses and larger partnerships. A conclusion here should not be that we're only doing big deals. An excellent mid-scale transaction won't get past us, I assure you.
We compete and perform best at the upper end of the scale, where we tend to achieve the best returns. This trend can also be seen by the average amount of equity we deploy per deal and the average returns on that deployment. From 2015- 2020, BIP deployed an average of $300 million per deal and achieved returns between 12% and 15%. Since 2021, however, this number doubled to $600 million, while average target returns now exceed 15%. Four key takeaways. Our opportunity set is vast and growing, allowing us to remain highly selective.
BIP is a trusted partner and counterparty to the best companies with access to the most attractive transactions. Our established operating toolkit creates significant value beyond what we underwrite at entry, particularly evident in our larger business acquisitions. These advantages, which all ultimately stem from scale, facilitate more capital deployed at higher target returns, supporting higher FFO per unit growth. Next up for you, we have a panel. It is titled Building the Backbone of AI, a Clear Example of BIP Scale in Action. I will now hand the stage back to Sam, who will serve as moderator. Thank you.
Please welcome our panel, moderated by Sam Pollock, with panelists Lief Williams and Sikander Rashid.
All right. Well, it is my pleasure to be on stage here with two of my colleagues who are the leaders of our AI strategy. Sikander Rashid is the global head. Lief Williams is one of our Senior Managing Directors here in Toronto, who has been leading a lot of the AI factory initiatives that we are going to talk a bit about and that he spoke about last year. The whole AI sector is something I guess we have been talking about, I think, for about two years now. When we first threw up the $7 trillion addressable market for AI infrastructure, we all were a little unsure whether or not that was a little high.
What we have clearly seen in the last couple of years, but especially this year, is that that number, if anything, is understated. The amount of capital, the opportunity set for us continues to grow at an exponential rate. What I thought we would do to start off with is maybe direct a question to both of you about your experience this past year with what you see changing, what you sort of see things going to in the next year or so. Maybe, Sikander, I will start with you. Maybe give your thoughts on what you have seen in the past year.
Yeah. Well, thank you, Sam. Thanks for having us, both Lief and I, and it is great to be here and see so many familiar faces. Look, 12 months in AI is a long time. A lot has happened. I would say 12 months ago, investors had three big questions. Will the AI models continue to get better? It is a topic Nate touched on earlier. Number two, do we need all of this infrastructure? Do we need all of these GPUs, the power, and the compute? Third, will AI actually generate revenue?
Fast-forward to today, I would say what is really exciting is the market has addressed all three of those questions. Number one, last year, you guys saw 20 GW of capacity. This year, we will beat that. NVIDIA, who is at the forefront of this, is recording $90 billion per quarter in revenues, up 120% since last year. Yes, we need this infrastructure. The demand is there. AI models continue to get better. We have gone from chatbots to agents, hopefully soon to physical AI. The models continue to get better, which is great for the technology and the adoption of it.
Lastly, yes, AI is generating revenue. The two frontier labs, OpenAI and Anthropic, 12 months ago had $5 billion-$7 billion in annualized revenues. Today, the number is $110 billion. Clearly AI has got revenue attached to it. That is all really exciting for AI infrastructures.
It's starting to prove out. Lief, what about you?
Yeah. One other thing I would add, Sam, is, and it's become more prevalent every month of the last year, has just been community engagement and just prevailing local sentiment on data centers. I would say 18 months ago, there was a perception in the race that the most important thing was just to get massive amounts of power at a site at any cost. Now it's probably no surprise that that's resulted in some backlash. We've now seen four states and over 300 counties and counting, who have implemented data center moratoriums or otherwise delayed permits to get these shovels in the ground and get these sites operational.
I think in practice, again, these AI factories, if you do them correctly, they do bring massive benefits, significant investments, large amount of taxable basis, employment, both during the construction period and when operational. But you need to articulate these benefits and you need to effectively ensure that you're doing it the right way. This really means bringing your own power, protecting the rate payer, and articulating the benefits that come from these projects.
