WSS: Earnings shows more signs of growth acceleration as AI inferencing drives demand and strategic moves
A busy week saw earnings season in full swing and the results coming in from across the ecosystem showed steady forward movement and signs of growth acceleration as AI workloads continue to drive cloud infrastructure and data centre colocation consumption. In the last few weeks, we took a closer look at the results coming from the hyperscalers (AWS, Microsoft, Google) and Meta was one we had not yet touched on. Meta does not have a raw cloud infrastructure business, but its CapEx disclosures are an important indicator for data centre spending and expansion. For the fourth straight quarter, Meta saw its CapEx climb and it came in at $19.2b in the recent quarter. This is expected to increase moving into 2026 and Meta confirmed intentions to spend up to $600b in CapEx by 2028 to build more data centre capacity across the United States.
The data centre segment is not heavily represented in the public markets after a wave of take-privates over the last several years. But there are now various publicly listed cryptocurrency firms moving into the data centre space and disclosing results that give us insight into the sector’s overall demand profile, particularly around GPUs and AI. There are a number to get through, but we have already commented on Applied Digital’s earnings report and this past week took a closer look at Core Scientific’s results. Core Scientific was set to, but will not be acquired by CoreWeave as its shareholders rejected a $9b acquisition offer. But Core Scientific is going to continue building data centre capacity for CoreWeave. This deal may not have gone through but given the many other cryptocurrency firms moving into the space, there should be no shortage of strategic opportunities. We will have more on the results coming from other cryptocurrency-oriented firms such as the likes of Northern Data, TeraWulf, Galaxy, Nebius and Cipher in the coming weeks.
The public markets also trade a number of webscale and managed infrastructure providers. In the past week, we reviewed the earnings coming from DigitalOcean and Akamai. Both these providers offer an alternative to hyperscale clouds and are growing at a steady, if not unspectacular pace. But there were signs in the last few quarters that things are starting to pick up pace. The AI inferencing opportunity is creating demand for these platforms and driving uptake in enabling services. Akamai is seeing the fruits of its move into compute infrastructure and Fastly is on somewhat of a similar journey. Its growth rate stepped up this quarter as well, though computing infrastructure is still a small part of the overall business. Akamai used the acquisition of Linode back in 2022 to push into compute infrastructure and we also saw Megaport move down a similar path. In the past week, Megaport acquired cloud provider Latitude.sh and will try to combine connectivity and compute infrastructure the way Akamai has done with delivery and compute infrastructure. The rationales are similar and the main driver is that controlling compute is the path to stickiness and growth into adjacent services. We will have more details on Megaport’s acquisition later this week in our regular commentary.
A lot of the growth acceleration is being driven by emerging AI inferencing demand, which is creating the need for smaller increments of computing infrastructure and in strategic or edge locations. This is a good fit for platforms like Akamai, Fastly and Megaport that have distributed local networks and extensive connectivity solutions. Akamai is going after this demand and recently released Akamai Inference Cloud, enabled by NVIDIA technology.
Still with earnings season, we also reviewed the results coming from Rackspace. It has managed to stem the revenue attrition it has been experiencing and is another provider that seems to have pushed the needle forward in recent quarters.
The past week also saw more global expansion activity. Microsoft is partnering with True IDC to build a cloud infrastructure region in Thailand and committing to AI infrastructure builds in the UAE. Meanwhile, Equinix launched a new data centre in Monterrey, Mexico and Global Technical Realty confirmed plans for a second campus in London, UK. All these developments are backed by tangible customer demand. But the fake data centre continues to be part of the landscape. Our latest commentary on this subject, courtesy of our contributor Appleby Strategy Group CTO Daniel Golding, looks at one of the more obvious markers: the lack of people and tools. If an operating platform does not have the right people, and enough people, the likelihood of success will be low. And even with people, the industry needs to start equipping them with the right tools. The sector is building for AI and it is time to start using AI to optimize operations and the build process.
or