Welcome to another episode of Rendez-vous with the Sustainable Equities Team. I'm Joe Toscano, an equity analyst on the team. And I'm Shawn Kumar, an equity analyst on the team. Today we're going to talk to you about data centers. Joe will kick us off.
Now that most of Q2 earnings season is behind us, we can review how the data center market is looking going into the second half. Upon review of trends leaving Q2, it is clear that AI infrastructure spending is not slowing down. And what is driving this growth in data center builds is the mega trend of generative AI throughout 2025. Hyperscaler and AI CapEx expectations have grown considerably. Starting before Q1 CapEx expectations from the five biggest AI spenders was expected to be around 340 billion. Coming out of Q1, this was up to 360 billion. And now exiting Q2, we are at 420 billion in total CapEx spend for 2025. There's been a lot of concerns regarding the return generated from the large CapEx outlays. We're looking at how these five performed revenue beat consensus expectations by an average of 3.5% across these names and earnings was ahead by almost 12%.
It's not all AI, but is helped by AI tailwinds while GPU chips get a lot of headlines. When it comes to AI, there are many components that are involved in building out data centers on the server side. In addition to GPUs and accelerated chips, there are CPUs, storage, memory, and network interface cards. The networking side includes switches that connect servers and racks, and a leaf spine architecture and routers to connect different networks together along with things like firewalls and load balancers to manage the data. Lastly, optical includes different types of copper and fiber optic cables for data transmission compute, fabric interconnect ties GPUs together within the server, and there's also rack direct cable connections and large data center interconnect as well. In addition to the digital side, there's a large billet of gray space to go with it, which you'll hear more about from Shawn.
Joe discussed the digital side inside a data center referred to as the white space and mentioned the gray space. The white space contains a lot of the areas that we have covered, including semiconductors, server hardware, and networking equipment. The gray space broadly covers the infrastructure it needed to run a data center, including the electrical cooling and ventilation and building operation equipment outside of the building. The value chain includes companies providing construction and engineering services, real estate, and critically the energy supply to meet the growing power needs of these data centers. Now, this supply can come from a variety of sources, including renewables, and there's typically a need for short-term backup power as well. Now, it's worth noting that while the size of the addressable market for the white space equipment such as the semiconductors is substantially larger, the infrastructure found in the gray space and outside of the building is just as critical for the rollout of these data centers more broadly. And as the technology evolves and generative AI gets more sophisticated, there's a greater and greater need for equipment focused on power management and for advanced cooling technologies that can handle the higher level of heat generation coming from these semiconductors. And lastly, securing the energy supply for a data center is often the bottleneck for expansion that many of the providers face.
Well, we hope you found this to be helpful, and thank you for joining us today.