Okay, I want to come back to a bunch of points there, but first I want to address one of the comments that Sikander was talking about, just the amount of dollars that we're seeing and just the rush into the sector. We're obviously not alone in having capital, and we need to differentiate ourselves. Obviously, we have a lot of capital, and we can bring that to bear, but what are the things we're doing today to differentiate ourselves in the businesses that we're establishing and buying?
Yeah, look, Sam, I think there's a couple of things. Scale obviously matters. Coming back to, maybe I'll start with the AI factory initiative, which Lief touched on. The way I describe it is whilst the technology firms are building the brain, we're building the body, and the components of this body are power, data center, real estate, and compute. Five years ago, these big technology firms or the sovereign governments could have worked with multiple vendors and pooled it all together for their cloud computing services.
Today, it's a race not just between the countries, whether it's U.S. and China, but also among all the technology firms. Their preference now is to work with vendors who have the ability to package it all together. The other element of this is 1 GW or 1 MW of power cost or data center cost $10 million. With the GPUs five years ago, today it's $50 million. The complexity and the size both have increased, and therefore our competitive advantage as a firm is we're one of the largest energy investors. With NVIDIA, we've also got compute capabilities.
We have the ability to offer this integrated computing services to some of the largest clients in the world.
Okay, so we've made an investment in Radiant. How did we do it, and why are we going to be able to make money and not blow our brains?
Yeah, so this was a question you asked us 18 months ago, and I'll give you the same answer. I think there's three things. Why are we so excited about this? Why we did this? First, AI infrastructure will be a $10 trillion, I know Sam said $7 trillion, we've already increased the number to $10 trillion. It'll be a $10 trillion CapEx spend. Half of that spend will be on compute or the kit that goes inside of a data center. So big dollars. Secondly, every company in every country is going to have its own AI, and that AI will be bespoke, customized, sovereign, secure, catering to the needs of that country or the company.
Thirdly, the chip industrialization cycle is five years long, so it takes time to manufacture new chips, design them, and build them. For all those reasons, we identified this as a sector where we could invest a lot of capital in partnership with NVIDIA, and that is why we formulated Radiant. It is effectively a compute platform that provides GPU-as-a-service, which includes data centers and the compute to some of the largest technology firms and sovereign governments in the world.
The focus is take-or-pay contracts, IG customers, and focusing on a return on and off of capital in the contracts to minimize technology risk, and we are really excited. The momentum is great, and it is really exciting.
Okay. Lief, you were here last year talking about AI factories. Sikander just kind of described what they are consisted of, both the building as well as the compute. Maybe describe for us how the cost in each of those facilities differs a little bit in various markets. Maybe we can touch on the two deals that we are currently working on that we have announced, one in the U.S. and one here in Canada.
Yeah. Look, to recap, an AI factory, this is a large-scale campus. It is purpose-built for AI workloads, and it is integrated. That means that you are finding a solution to bring power to the site, whether it is grid-connected or whether it is behind the meter. You are building the data center and all of the infrastructure that goes inside that, so generators, batteries, mechanical plant. The compute, in some cases, you would also be bringing the compute and offering that as a service to your end customers.
We think that this is an important infrastructure asset for a couple of reasons. Number one, it produces attractive risk-adjusted returns. I would say to bring a gigawatt of load, just as an example unit, that is a very difficult thing, and there is massive scarcity value in that asset. As a result, you can earn attractive risk-adjusted returns of over 10% unlevered on a yield on cost basis. Secondly, as Siko described, these are very capital intensive, and so $50 billion-$60 billion per AI factory.
Thirdly, these have all the infrastructure characteristics that we look for, so investment-grade counterparties, long-term contracts, inflation protections. Maybe to talk a little bit about two of the AI factories that we have announced. If you thought about where should an AI factory be built, and yet you were working from a blank sheet of paper, these two projects are exactly what you would come up with. We have got one site in Paducah, Kentucky, and one site in Keephills, in Alberta. In both cases, like I said, they are exactly what you would look for.
