WSS: Earnings season reveals data points, KPIs, pointing to growth, accelerated investment in infrastructure amid strong demand signals
Earnings season is in full swing and the results coming from the sector’s bellwether companies provide valuable insights into what direction the sector is tracking. The hyperscale clouds have seen growth accelerate over the last several quarters and things mostly held steady in the recent quarter. AWS and Microsoft Azure grew at basically the same elevated rate seen in the previous quarter and Google Cloud showed some meaningful uptick as did Oracle Cloud back in September. Growth has been driven by improved macroeconomic conditions, an end to the cost-cutting wave coming out of the pandemic and the rise in demand for AI tools. Public cloud growth is always good news for data centre colocation providers, and the two data centre bellwethers – Equinix and Digital Realty – reported results in the last week. Both companies showed that hyperscale growth is translating into leasing momentum and interconnection uptake, while pushing investment levels upwards as capacity requirements increase and time-to-market becomes crucial. While hyperscalers continue to self-build, data centre colocation remains a strategic must given the various constraints around energy and the unpredictability of the process of building data centres.
The intersection of hyperscale cloud and data centre colocation is a critical one for the sector and the healthy demand profile for cloud and AI is pushing the demand and supply situation to a place of imbalance. All the hyperscalers have either explicitly stated or hinted that demand is exceeding supply. And it is this imbalance that will have an impact on the growth trajectory. Microsoft, for example, had a slightly better quarter because some data centre inventory came online ahead of schedule, while Oracle pointed to data centre capacity shortages as one of the reasons why demand for infrastructure is signed, but not booked, and may not be for some time. Hyperscale revenue growth is a leading indicator of course, but the CapEx numbers are also an important metric from an infrastructure perspective. CapEx continues to climb for all the hyperscalers and Microsoft, for example, hit $20b in the recent quarter. The numbers are big, but should be taken with a grain of salt. While CapEx is mostly going to building out public cloud infrastructure, Google and Microsoft had broken out the split between how much is going to servers (both CPUs and GPUs) and how much is going to data centres. There are hints that the demand for GPUs is pushing the CapEx numbers up. That is not to say there is not plenty of CapEx going into data centres. Clearly, data centres are getting a big chunk of the CapEx, but there is plenty going to the hardware part of the Internet infrastructure ecosystem.
GPU-based infrastructure is leading to new company formation and transition and we have commented on this and tracked it closely. Just this past week, Rackspace became the latest to jump into the game and rolled out a GPU Cloud offering. Rackspace has de-emphasized direct public cloud services, opting to manage public cloud platforms, while sticking with private cloud services. It has let the legacy OpenStack-based public cloud service decline and it has been rapidly shrinking. In a bit of an upset, the rise in AI has caused Rackspace to get back into direct service delivery. The GPU service is available on-demand and comes with the ability to bid on pricing. It will be interesting to see where this goes. Does Rackspace get back into cloud in a meaningful way? Or is this more a service that drives uptake and mindshare, and then funnel into its main portfolio of services. Time will tell.
There is change elsewhere when it comes to service provider cloud infrastructure services. This is a smaller and focused market, but there is value being driven (see our managed infrastructure marketshare report for data points). Things however have been knocked off course in recent months as VMware continues to wreak havoc on the ecosystem and left service providers thinking about alternatives and backup plans. US Signal is a VMware customer, but rolled out a new cloud service built on Apache CloudStack. The move could be a harbinger of things to come from other infrastructure service providers.
Meanwhile, in APAC, the market continues to see steady growth. Among Equinix’s three main regions – Americas, Europe and APAC – APAC is by far growing the fastest at 15% y/y in the recent quarter. And the hyperscale investments in Southeast Asia of late speak to where the growth and expansion is happening. Equinix has pushed into Malaysia and Indonesia, and has confirmed it will now invest $500m to enter the market in Thailand. Southeast Asia has built momentum off the innovation and scale in Singapore, and data centre operators GDS has been a beneficiary, serving hyperscale deployments in Malaysia, Indonesia and Singapore. GDS is spinning out GDS International into an independent entity and just raised $1b to fund its expansion efforts. And Bridge Data Centres set up a JV with Mah Sing for a data centre campus project in Malaysia.
With earnings season in full swing, we will be looking at the results coming from power companies in the US, which increasingly have details on data centre and hyperscale demand. The demand is high enough that utilities are putting new conditions on hyperscalers when they commit to such large volumes of energy. The issue is not just the availability of power – though obviously that is increasingly a factor – but the reality around constraints is as much about the supporting delivery infrastructure and the time and capital required to build it out. AEP in Ohio is making progress on this and we have some of the latest.
or