AIB Data Centers Inc. LD Micro 20th Annual Main Event Conference

Thank you, Kendall. It is a pleasure to be here to chat with everybody today and discuss what is going on with AIB Data Centers. I will get started. In the AI economy, power is scarce.

I am going to spend the next 20 minutes or so talking about just that, our people, our power, and our pipeline. I am also going to talk a little bit about our progress and the potential of what we have going on and what our future looks like. First of all, just as a company overview, we call ourselves a right-sized developer. We stay in that 40-90 MW zone in terms of our developments. We do not chase anything hyperscaler, anything to do with hyperscalers. We have found that our sweet spot is within that under 100 MW, and usually between 50 MW and 75 MW. There are a few reasons for that. It is a simpler lease negotiation. There is an easier supply chain to develop and navigate. You are not buying 3,000 of something. You are not buying 3,000 generators. You are buying 300 generators.

That type of scale makes that supply chain that much easier to navigate. We also find as well that there is a lot of, it is well known in our geopolitical environment, there is a lot of pushback on certain types of data center developments. We find that at this size, we are embraced by the communities that we go into. We very much pride ourselves on becoming part of the communities where our developments are built, and that we engage with that community, and we support that community as they support us. Just to reiterate, we are power first. We do not even look at a piece of land unless there is going to be power there within 12 months. We make sure that we have those utility agreements in place. There is a lot of what our CEO Jerry Tang calls fake power, where it is a maybe power.

We are only securing sites where power is assured within the next 12 months. We have contracted demand in terms of we sell our capacity to creditworthy customers under long-term agreements. We will talk a little bit about the agreement that we just signed with Nebius in the slides that are upcoming. Then we look at our project level funding and financing. We structure our financing to have reliance upon the equity into the development and not fully reliant on raising common stock equity to fund our projects, and we will talk about that as well in the slides to come. On the other side of the slide, on the right-hand side, is the portfolio at a glance. Right now we have 120 MW under contract of power, and that is secured. It is either under ESA or an FEA type agreement.

Right now, we have contracted with a tenant at 50 MW of critical IT load. That agreement is a 12-year term with 2 renewals. We have 5 other sites currently under evaluation. I am going to talk for a minute about what is shaping the demand in this AI space that we have. You can see from this graph on the right-hand side, it is talking about the global demand for data center power in gigawatts. Right now, 2025, we are sitting at 82 gigawatts. By 2030, that is expected to jump up and balloon up to 220 gigawatts. Of that, you can see also within that, the dark blue space is the AI workloads utilizing that data center capacity is expected to grow up to 70% of that entire global data center space.

In full market context, data center electricity expected to be used worldwide by 2030 is around 945 terawatts, against right now, where we are at 415. 70% of that capacity is expected to be used by AI-related utilizations, AI-related workloads. Also, as part of that, there is $1.7 trillion expected in development, CapEx going into AI data centers through 2030. Actually, Goldman Sachs has projected $1.4 trillion by the end of 2028. There is a lot of speculation in terms of the investment, and those numbers move, but you can clearly see the trajectory here. AI, we are just at the beginning of what is happening in this space. We are in a space as well, I am going to talk about 3 of the constraints that we have. Power being one of them, and we have talked about that. We secure our sites with power availability.

There is lots of land out there. You just have to know where to look and how to find it and what you are looking for. There is also chips. That is another constraint within our industry. That is not our issue. We do not do anything with hardware. That would be at Nebius or our off-takers. That would be their constraint. The third constraint that has recently arisen is capital. We are now competing for capital. With an off-taker like Nebius, which we will talk about more, we feel that the risk, an off-taker like Nebius with the terms of the lease that we have secured, financing, we do not anticipate to be an issue for this project.

When you look at here, power being scarce, and we look at the data center vacancy, because we monitor different markets, Northern Virginia, Atlanta, Dallas, Fort Worth, and Chicago, and you can clearly see these low levels of vacancy. When you have queues of 4-plus years average wait time per new grid connection, that is a real constraint. That 88% of capacity at this point is actually pre-leased, even it is before construction in most markets. You have the grid time as a constraint, and once you actually have your power secured, then you move on to equipment procurement. Timelines can create a problem and permitting. That can be a constraint as well. Where we fit in the AI infrastructure in our food chain or our ecosystem, so to speak. Land and power is the primary input.

We come in as a developer to secure that land and power, and then we begin our design. We begin the construction of our facility, the building, the electrical infrastructure, and the cooling systems. Then we sell that contracted critical IT capacity under long-term agreements. We really do not start construction until those long-term agreements are in place, because we do not have to. You really can not, because that lease, depending on the off-taker you have, that is the piece of the project that is financeable. Then our customer becomes in the next link in the chain. They install and run their own GPUs, and for us, that is Nebius as our anchor tenant. Then you have the end market users as AI developers and different enterprises that are training and running the AI models as part of their compute solution.

Like I said, we own the building, the power infrastructure, the cooling systems. The customer owns the servers and the GPUs. Then how we are generating revenue and monetizing is we contract out that space and lease it out. We also have cash flows during our construction. In this case, for example, Nebius has contracted us to build out their white space. They fund the build, but they pay us to build out their white space. This project will probably start generating operational cash flow in month three. What we like most about our model, and this is what we call our rinse and repeat, so it is a repeatable sequence that we have set ourselves up for. In just taking you through the steps, number one is to secure the power. We have that complete. We have a 15-year ESA signed at our anchor site in South Carolina.

We have the customer capacity contracted with Nebius. That part is complete as well. These lease negotiations take anywhere from 3 to 6 months. Because we are in that smaller footprint at a smaller megawatt size, our data centers, for example, are the size of a Home Depot, and we use less water than an Olive Garden for our data centers. So much easier to negotiate those leases. At this point, those two steps are done. Now we are working very diligently on securing our project funding. That has been underway for months. We have been working with several banks. You really cannot kick off the process, though, until you have that signed lease in hand, because that is the financeable piece of the project.

Afterwards, there is also the permitting that we talked about, the procurement, and the phased construction. We understand at this point we have a modular build. We have two data halls planned for our first project. Data hall one is due within 10 months. The other 25 megawatts, data hall two, is due 4 months later. We are in that step right now, and we understand how critically important it is to be on time and on budget. From there, it is generating the long-term revenue streams. In our piece, we have a property manager hired to run the data center. It is a gross modified lease. Power is a complete pass-through, so we have no risk in power in terms of fluctuating rates, because that is a complete pass-through to the tenant. We are responsible for the property management, the property taxes, and the insurance.