Massive campuses, really well interconnected to the grid, both the electric grid and the gas grid. We've got great partners who are working with us on this. They're just opportunities to bring scale, or can be expanded over the long term, and we think are kind of examples of doing this in the right way. Just a couple of anecdotes about each of them. The Paducah campus, this is a former Department of Energy campus. It served 3 GW of load at its peak. I've not seen any other site anywhere in North America that had that much load in the past.
It's just really well connected from that perspective. The Keephills site, this is located really at the heart of Alberta's electric transmission system. The whole grid was built around the power plants in this area. Typically, for an AI factory, you look for one or two large high-voltage transmission lines. Our site at Keephills has seven high-voltage transmission lines connecting into our interconnection point. So really well-situated to protect ratepayers and really well-situated for our customers.
Okay.
The only comment I'd add to that is, to what Lief described really well, is both of these opportunities, unlike traditional data center investments, we've leveraged relationships with NextEra Energy or the Department of Energy or TransAlta, one of our companies, to create opportunities where we've minimized speculative capital outlay, which would be another differentiator of how we're doing things compared to everyone else.
Okay. Well, maybe just tying on two things. The site isn't as well-endowed as Keephills is with interconnections. So for that, we've identified Bloom Energy as a great partner, and someone we've built a great relationship with to provide behind-the-meter power solutions for many of these facilities. Siko, maybe just talk about that relationship and how it works and why has it been so powerful.
Yeah, absolutely. Big picture, the U.S. alone will need 100 GW of power in the next 10 years for AI infrastructure alone. That's obviously a big number. The utilities throughout the country can only support 30 GW of that. We identified a 7 GW gap in power availability in the U.S. and looked at a host of solutions. We picked Bloom and approached them with this idea of becoming their capital partner because time to market in this market environment is everything. Time to tokens is revenue, as you would hear from some of the frontier labs.
Bloom has the fuel cells, by definition, don't burn gas. They consume gas. By virtue of that process, the permitting timelines for this technology are relatively short. It's been a phenomenal partnership. We're the capital partner. We invest in investment-grade customers, price the contracts such that we get a return on and off of our capital over the contract term, which is normally 10- 15 years. We started with $5 billion. We've upsized the partnership to $25 billion, and that's fantastic.
That's great. We have a little time left. In true Brookfield fashion, we always talk about the downside. What I wanted us to cover to end off was just, and you both have worked with us in our traditional infrastructure business for many years. What's different about these assets versus what you've historically been involved with, and how are we mitigating risk? Maybe we'll start with Lief, and then Siko, you can finish off.
Yeah. What I would say, I feel in a lot of ways it's returning to infrastructure's roots as concession assets. We're underwriting a DCF that effectively has a stream of cash flows, and we're assuming very little or zero terminal value in a lot of cases. Again, I think it's really the underwriting approach where we're taking a very infrastructure-centric focus.
Okay. As much as I would love to tell you we've used AI to come up with a new underwriting formula, we haven't, unfortunately. It's still the same formula we've learned from you over the years, Sam, and it's basically focusing on take-or-pay contracts, investment-grade counterparties, and more importantly, I think this is very important for AI infrastructure, pricing the contracts, whether it's power, data centers, or compute, or silicon, pricing them such that we target to get a return on and off of capital over the initial contract term and ensuring the term is commensurate with the useful life of the asset is very, very critical and that's our focus.
Another thing I would add is counterparties and documentation are critical. So our attention to detail has to be razor sharp to ensure that we're not entering any agreements that people can somehow get out of, because these are long-term agreements that we're entering into. So that concludes what we want to talk about AI infrastructure. Next up, we have David Krant, who's going to talk about organic growth and tie together all the different deployment opportunities that we've been talking about.
Please welcome Chief Financial Officer, Brookfield Infrastructure Partners, David Krant.
All right. Thank you, Sam, and good afternoon, everyone. I'm going to spend the next few minutes on one straightforward idea, and that is that BIP has more ways to grow today, and that gives us more ability to choose where we put our capital. The result is a broader opportunity set, which will allow us to invest capital at better risk-adjusted returns and further drive FFO per unit growth. Let me start by laying out the three channels available to us. As you saw earlier, the first is traditional M&A.
This is buying businesses that we've done for the last 17 years, utilities, transportation, midstream, digital infrastructure. It's driven our growth for many years, and it will for years to come. The second you just heard of is AI infrastructure, and the third is organic growth capital. This is investing inside the 50 portfolio companies we already own today. Now, each of these channels differ in terms of their timing, the risks, and the return profiles, and that's exactly the point. We do not need all channels to be open at the same time in order for us to grow.
Rather, we can choose where we invest that capital to earn the best risk-adjusted returns. Let's go through each of these, starting with the one that's done the majority of our growth to date, and that's traditional M&A. Across the period shown, we have averaged about $2.2 billion of investment into growth initiatives. If we focus in on where that's come from, while almost 80% of it, the size of these investments have trended larger, and as a result, have provided us with very strong risk-adjusted returns. Traditional M&A will remain a very important part of what we do.
But as you heard, the opportunity set is now more broad, and that's most notable in AI infrastructure. As you've heard, AI infrastructure is not just one single asset class, rather a set of connected opportunities across three channels so far. Today it's been power and transmission, AI factories, and compute infrastructure. Each of these opportunities are similar in that they are large scale and have highly contracted revenue profiles, which is something we love. But what may differ is the deployment cycle or the time it converts to earnings.
As we look out over the next five years, which is the time it may take to fully contribute, we expect to invest upwards of $2.5 billion at our share over that five-year period. That's simple math. That's about $500 million per year that we expect to deploy into this new growth channel. Because these opportunities are large, we will invest alongside partners, allowing BIP to concentrate its investment into those highest returning projects. Finally, the third channel, which I'll spend a little bit more time on today, is our organic growth.
Looking at our deployment this year, as Sam highlighted earlier, you can see we've deployed nearly $1.5 billion into growth. The majority of that has come from internally sourced organic growth projects. If you add on top of that the funding of our Intel joint venture in Arizona, we've deployed nearly $1 billion into assets we own today, building high-quality new businesses. Said differently, the majority of this year's growth has been organic. That doesn't mean traditional M&A won't be coming, but it does mean that we have focused on internally sourced projects, and that is a good thing.
There's a few reasons we like organic growth projects. The first is that it's lower risk. We know the business, the team, the customer, and we've got a 15-year track record of building these types of projects on scope, schedule, and budget. The second is less competition. As the incumbent owner and operator of these businesses, we are uniquely positioned to win on these mandates. When you combine lower risk with less competition, that gives you excellent risk-adjusted returns.
What we don't underwrite in these projects is the fact that we're building them at several turns discount to what they would go for in the prevailing market. That value will be captured through capital recycling, something we talked about last year. Because these projects are now at scale, we've built our backlog into a record level. In fact, five years ago at this event, we had set a pretty ambitious target for ourselves to grow our backlog to $3 billion. By 2024, we'd achieved that and more. We did $4 billion, a third above our target.
Fast-forward to today, our backlog stands at $6 billion, and that excludes any capital for the Intel joint venture. The important part here is not just the numbers, but it's the trajectory. As our portfolio of companies have grown, as the businesses we've acquired have grown, so has our ability to source extremely attractive projects from within them. If we dive a bit more into the backlog, we'd like to think of it as a funnel. At the bottom of the funnel is the $6 billion of approved projects that we have underway today.
This provides highly visible earnings over the next two to three years, as we've already secured the customer, the financing, and we're well underway. If we work our way up, say, the layer at the planning and commercialization phase, we have another $3 billion of projects at our share that we're in active discussions and final negotiations with customers on. This is really where the relative ranking and prioritization of projects gets done to find the best opportunities within our portfolio.
Taking one step further back, there's $6 billion or more of projects that are in the early stage of opportunities. This is where we may be looking at sites, having initial discussions with our customers to help unlock the growth within their business. The point here is really around the depth of the funnel. It gives us good visibility, not just into the earnings in the next two to three years, but it gives us the confidence that we'll replenish this backlog with a significant amount of capital, as well as grow it over time.
If we combine our scale with the fact that we have a broadening opportunity set in front of us, it gives us the conviction to believe that we can deploy a significant amount of capital at really attractive risk-adjusted returns, and therefore drive further FFO per unit growth. If we pull it all together, starting with the deployment channel, you can see we've invested, or as you recall, we invested about $2.2 billion annually over the last three years. Going forward, we expect that number to be anywhere between $2 billion and $3 billion per year.
Based on our historical deployment, which was traditionally M&A, you can see that range is between $1 billion and $2 billion per year. Again, will vary because, again, we are value-based investors, and so we will choose where we put that capital. On top of that, you have the AI infrastructure opportunity set of about $500 million per year and funding our backlog between $500 million and $1 billion in a given year. Looking at that, what's important here is that we have the choice to invest capital where we see the best returns.
The choice won't just come through in the deployment figures. It will really come through in the returns that we generate. Across each of these three channels, whether it be traditional M&A, AI infrastructure, or our organic growth backlog, we're targeting a minimum of 15% equity returns. To help frame what does that mean for per unit growth in the years to come, we thought we'd show a few scenarios. On the conservative end, let's assume we invest the low amount, $2 billion per year, and we deploy it only at 15%.
That will drive 5% per unit growth that year. If we are able to flex our investment amount up to $3 billion or generate returns above the 15%, that's going to translate into higher per unit growth. At the higher end of the range, you'll see upwards of 8% per year growth from the business. The point here is that the broader opportunity set improves both the amount we can deploy, but also the returns we can generate. And that combination gives us good visibility into today, into 2026 earnings being at 10% per unit or higher, or into the next two to three years.
As we look out, the numbers are quite attractive. As a reminder, the base business grows from inflation indexation and volume surplus. That is 4%-6% per year without investing a single dollar into growth. Now, if you use the low end of our target range of $2 billion at 15%, well, that is roughly 5% per unit growth, making a very visible path to our 10% long-term target. As I said, there is upside here. If we are able to deploy more capital, which we hope we do, and we are able to do so at better returns, well, we have a business plan that can get us to 12% per year growth or higher.
That will allow us to continue to inflect in our growth rates and support long-term distribution growth as well. As a reminder, since 2009, we have grown our distribution for 17 consecutive years at a 9% average rate. As we look ahead, as our growth rate continues to accelerate, we hope that our distribution track record will as well. I will wrap up today with four key takeaways as every Brookfield presentation has seemed to do. First, organic growth is at record levels, and that is a good thing for our business.
Second, our scale has made us a partner of choice, leading to bigger investment opportunities at higher attractive returns. Third, AI infrastructure is an evolving asset class that is allowing us to deploy more capital at really good risk-adjusted returns. Finally, all this results in our growth rate being at the early stages of an inflection point. With that, I will now invite Sam back up for closing remarks and Q&A. Thank you, everyone.
Thank you. Thanks. All right. Well, thank you for your time today. Before we take questions, I thought I would just close with a few thoughts on why we think this is an attractive entry point for investors to come into the BIP units. First, our BIP units offer a dividend yield of more than 5% today, providing investors with a meaningful current return while participating in the growth of the business. Second, as Dave just highlighted, the growth rate in our FFO has begun to accelerate, and that should support an attractive dividend growth rate in the future. Hopefully we can get back up to that 9%.
Lastly, we're hoping that the simplification of our corporate structure, which should take place by end of year, will help re-rate the stock a bit by making it easier to own, increase liquidity, and broaden the potential for new investors to come into the stock. With that, I'll conclude and take any questions that there are. Oh, Cherilyn up front here.
Hi, Sam. Thanks. Cherilyn Radbourne from TD Cowen. One of the themes in your presentation was that scale brings you into better opportunities. The corollary of that is that you end up with a bigger business, so you need more sort of optionality on the exit side to fully monetize that. Can you sort of close the loop for us there?
Sure. No, it's a great question, and it's one that we get a lot, because the natural reaction to when you say you buy something that's bigger is that, as you add value and look to sell 10 years from now, it's even bigger. If you got in because there are very few people who can get in, well then how do you expect to sell it? What I'd say, and I think this is what Scott said in his remarks a bit, is that these businesses typically aren't just one business.
Often what happens when we buy a large enterprise is it's a One of the things that we tend to do is examine opportunities where we can split the business, either by geography or different business lines, and make that available to a more mid-market type buyer. A good example of that would be a business we bought a number of years ago in the U.S., a railroad, a short line railroad, which was a business that had been rolled up over many years.
What we saw is an opportunity to split that business into three different sections by region, and our plan will be, in a couple years' time, once we've finished optimizing the business, to sell that company in those three different pieces. But if someone saw that as a public company, initially they weren't thinking of it as three different businesses. They just thought of it as one company. Hopefully I answered your question. There's two over here.
Hi, Frederic Bastien at Raymond James. Sam, I would like to touch on the Canada Investment Summit that was held a couple weeks ago. Of the federal initiatives announced, including the new investment tax write-off, proposed airport privatization, and the National Secure Digital Network, which do you expect will create the strongest opportunities for Brookfield Infrastructure?
Well, look, I think there are opportunities in all of the above. I think for us, we have been advocating for a change to the tax code for a while, and I think that is going to spur a lot of new investment into the country, so I think that is very positive. Obviously, one of the things Canada has not been doing as well as some other countries has been monetizing mature assets to generate proceeds that they can then invest in new infrastructure. We have seen that done in Australia, we have seen it done in the U.K., we have seen it done in many other places, and we are seeing right now taking place in the Middle East.
Canada should be taking advantage of all the capital that is available and that is interested in mature assets. So I think the airports, should they proceed with that, will find a lot of interested buyers, including ourselves. I think obviously, the other important factor is just streamlining the consultation and approval processes will make a big, big difference in encouraging developers to take risks with new projects. I think that is something that has not taken place over the last decade, and I think that will have a big impact on the future of Canada. There is one over here.
Good afternoon. Maurice Choy from RBC Capital Markets. Just one question from me. I wanted to unpack a comment earlier made in the Bruce and Howard session that inflation today is caused by two ongoing wars and inflation, and the rest of the system is, quote, "not that much." You said earlier that the AI opportunity has now increased to $10 trillion, and because of the war, likely global energy infrastructure CapEx is also likely to increase. All this suggests that there is likely a persistent supply chain-driven inflation in the years ahead, even if the war ends.
So I wonder if you had any thoughts on that. As a quick follow-up, on one of the slides that David presented on slide 48, I think the first bucket for your FFO growth was inflation indexation of 3% or 4%. I wonder if you could possibly see that increasing higher if inflation does persist.
Well, the first thing I would say, it's never wise to disagree with your boss. If he says inflation's not going up, then I agree with him. Look, his point was that there are obviously inflationary pressures from the war and from CapEx going into various industries. What you said is correct. At the same time, there are other factors that are deflationary. There's overcapacity in certain sectors. There's demographic factors taking place. How all those weigh in over a longer period of time is hard to say.
But I do think that generally speaking, once some of these near-term factors dissipate, and look, the war should end at some point relatively soon, and I think the investment boom will moderate as it always does. I think longer-term, you will see inflation taper down. The benefit, though, is that if it doesn't, there's no better asset class, no better stock to own than BIP, because almost all our revenues are indexed to inflation, and we'll be big beneficiaries of that. So thank you for that lob ball. I think that's all we have time-wise. So thank you very much. Appreciate your time, and be happy to speak to anyone outside.